<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.1 20151215//EN" "http://jats.nlm.nih.gov/publishing/1.1/JATS-journalpublishing1.dtd">
<article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" article-type="review-article" dtd-version="1.1" specific-use="sps-1.9" xml:lang="pt">
  <front>
    <journal-meta>
      <journal-id journal-id-type="nlm-ta">Arq Bras Cardiol</journal-id>
      <journal-id journal-id-type="publisher-id">abc</journal-id>
      <journal-title-group>
        <journal-title>Arquivos Brasileiros de Cardiologia</journal-title>
        <abbrev-journal-title abbrev-type="publisher">Arq. Bras. Cardiol.</abbrev-journal-title>
      </journal-title-group>
      <issn pub-type="ppub">0066-782X</issn>
      <issn pub-type="epub">1678-4170</issn>
      <publisher>
        <publisher-name>Sociedade Brasileira de Cardiologia - SBC</publisher-name>
      </publisher>
    </journal-meta>
    <article-meta>
      <article-id specific-use="scielo-v3" pub-id-type="publisher-id">p3L3jTtkvGXKjbhwPx3bYWp</article-id>
      <article-id specific-use="scielo-v2" pub-id-type="publisher-id">S0066-782X2026000701401</article-id>
      <article-id pub-id-type="other">01401</article-id>
      <article-id pub-id-type="doi">10.36660/abc.20250790</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Artigo de Revis&#227;o</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Intelig&#234;ncia Artificial Aplicada &#224; Avalia&#231;&#227;o de Risco Cardiovascular Baseada em Eletrocardiograma: Uma Revis&#227;o Sistem&#225;tica</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0009-0006-8808-1934</contrib-id>
          <name>
            <surname>Pinheiro</surname>
            <given-names>Maria Clara Mantoan</given-names>
          </name>
          <role>Concep&#231;&#227;o e desenho da pesquisa e Reda&#231;&#227;o do manuscrito</role>
          <role>Obten&#231;&#227;o de dados</role>
          <role>An&#225;lise e interpreta&#231;&#227;o dos dados e Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante</role>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-0269-569X</contrib-id>
          <name>
            <surname>Felberg</surname>
            <given-names>L&#237;via</given-names>
          </name>
          <role>Concep&#231;&#227;o e desenho da pesquisa e Reda&#231;&#227;o do manuscrito</role>
          <role>Obten&#231;&#227;o de dados</role>
          <role>An&#225;lise e interpreta&#231;&#227;o dos dados e Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante</role>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-3800-7107</contrib-id>
          <name>
            <surname>Bozzi</surname>
            <given-names>Isadora Cristine Reis Sguizzato</given-names>
          </name>
          <role>Obten&#231;&#227;o de dados</role>
          <role>An&#225;lise e interpreta&#231;&#227;o dos dados e Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante</role>
          <role>An&#225;lise estat&#237;stica</role>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-2740-0042</contrib-id>
          <name>
            <surname>Ribeiro</surname>
            <given-names>Ant&#244;nio Luiz Pinho</given-names>
          </name>
          <role>An&#225;lise e interpreta&#231;&#227;o dos dados e Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante</role>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0003-1349-1745</contrib-id>
          <name>
            <surname>Paix&#227;o</surname>
            <given-names>Gabriela Miana de Mattos</given-names>
          </name>
          <role>Concep&#231;&#227;o e desenho da pesquisa e Reda&#231;&#227;o do manuscrito</role>
          <role>Obten&#231;&#227;o de dados</role>
          <role>An&#225;lise e interpreta&#231;&#227;o dos dados e Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante</role>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff1">
        <label>1</label>
        <institution content-type="orgname">Universidade Federal de Minas Gerais</institution>
        <addr-line>
          <named-content content-type="city">Belo Horizonte</named-content>
          <named-content content-type="state">MG</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">Universidade Federal de Minas Gerais, Belo Horizonte, MG &#8211; Brasil</institution>
      </aff>
      <author-notes>
        <corresp id="c01"><label>Correspond&#234;ncia:</label> Gabriela Miana de Mattos Paix&#227;o Universidade Federal de Minas Gerais - Rua Alfredo Balena, 190. CEP 30130-100, Santa Efig&#234;nia, Belo Horizonte, MG &#8211; Brasil E-mail: <email>gabimiana@gmail.com</email> </corresp>
        <fn fn-type="coi-statement">
          <label>Potencial Conflito de Interesse:</label>
          <p> Declaro n&#227;o haver conflito de interesses pertinentes.</p>
        </fn>
        <fn fn-type="edited-by">
          <label>Editor respons&#225;vel pela revis&#227;o:</label>
          <p> Marcio Bittencourt</p>
        </fn>
      </author-notes>
      <pub-date date-type="pub" publication-format="electronic">
        <day>03</day>
        <month>08</month>
        <year>2026</year>
      </pub-date>
      <pub-date date-type="collection" publication-format="electronic">
        <year>2026</year>
      </pub-date>
      <volume>123</volume>
      <issue>7</issue>
      <elocation-id>e20250790</elocation-id>
      <history>
        <date date-type="received">
          <day>30</day>
          <month>01</month>
          <year>2026</year>
        </date>
        <date date-type="rev-recd">
          <day>04</day>
          <month>05</month>
          <year>2026</year>
        </date>
        <date date-type="accepted">
          <day>20</day>
          <month>05</month>
          <year>2026</year>
        </date>
      </history>
      <permissions>
        <license license-type="open-access" xlink:href="https://creativecommons.org/licenses/by/4.0/" xml:lang="pt">
          <license-p>Este &#233; um artigo de acesso aberto distribu&#237;do sob os termos da licen&#231;a de atribui&#231;&#227;o pelo Creative Commons</license-p>
        </license>
      </permissions>
      <abstract>
        <title>RESUMO</title>
        <p>As doen&#231;as cardiovasculares s&#227;o a principal causa de mortalidade no mundo, destacando a necessidade de estrat&#233;gias eficazes de predi&#231;&#227;o de risco e detec&#231;&#227;o precoce. Embora o eletrocardiograma (ECG) seja um exame amplamente dispon&#237;vel e de baixo custo, sua interpreta&#231;&#227;o tradicional &#233; limitada pela subjetividade. A intelig&#234;ncia artificial (IA) surgiu como uma abordagem promissora, capaz de extrair informa&#231;&#245;es progn&#243;sticas ocultas dos sinais de ECG. Esta revis&#227;o sistem&#225;tica teve como objetivo avaliar estudos originais que aplicaram t&#233;cnicas de IA a ECGs para predi&#231;&#227;o de risco cardiovascular e mortalidade. Foram inclu&#237;dos estudos originais que utilizaram sinais de ECG como &#250;nica vari&#225;vel de entrada para modelos de IA, com foco em desfechos de risco cardiovascular. Uma busca sistem&#225;tica foi realizada em diferentes bases de dados, e os dados fora msintetizados de forma narrativa. Onze estudos foram inclu&#237;dos, predominantemente coortes retrospectivas que aplicaram redes neurais convolucionais (CNNs) para prever risco cardiovascular ou mortalidade. As amostras eram majoritariamente compostas por popula&#231;&#245;es adultas de pa&#237;ses de alta renda. Os desfechos prim&#225;rios inclu&#237;ram mortalidade por todas as causas, morte cardiovascular e eventos cardiovasculares adversos maiores (MACE). Os valores de AUROC variaram de 0,63 a 0,961 nos conjuntos de treinamento, com alguns modelos superando escores tradicionais de risco. Os modelos de IA&#8209;ECG demonstraram potencial para detectar doen&#231;a subcl&#237;nica, permitindo estratifica&#231;&#227;o precoce de risco mesmo em ECGs normais. No entanto, persistem desafios relacionados &#224; diversidade populacional, interpretabilidade dos modelos e valida&#231;&#227;o prospectiva. A aplica&#231;&#227;o de IA &#224; an&#225;lise de ECG representa um avan&#231;o promissor na avalia&#231;&#227;o personalizada do risco cardiovascular. Contudo, mais pesquisas s&#227;o necess&#225;rias para garantir a seguran&#231;a, a efic&#225;cia e a integra&#231;&#227;o cl&#237;nica equitativa dessas tecnologias.</p>
      </abstract>
      <abstract abstract-type="graphical">
        <p>
          <fig id="f01">
            <label>Figura Central</label>
            <caption>
              <title>: Intelig&#234;ncia Artificial Aplicada &#224; Avalia&#231;&#227;o de Risco Cardiovascular Baseada em Eletrocardiograma: Uma Revis&#227;o Sistem&#225;tica</title>
            </caption>
            <alternatives>
              <graphic xlink:href="https://minio.scielo.br/documentstore/1678-4170/p3L3jTtkvGXKjbhwPx3bYWp/28c9c00995507c940de3fbbcde139288ec2da357.tif"/>
              <graphic xlink:href="https://minio.scielo.br/documentstore/1678-4170/p3L3jTtkvGXKjbhwPx3bYWp/ebc45368f00c79522bb8e681c04870880dc22fa7.png" specific-use="scielo-web"/>
              <graphic xlink:href="https://minio.scielo.br/documentstore/1678-4170/p3L3jTtkvGXKjbhwPx3bYWp/a45b157ce16119caeb4915fd938069fd3dfede04.jpg" specific-use="scielo-web" content-type="scielo-267x140"/>
            </alternatives>
          </fig>
        </p>
      </abstract>
      <kwd-group xml:lang="pt">
        <title>Palavras-chave:</title>
        <kwd>Intelig&#234;ncia Artificial</kwd>
        <kwd>Fatores de Risco de Doen&#231;as Card&#237;acas</kwd>
        <kwd>Eletrocardiografia</kwd>
      </kwd-group>
      <funding-group>
        <funding-statement><bold>Fontes de Financiamento:</bold> O presente estudo n&#227;o teve fontes de financiamento externas.</funding-statement>
      </funding-group>
      <counts>
        <fig-count count="10"/>
        <table-count count="10"/>
        <equation-count count="0"/>
        <ref-count count="24"/>
      </counts>
    </article-meta>
  </front>
  <body>
    <sec sec-type="intro">
      <title>Introdu&#231;&#227;o</title>
      <p>As doen&#231;as cardiovasculares (DCVs) s&#227;o a principal causa de mortalidade no mundo, englobando tanto condi&#231;&#245;es transmiss&#237;veis quanto n&#227;o transmiss&#237;veis.<sup><xref ref-type="bibr" rid="B1">1</xref></sup> Portanto, a estratifica&#231;&#227;o do risco cardiovascular e a detec&#231;&#227;o precoce dessas doen&#231;as s&#227;o essenciais. Nesse contexto, o eletrocardiograma (ECG) destaca-se como uma ferramenta bem estabelecida, devido &#224; sua ampla disponibilidade, baixo custo e simplicidade operacional. No entanto, a interpreta&#231;&#227;o do ECG apresenta uma limita&#231;&#227;o importante: sua subjetividade, uma vez que depende de crit&#233;rios visuais que variam conforme o examinador.</p>
      <p>Nos &#250;ltimos anos, a intelig&#234;ncia artificial (IA) ganhou destaque significativo, especialmente na &#225;rea da medicina, abrindo novas fronteiras na an&#225;lise de dados. Abordagens baseadas em <italic>machine learning</italic> (ML) e <italic>deep learning</italic> (DL) demonstraram capacidade de reconhecer padr&#245;es complexos e sutis que podem passar despercebidos por observadores humanos. Isso levou &#224; hip&#243;tese de que integrar IA &#224; an&#225;lise de ECG (IA-ECG) pode permitir a extra&#231;&#227;o de informa&#231;&#245;es progn&#243;sticas e de estratifica&#231;&#227;o de risco cardiovascular anteriormente n&#227;o identificadas.<sup><xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref></sup></p>
      <p>Os modelos atuais de estimativa de risco cardiovascular apresentam limita&#231;&#245;es importantes, pois nem sempre refletem com precis&#227;o a realidade individual do paciente e, muitas vezes, n&#227;o incorporam dados do ECG. O uso de m&#233;todos adicionais, como o escore de c&#225;lcio coronariano, apresenta bom desempenho na reestratifica&#231;&#227;o do risco cardiovascular; entretanto, o custo permanece uma barreira significativa, especialmente em pa&#237;ses de baixa e m&#233;dia renda.<sup><xref ref-type="bibr" rid="B4">4</xref></sup> Como resultado, um n&#250;mero consider&#225;vel de pacientes permanece desassistido, sobretudo aqueles de popula&#231;&#245;es diversas cujas caracter&#237;sticas sociais e raciais n&#227;o s&#227;o adequadamente consideradas. Isso refor&#231;a a necessidade de valida&#231;&#227;o em popula&#231;&#245;es heterog&#234;neas &#8211; uma pr&#225;tica ainda ausente em alguns dos estratificadores de risco existentes.<sup><xref ref-type="bibr" rid="B4">4</xref></sup></p>
      <p>Embora a maioria dos eventos ateroscler&#243;ticos possa ser prevenida por meio de estrat&#233;gias de promo&#231;&#227;o da sa&#250;de e do controle de fatores de risco conhecidos, vari&#225;veis individuais persistem, incluindo baixa ades&#227;o ao tratamento e desigualdade social.<sup><xref ref-type="bibr" rid="B2">2</xref></sup> Nesse contexto, a aplica&#231;&#227;o de IA &#224; an&#225;lise de ECG pode oferecer uma alternativa para apoiar o desenvolvimento de modelos de predi&#231;&#227;o mais sens&#237;veis, personalizados e integrados.<sup><xref ref-type="bibr" rid="B3">3</xref></sup></p>
      <p>O objetivo deste artigo &#233; fornecer uma revis&#227;o sistem&#225;tica atualizada e cr&#237;tica das principais aplica&#231;&#245;es de IA na interpreta&#231;&#227;o de ECG, com foco na predi&#231;&#227;o de eventos cardiovasculares adversos maiores (MACE). O artigo examina os avan&#231;os recentes, os desafios associados &#224; implementa&#231;&#227;o em larga escala e as perspectivas para sua ado&#231;&#227;o na pr&#225;tica cl&#237;nica. Os principais achados desta revis&#227;o sistem&#225;tica est&#227;o resumidos na <xref ref-type="fig" rid="f01">Ilustra&#231;&#227;o Central</xref>.</p>
    </sec>
    <sec sec-type="methods">
      <title>M&#233;todos</title>
      <p>Esta revis&#227;o sistem&#225;tica foi conduzida de acordo com as diretrizes do <italic>Preferred Reporting Items for Systematic Reviews and Meta-Analyses</italic> (PRISMA).<sup><xref ref-type="bibr" rid="B5">5</xref></sup> O protocolo foi registrado no PROSPERO (1039916).</p>
      <p>A metodologia desta revis&#227;o sistem&#225;tica seguiu o modelo &#8220;PICOTS&#8221; (<italic>Population, Intervention, Comparator, Outcome, Timing,</italic> e <italic>Study design</italic>). Foram inclu&#237;dos estudos envolvendo popula&#231;&#245;es adultas que aplicaram modelos de IA utilizando exclusivamente sinais de ECG como entrada, sem a necessidade de um grupo comparador expl&#237;cito. Os desfechos prim&#225;rios foram mortalidade por todas as causas, morte cardiovascular, insufici&#234;ncia card&#237;aca (IC), infarto do mioc&#225;rdio (IAM), acidente vascular cerebral (AVC) e doen&#231;a cardiovascular ateroscler&#243;tica (DCVA). Os per&#237;odos de seguimento variaram entre os estudos, com predomin&#226;ncia de delineamentos de coorte retrospectiva.</p>
      <sec>
        <title>Crit&#233;rios de elegibilidade</title>
        <p>Inclu&#237;mos artigos de pesquisa original publicados em ingl&#234;s nos &#250;ltimos cinco anos que avaliaram diretamente o uso de IA &#8211; especificamente redes neurais convolucionais (CNNs, do ingl&#234;s <italic>convolutional neural networks</italic>) ou redes neurais profundas (DNNs, do ingl&#234;s <italic>deep neural networks</italic>) &#8211; aplicadas exclusivamente a ECGs para prever risco cardiovascular em popula&#231;&#245;es adultas (&#8805;18 anos). Os estudos eleg&#237;veis avaliaram o risco ou a predi&#231;&#227;o de mortalidade por todas as causas, DCVA e eventos cardiovasculares adversos maiores (MACE), incluindo IAM, IC, AVC e morte cardiovascular.</p>
        <p>Foram exclu&#237;dos estudos que: (1) integraram dados de ECG com outras informa&#231;&#245;es cl&#237;nicas, laboratoriais ou de imagem; (2) focaram exclusivamente no diagn&#243;stico de arritmias; (3) foram restritos a popula&#231;&#245;es espec&#237;ficas de pacientes; ou (4) eram revis&#245;es narrativas, relatos de caso ou n&#227;o disponibilizavam o texto completo.</p>
      </sec>
      <sec>
        <title>Fontes de informa&#231;&#227;o e estrat&#233;gia de busca</title>
        <p>A busca na literatura foi realizada nas bases PubMed, Cochrane Library, Scopus e ScienceDirect. As buscas foram limitadas a artigos publicados entre 2020 e 2025, e a &#250;ltima busca foi realizada em maio de 2025. Os seguintes termos em ingl&#234;s e seus sin&#244;nimos foram utilizados como descritores: <italic>&#8220;Artificial Intelligence&#8221;, &#8220;Machine Learning&#8221;, &#8220;Deep Learning&#8221;, &#8220;Electrocardiography&#8221;, &#8220;Electrocardiogram&#8221;, &#8220;ECG&#8221;, &#8220;Cardiovascular Risk Factors&#8221;, &#8220;Risk Assessment&#8221;, &#8220;Cardiovascular Risk&#8221;, &#8220;Heart Disease Risk&#8221;</italic>. Os operadores booleanos &#8220;AND&#8221; e &#8220;OR&#8221; foram aplicados para combinar os termos.</p>
      </sec>
      <sec>
        <title>Processo de sele&#231;&#227;o</title>
        <p>O processo de sele&#231;&#227;o consistiu em quatro etapas: identifica&#231;&#227;o, triagem, avalia&#231;&#227;o de elegibilidade e inclus&#227;o. A busca inicial identificou 1118 registros, dos quais 1056 foram exclu&#237;dos ap&#243;s a triagem de t&#237;tulos por n&#227;o abordarem o foco central da revis&#227;o. Dos 62 registros restantes, 45 foram exclu&#237;dos ap&#243;s a avalia&#231;&#227;o dos resumos e remo&#231;&#227;o de duplicatas. Em seguida, seis registros adicionais foram exclu&#237;dos ap&#243;s a leitura do texto completo por n&#227;o atenderem aos crit&#233;rios de inclus&#227;o. Por fim, onze estudos cumpriram todos os crit&#233;rios de elegibilidade e foram inclu&#237;dos na revis&#227;o.</p>
        <p>Dois revisores conduziram independentemente todas as fases do processo de sele&#231;&#227;o. Quaisquer discrep&#226;ncias foram resolvidas por meio de discuss&#227;o com um terceiro revisor. Todo o fluxo de sele&#231;&#227;o est&#225; apresentado no diagrama PRISMA (<xref ref-type="fig" rid="f02">Figura 1</xref>).</p>
        <p>
          <fig id="f02">
            <label>Figura 1</label>
            <caption>
              <title>&#8211; Diagrama de fluxo PRISMA: Processo de sele&#231;&#227;o de estudos</title>
            </caption>
            <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf01.tif"/></fig>
        </p>
      </sec>
      <sec>
        <title>Processo de coleta de dados e itens de dados</title>
        <p>A extra&#231;&#227;o de dados foi realizada de forma independente por dois revisores, utilizando formul&#225;rios padronizados criados no Microsoft Excel. Os dados extra&#237;dos inclu&#237;ram o delineamento do estudo, caracter&#237;sticas da popula&#231;&#227;o, tipo de modelo de IA (CNN ou DNN), detalhes da aquisi&#231;&#227;o do ECG e os desfechos cardiovasculares avaliados. Vari&#225;veis adicionais coletadas inclu&#237;ram o cen&#225;rio do estudo, tamanho da amostra, os procedimentos de treinamento e valida&#231;&#227;o dos modelos de IA e se houve valida&#231;&#227;o externa.</p>
      </sec>
    </sec>
    <sec sec-type="results">
      <title>Resultados</title>
      <p>Um total de onze estudos originais foi inclu&#237;do nesta revis&#227;o sistem&#225;tica. Suas caracter&#237;sticas est&#227;o resumidas na <xref ref-type="table" rid="t1">Tabela 1</xref>. Dez estudos eram coortes retrospectivas<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B15">15</xref></sup>e um estudo tinha delineamento prospectivo.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> Os estudos analisaram uma ampla variedade de sinais de ECG, predominantemente ECGs de 12 deriva&#231;&#245;es, com alguns tamb&#233;m avaliando registros de deriva&#231;&#227;o &#250;nica (Deriva&#231;&#227;o I).<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref></sup></p>
      <p>
        <table-wrap id="t1">
          <label>Tabela 1</label>
          <caption>
            <title>&#8211; Caracter&#237;sticas dos estudos de intelig&#234;ncia artificial aplicada na eletrocardiografia (IA-ECG) inclu&#237;dos</title>
          </caption>
          <table frame="hsides" rules="groups">
            <colgroup>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
            </colgroup>
            <thead>
              <tr>
                <th align="left">Autores</th>
                <th>Ano de publica&#231;&#227;o</th>
                <th>T&#237;tulo do artigo</th>
                <th>Tipo do estudo</th>
                <th>T&#233;cnicas de IA</th>
                <th>Tipo de sinal de input</th>
                <th>Popula&#231;&#227;o</th>
                <th>N&#250;mero de ECGs</th>
                <th>Idade m&#233;dia (anos)</th>
                <th>Homens (%)</th>
                <th>Seguimento (m&#233;dia de anos)</th>
                <th>Valida&#231;&#227;o externa</th>
                <th>Popula&#231;&#227;o de valida&#231;&#227;o</th>
              </tr>
            </thead>
            <tbody>
              <tr>
                <td>Sau et al.<sup>6</sup></td>
                <td align="center">2024</td>
                <td align="center">Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es e de deriva&#231;&#227;o &#250;nica (DI)</td>
                <td align="center">BIDMC (EUA)</td>
                <td align="center">1 163 401</td>
                <td align="center">NR</td>
                <td align="center">NR</td>
                <td align="center">5.46</td>
                <td align="center">Sim</td>
                <td align="center">CODE (Brasil) SaMi-Trop (Brasil) ELSA-Brazil (Brasil) UK Biobank (Reino Unido)</td>
              </tr>
              <tr>
                <td>Hughes et al.<sup>7</sup></td>
                <td align="center">2023</td>
                <td align="center">A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es e de deriva&#231;&#227;o &#250;nica (DI)</td>
                <td align="center">Stanford Health Care (EUA)</td>
                <td align="center">500 000</td>
                <td align="center">59</td>
                <td align="center">54%</td>
                <td align="center">5</td>
                <td align="center">Sim</td>
                <td align="center">Cedars-Sinai Medical Center (EUA) Columbia University Irving Medical Center (EUA)</td>
              </tr>
              <tr>
                <td>Al-Alusi et al.<sup>8</sup></td>
                <td align="center">2025</td>
                <td align="center">A deep learning digital biomarker to detect hypertension and stratify cardiovascular risk from the electrocardiogram</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">Massachusetts General Hospital (EUA)</td>
                <td align="center">752 415</td>
                <td align="center">57,3 &#177; 16,8</td>
                <td align="center">51,6%</td>
                <td align="center">5.6&#8211;7.8</td>
                <td align="center">Sim</td>
                <td align="center">Brigham and Women&#8217;s Hospital (EUA)</td>
              </tr>
              <tr>
                <td>Sau et al.<sup>9</sup></td>
                <td align="center">2025</td>
                <td align="center">Artificial intelligence-enhanced electrocardiography for the identification of a sex-related cardiovascular risk continuum: a retrospective cohort study</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">BIDMC (EUA)</td>
                <td align="center">1 163 401</td>
                <td align="center">57,7 &#177; 18,69</td>
                <td align="center">47,9%</td>
                <td align="center">3.4&#8211;4.8</td>
                <td align="center">Sim</td>
                <td align="center">UK Biobank (Reino Unido)</td>
              </tr>
              <tr>
                <td>Gao et al.<sup>16</sup></td>
                <td align="center">2025</td>
                <td align="center">Electrocardiograph analysis for risk assessment of heart failure with preserved ejection fraction: A deep learning model</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">First Hospital of Hebei Medical University (China)</td>
                <td align="center">238</td>
                <td align="center">NR</td>
                <td align="center">50,8%</td>
                <td align="center">2</td>
                <td align="center">N&#227;o</td>
                <td align="center">Valida&#231;&#227;o interna prospectiva</td>
              </tr>
              <tr>
                <td>Raghunath et al.<sup>10</sup></td>
                <td align="center">2020</td>
                <td align="center">Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">Geisinger (USA)</td>
                <td align="center">1 169 662</td>
                <td align="center">58 &#177; 18</td>
                <td align="center">47%</td>
                <td align="center">34</td>
                <td align="center">N&#227;o</td>
                <td align="center">Valida&#231;&#227;o interna cruzada 5-Fold</td>
              </tr>
              <tr>
                <td>Dhingra et al.<sup>11</sup></td>
                <td align="center">2025</td>
                <td align="center">Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">YNHHS (USA)</td>
                <td align="center">286 880</td>
                <td align="center">57</td>
                <td align="center">43,4%</td>
                <td align="center">10</td>
                <td align="center">Sim</td>
                <td align="center">UK Biobank (Reino Unido) ELSA-Brazil (Brasil)</td>
              </tr>
              <tr>
                <td>Lin et al.<sup>12</sup></td>
                <td align="center">2025</td>
                <td align="center">A multitask deep learning model utilizing electrocardiograms for major cardiovascular adverse events prediction</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">CGMH (Taiwan)</td>
                <td align="center">2 821 889</td>
                <td align="center">60 &#177; 14,5</td>
                <td align="center">50,2%</td>
                <td align="center">12</td>
                <td align="center">Sim</td>
                <td align="center">TSGH (Taiwan)</td>
              </tr>
              <tr>
                <td>Butler et al.<sup>13</sup></td>
                <td align="center">2023</td>
                <td align="center">A generalizable electrocardiogram-based artificial intelligence model for 10-year heart failure risk prediction</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">ARIC (EUA)</td>
                <td align="center">14 613</td>
                <td align="center">54,2</td>
                <td align="center">44,7%</td>
                <td align="center">10</td>
                <td align="center">Sim</td>
                <td align="center">MESA (USA)</td>
              </tr>
              <tr>
                <td>Sun et al.<sup>14</sup></td>
                <td align="center">2023</td>
                <td align="center">Towards artificial intelligence-based learning health system for population-level mortality prediction using electrocardiograms</td>
                <td align="center">Retrospectivo</td>
                <td align="center">DL</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">Alberta (Canad&#225;)</td>
                <td align="center">1 605 268</td>
                <td align="center">65,80 &#177; 17,25</td>
                <td align="center">56,81%</td>
                <td align="center">13</td>
                <td align="center">N&#227;o</td>
                <td align="center">NR</td>
              </tr>
              <tr>
                <td>Lin et al.<sup>15</sup></td>
                <td align="center">2025</td>
                <td align="center">ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram</td>
                <td align="center">Retrospectivo</td>
                <td align="center">CNN</td>
                <td align="center">ECG de 12 deriva&#231;&#245;es</td>
                <td align="center">CGMH (Taiwan)</td>
                <td align="center">4 932 573</td>
                <td align="center">60,5 &#177; 16,8</td>
                <td align="center">52,5%</td>
                <td align="center">12</td>
                <td align="center">Sim</td>
                <td align="center">Tri-Service General Hospital (Taiwan)</td>
              </tr>
            </tbody>
          </table>
          <table-wrap-foot>
            <fn id="TFN1">
              <p>IA: Intelig&#234;ncia Artificial; ARIC: Atherosclerosis Risk in Communities Study (Estudo de Risco de Aterosclerose em Comunidades); BIDMC: Beth Israel Deaconess Medical Center; BWH: Brigham and Women&#8217;s Hospital; CGMH: Chang Gung Memorial Hospital; CODE: Clinical Outcomes in Digital Electrocardiology; DL: Deep Learning (Aprendizado Profundo); DNN: Deep Neural Network (Rede Neural Profunda); ECG: Eletrocardiograma; ELSA-Brasil: Estudo Longitudinal de Sa&#250;de do Adulto; HECIUH: Health Examination Center at Inha University Hospital; HHMU: Hospital da Universidade M&#233;dica de Hebei; MESA: Multi-Ethnic Study of Atherosclerosis (Estudo Multi&#233;tnico de Aterosclerose); MGH: Massachusetts General Hospital; NR: N&#227;o reportado; SaMi-Trop: coorte brasileira de pacientes com doen&#231;a de Chagas; TSGH: Tri-Service General Hospital; UK Biobank: Biobanco do Reino Unido; YNHHS: Yale New Haven Health System.</p>
            </fn>
          </table-wrap-foot>
        </table-wrap>
      </p>
      <p>Entre os estudos inclu&#237;dos, a maioria (90,9%) empregou CNNs como a principal arquitetura para an&#225;lise dos sinais de ECG.<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref></sup> Especificamente, as CNNs foram utilizadas isoladamente ou em combina&#231;&#227;o com outros m&#233;todos, como camadas <italic>Long Short-Term Memory</italic> (LSTM), <italic>Residual Networks</italic> (ResNet) ou <italic>Variational Autoencoders</italic> (VAEs). Um estudo utilizou abordagens mais amplas de DL sem empregar explicitamente arquiteturas CNN, enquadrando-se na categoria de DNNs ou DL geral.<sup><xref ref-type="bibr" rid="B14">14</xref></sup></p>
      <p>Os tamanhos amostrais variaram substancialmente, de 238 pacientes a mais de 4 milh&#245;es de ECGs. As popula&#231;&#245;es dos estudos consistiam principalmente de pacientes adultos provenientes de sistemas hospitalares associados a grandes biobancos, com idades m&#233;dias geralmente variando entre 54 e 65 anos. No entanto, alguns estudos n&#227;o forneceram informa&#231;&#245;es demogr&#225;ficas detalhadas.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> Quanto ao per&#237;odo de seguimento, o mais longo relatado foi de 34 anos,<sup><xref ref-type="bibr" rid="B10">10</xref></sup> enquanto o mais curto foi de dois anos.<sup><xref ref-type="bibr" rid="B16">16</xref></sup></p>
      <p>Como mostrado na <xref ref-type="fig" rid="f03">Figura 2</xref>, a maioria das coortes utilizadas para treinar os modelos de IA era baseada nos Estados Unidos da Am&#233;rica (EUA).<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B12">12</xref></sup> Dois estudos utilizaram popula&#231;&#245;es de Taiwan,<sup><xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B17">17</xref></sup> enquanto um estudo incluiu popula&#231;&#245;es da China<sup><xref ref-type="bibr" rid="B10">10</xref></sup> e outro do Canad&#225;.<sup><xref ref-type="bibr" rid="B14">14</xref></sup></p>
      <p>
        <fig id="f03">
          <label>Figura 2</label>
          <caption>
            <title>&#8211; Distribui&#231;&#227;o geogr&#225;fica das coortes do estudo de IA-ECG. O mapa destaca a localiza&#231;&#227;o dos conjuntos de dados usados para treinamento (azul) e valida&#231;&#227;o externa (vermelho) dos modelos de IA-ECG. A maioria das coortes de treinamento originou-se dos EUA, enquanto os conjuntos de dados de valida&#231;&#227;o concentraram-se no Reino Unido, Brasil, Taiwan e EUA.</title>
          </caption>
          <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf02.tif"/></fig>
      </p>
      <p>Os estudos que empregaram valida&#231;&#227;o externa tamb&#233;m inclu&#237;ram participantes de pa&#237;ses de alta renda, como o Reino Unido<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup> e os EUA.<sup><xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B13">13</xref></sup> Apenas dois estudos utilizaram uma coorte do Brasil para valida&#231;&#227;o externa,<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B12">12</xref></sup> e dois estudos inclu&#237;ram popula&#231;&#245;es de Taiwan.<sup><xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref></sup></p>
      <p>Todos os desfechos relevantes para esta revis&#227;o est&#227;o apresentados na <xref ref-type="table" rid="t2">Tabela 2</xref>. Al&#233;m disso, as Figuras 3 e 4 exibem <italic>forest plots</italic> que resumem o desempenho preditivo dos modelos de IA aplicados ao ECG nos estudos inclu&#237;dos. A <xref ref-type="fig" rid="f04">Figura 3</xref> apresenta m&#233;tricas de discrimina&#231;&#227;o, incluindo o &#237;ndice de concord&#226;ncia (&#237;ndice-C) e a &#225;rea sob a curva ROC (AUC/AUROC), enquanto a <xref ref-type="fig" rid="f05">Figura 4</xref> mostra medidas de efeito expressas como <italic>hazard ratios</italic> (HR) para diversos desfechos cardiovasculares, incluindo intervalos de confian&#231;a de 95% quando dispon&#237;veis.</p>
      <p>
        <table-wrap id="t2">
          <label>Tabela 2</label>
          <caption>
            <title>&#8211; Resumo do desempenho da intelig&#234;ncia artificial aplicada na eletrocardiografia (IA-ECG) nos diferentes desfechos cl&#237;nicos</title>
          </caption>
          <table frame="hsides" rules="groups">
            <colgroup>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
              <col/>
            </colgroup>
            <thead>
              <tr>
                <th align="left">Autores</th>
                <th>Mortalidade por todas as causas</th>
                <th>Morte cardiovascular</th>
                <th>DCVA</th>
                <th>Insufici&#234;ncia card&#237;aca</th>
                <th>ICFEp</th>
                <th>IAM</th>
                <th>AVC</th>
                <th>MACE</th>
              </tr>
            </thead>
            <tbody>
              <tr>
                <td>Sau et al.<sup>6</sup></td>
                <td align="center">&#205;ndice-C 0,775 (0,773&#8211;0,776) Taxa de evento: NR</td>
                <td align="center">&#205;ndice-C 0,832 (0,831&#8211;0,834) Taxa de evento: NR</td>
                <td align="center">&#205;ndice-C 0,696 (0,694&#8211;0,698) Modelo SEER &#237;ndice-C 0,547 (0,527&#8211;0,567) Taxa de evento: NR</td>
                <td align="center">&#205;ndice-C 0,787 (0,785&#8211;0,789) Taxa de evento: NR</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Hughes et al.<sup>7</sup></td>
                <td align="center">-</td>
                <td align="center">Stanford (12-deriva&#231;&#245;es) AUC 0,832 (0,810&#8211;0,854), &#237;ndice-C 0,815 (0,798&#8211;0,826); Cedars-Sinai (12-deriva&#231;&#245;es) AUC 0,781 (0,767&#8211;0,795), &#237;ndice-C 0,777 (0,768&#8211;0,783); Columbia (12-deriva&#231;&#245;esd) AUC 0,825 (0,821&#8211;0,829), &#237;ndice-C 0,808 (0,806&#8211;0,810) Stanford (DI) AUC 0,797 (0,773&#8211;0,824), &#237;ndice-C 0,781 (0,760&#8211;0,793); Cedars-Sinai (DI) AUC 0,778 (0,765&#8211;0,792), &#237;ndice-C 0,771 (0,764&#8211;0,778); Columbia (DI) AUC 0,761 (0,757&#8211;0,765), &#237;ndice-C 0,755 (0,753&#8211;0,757) Taxa de evento: NR</td>
                <td align="center">Stanford (12-deriva&#231;&#245;es) AUC 0,668 (0,647&#8211;0,687), &#237;ndice-C 0,661 (0,652&#8211;0,669); Cedars-Sinai (12-deriva&#231;&#245;es) AUC 0,632 (0,591&#8211;0,674), &#237;ndice-C 0,635 (0,619&#8211;0,652)Taxa de evento: NR</td>
                <td align="center">Stanford HR 3,2 (2,9&#8211;3,5) Taxa de evento: NR</td>
                <td align="center">-</td>
                <td align="center">Stanford HR 2.0 (1.7&#8211;2.5) Taxa de evento: NR</td>
                <td align="center">Stanford HR 1.6 (1.4&#8211;1.9) Taxa de evento: NR</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Al-Alusi et al.<sup>8</sup></td>
                <td align="center">HR 1,47 (1,36&#8211;1,60), &#205;ndice-C 0,804 (0,796&#8211;0,813) Taxa de evento: 32,7% mortes no grupo de alto risco 13% no grupo de baixo risco</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">HR 2,26 (1,90&#8211;2,69), &#205;ndice-C 0,804 (0,796&#8211;0,813) Taxa de evento: 5,9%</td>
                <td align="center">-</td>
                <td align="center">HR 1,87 (1,69&#8211;2,07), &#205;ndice-C 0,733 (0,725&#8211;0,742) Taxa de evento: 14,1%</td>
                <td align="center">HR 1,30 (1,18&#8211;1,44) Taxa de evento: 6,5%</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Sau et al.<sup>9</sup></td>
                <td align="center">Homens HR 1,22 (1,19&#8211;1,25), mulheres HR 1,17 (1,19&#8211;1,31) Taxa de evento: 18,4%</td>
                <td align="center">Homens HR 1,00 (0,63&#8211;1,58), mulheres HR 1,78 (1,18&#8211;2,70) Taxa de evento: 3,7%</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Gao et al.<sup>16</sup></td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">ACC: 75% (Coorte A), 71,8% (Coorte B) Taxa de evento: NR</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Raghunath et al.<sup>10</sup></td>
                <td align="center">ECG AUC 0,855 Eventos: 14207</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Dhingra et al.<sup>11</sup></td>
                <td align="center">YNHHS HR 1,19 (1,15&#8211;1,24) UKB HR 2.13 (1,41&#8211;3,24) ELSA HR 3,64 (2,27&#8211;5,83) Taxa de evento: YNHHS 7,5%, UKB 0,8%, ELSA 1,7%</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">YNHHS HR 6,51 (6,11&#8211;6,93), &#237;ndice-C 0,718 (0,697&#8211;0,738); UKB HR 18,33 (9,90&#8211;33,97), &#237;ndice-C 0,769 (0,670&#8211;0,867); ELSA HR 32,06 (15,36&#8211;66,92), &#237;ndice-C 0,810 (0,714&#8211;0,907) Taxa de evento: YNHHS 4,2%</td>
                <td align="center">-</td>
                <td align="center">YNHHS HR 1,44 (1,04&#8211;2,00) UKB HR 3,16 (1,98&#8211;5,02) ELSA HR 3,53 (1,4&#8211;8,85) Taxa de evento: YNHHS 0,1%, UKB 0,5%, ELSA 0,4%</td>
                <td align="center">UKB HR 2,30 (1,36&#8211;3,9) ELSA HR 5,74 (2,59&#8211;12,72) Taxa de evento: YNHHS 1,6%, UKB 0,5%, ELSA 0,4%</td>
                <td align="center">YNHHS HR 2,10 (2,04&#8211;2,17) UKB HR 2,79 (2,17&#8211;3,6) ELSA HR 4,04 (2,77&#8211;5,89) Taxa de evento: YNHHS 10,4%, UKB 1,8%, ELSA 2,5%</td>
              </tr>
              <tr>
                <td>Lin et al.<sup>12</sup></td>
                <td align="center">AUROC 0,89, Sensibilidade 0,81, Especificidade 0,83, AUROC (Externa) 0,83 Taxa de evento: 6,3% (3 meses), 8,5% (6 meses), 10,0% (9 meses), 11,4% (1 ano)</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">AUROC 0,90 Sensibilidade 0,83 Especificidade 0,81 Taxa de evento: 0,6% (3 meses), 0,8% (6 meses), 1,1% (9 meses), 1,2% (1 ano)</td>
                <td align="center">-</td>
                <td align="center">AUROC 0,85 Sensibilidade 0,76 Especificidade 0,79 Taxa de evento: 0,3% (3 meses), 0,4% (6 meses), 0,5% (9 meses), 0,5% (1 ano)</td>
                <td align="center">AUROC 0,76 Sensibilidade 0,75 Especificidade 0,63 Taxa de evento: 0,3% (3 meses), 0,4% (6 meses), 0,6% (9 meses), 0,7% (1 ano)</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Butler et al.<sup>13</sup></td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">ARIC AUC 0,76 (0,72&#8211;0,83) MESA AUC 0,77 (0,74&#8211;0,79) Taxa de evento: 3,6% (MESA), 5,5% (ARIC)</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Sun et al.<sup>14</sup></td>
                <td align="center">30 dias: AUROC 0,843 (0,832&#8211;0,858), Brier Score 5,97 (5,83&#8211;6,12); 1 ano: AUROC 0,812 (0,807&#8211;0,81,6), Brier Score 11,44 (11,3-11,6); 5 anos: AUROC 0,798 (0,791&#8211;0,802), Brier Score 17,52 (17,29&#8211;17,83) Taxa de evento: 7,6% (30 dias) 17,3%, (1 ano) 32,9%, (5 anos)</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
              <tr>
                <td>Lin et al.<sup>15</sup></td>
                <td align="center">CGMH: &#205;ndice C 0,860 (0,859&#8211;0,861) TSGH: &#205;ndice C 0,813 (0,807&#8211;0,814) Taxa de evento: CGMH 5,9% (3 meses), 7,8% (6 meses), 9,3% (9 meses), 10,5% (1 ano); TSGH 6,8% (3 meses), 8,5% (6 meses), 9,5% (9 meses), 10,5% (1 ano)</td>
                <td align="center">CGMH: &#237;ndice-C 0,891 (0,890&#8211;0,893) Taxa de evento: NR</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
                <td align="center">-</td>
              </tr>
            </tbody>
          </table>
          <table-wrap-foot>
            <fn id="TFN2">
              <p>Os valores s&#227;o apresentados como estimativas pontuais com intervalos de confian&#231;a de 95% entre par&#234;nteses, salvo indica&#231;&#227;o em contr&#225;rio. ACC: acur&#225;cia; ARIC: Atherosclerosis Risk in Communities Study (Estudo de Risco de Aterosclerose em Comunidades); DCVA: doen&#231;a cardiovascular ateroscler&#243;tica; AUC/AUROC: &#225;rea sob a curva ROC; CGMH: Chang Gung Memorial Hospital; ECG: eletrocardiograma; ELSA-Brasil: Estudo Longitudinal de Sa&#250;de do Adulto; ICFEp: insufici&#234;ncia card&#237;aca com fra&#231;&#227;o de eje&#231;&#227;o preservada; HR: hazard ratio (raz&#227;o de risco); IAM: infarto agudo do mioc&#225;rdio; MACE: eventos cardiovasculares adversos maiores; MESA: Multi-Ethnic Study of Atherosclerosis (Estudo Multi&#233;tnico de Aterosclerose); NR: n&#227;o reportado; SEER: Surveillance, Epidemiology, and End Results (base de dados dos EUA); TSGH: Tri-Service General Hospital; UKB: UK Biobank (Biobanco do Reino Unido); YNHHS: Yale New Haven Health System</p>
            </fn>
          </table-wrap-foot>
        </table-wrap>
      </p>
      <p>
        <fig id="f04">
          <label>Figura 3</label>
          <caption>
            <title>&#8211; M&#233;tricas de discrimina&#231;&#227;o (AUC e &#237;ndice C) para modelos de IA-ECG. Gr&#225;fico resumindo o desempenho preditivo de modelos de IA baseados em ECG em diferentes estudos e horizontes temporais. Os marcadores azuis representam o &#237;ndice C (&#237;ndice de concord&#226;ncia de Harrell) e os marcadores laranja representam a &#225;rea sob a curva ROC (AUC/AUROC). Os dados s&#227;o apresentados como estimativas pontuais com intervalos de confian&#231;a de 95% (IC 95%). A linha vertical tracejada em 0,5 indica o limiar para predi&#231;&#227;o n&#227;o discriminat&#243;ria (aleat&#243;ria).</title>
          </caption>
          <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf03.tif"/></fig>
      </p>
      <p>
        <fig id="f05">
          <label>Figura 4</label>
          <caption>
            <title>&#8211; Raz&#245;es de risco para desfechos cl&#237;nicos associados a modelos de IA-ECG. Gr&#225;fico mostrando a associa&#231;&#227;o entre categorias de risco derivadas de IA e eventos cl&#237;nicos (mortalidade, insufici&#234;ncia card&#237;aca, eventos cardiovasculares adversos maiores, etc.). Os pontos representam as raz&#245;es de risco (HR) e as barras horizontais indicam os intervalos de confian&#231;a de 95% (IC 95%). A linha vertical tracejada em 1,0 representa a hip&#243;tese nula (nenhuma diferen&#231;a no risco).</title>
          </caption>
          <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf04.tif"/></fig>
      </p>
      <sec>
        <title>Mortalidade por todas as causas</title>
        <p>Oito estudos relataram resultados promissores na predi&#231;&#227;o de mortalidade por todas as causas utilizando diversos modelos de IA aplicados ao ECG. Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> demonstraram alta acur&#225;cia, com AUROC de 0,89 para mortalidade em 1 ano e 0,83 para mortalidade por todas as causas em 1 ano na valida&#231;&#227;o externa. Sun et al.<sup><xref ref-type="bibr" rid="B14">14</xref></sup> encontraram AUROCs de 0,843 para mortalidade em 30 dias, 0,812 para mortalidade em 1 ano e 0,798 para mortalidade em 5 anos, indicando desempenho robusto em m&#250;ltiplos horizontes temporais. Raghunath et al.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> relataram uma AUC de 0,855 utilizando apenas dados de ECG e 0,876 quando idade e sexo foram incorporados. No estudo Surv-ECG, Lin et al.<sup><xref ref-type="bibr" rid="B15">15</xref></sup> alcan&#231;aram um &#237;ndice-C de 0,860 para mortalidade por todas as causas. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> relataram um &#237;ndice-C de 0,775, refor&#231;ando ainda mais o potencial preditivo dos modelos baseados em IA para risco de mortalidade.</p>
        <p>Sau et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> identificaram HR de 1,22 para homens e 1,17 para mulheres, sugerindo que a discord&#226;ncia entre sexo biol&#243;gico e sexo estimado pelo ECG pode atuar como um marcador inovador de risco cardiovascular n&#227;o reconhecido. Corroborando esses achados, Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> relataram um HR de 1,19 para mortalidade por todas as causas utilizando um modelo de IA aplicado a imagens de ECG, em um estudo multinacional que tamb&#233;m demonstrou valor preditivo para outros desfechos cardiovasculares, incluindo IAM e IC. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> mostraram que indiv&#237;duos classificados como alto risco pelo modelo HTN-AI apresentaram uma taxa de mortalidade em 10 anos significativamente maior (21,0%) em compara&#231;&#227;o ao grupo de baixo risco (5,4%), destacando o valor da IA na identifica&#231;&#227;o de indiv&#237;duos de maior risco.</p>
        <p>Houve marcada heterogeneidade entre os estudos, com popula&#231;&#245;es variando de 286 880 a quase 5 milh&#245;es de ECGs e um tempo m&#233;dio de seguimento entre 3 e 34 anos. Embora v&#225;rios estudos tenham relatado valida&#231;&#227;o externa, na maioria dos casos tratava-se de replica&#231;&#245;es dentro do mesmo sistema de sa&#250;de ou biobanco. Apenas uma minoria testou seus modelos em popula&#231;&#245;es verdadeiramente independentes, com caracter&#237;sticas demogr&#225;ficas e cl&#237;nicas distintas.<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup></p>
      </sec>
      <sec>
        <title>Morte cardiovascular</title>
        <p>A capacidade dos modelos de IA aplicados ao ECG para prever mortalidade cardiovascular tamb&#233;m foi validada. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> relataram um &#237;ndice-C de 0,832 para a predi&#231;&#227;o de morte cardiovascular. Em Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> o modelo apresentou uma AUC de 0,83 e um &#237;ndice-C de 0,82 para morte cardiovascular em 5 anos utilizando o ECG de 12 deriva&#231;&#245;es completo, com AUC de 0,80 quando apenas a Deriva&#231;&#227;o I foi utilizada. Sau et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> relataram um HR de 1,78 para mulheres em compara&#231;&#227;o a 1,00 para homens entre indiv&#237;duos com ECG normal, refor&#231;ando o papel da discord&#226;ncia de sexo como um potencial preditor independente de mortalidade cardiovascular, particularmente em mulheres. Lin et al.,<sup><xref ref-type="bibr" rid="B15">15</xref></sup> no estudo Surv-ECG, refor&#231;aram ainda mais essa evid&#234;ncia ao relatar um &#237;ndice-C de 0,891 para predi&#231;&#227;o de morte cardiovascular.</p>
        <p>Os estudos que relataram morte cardiovascular tamb&#233;m apresentaram heterogeneidade significativa. As popula&#231;&#245;es diferiram em tamanho e dura&#231;&#227;o do seguimento. As taxas de eventos n&#227;o foram especificadas na maioria dos estudos, e muitas valida&#231;&#245;es classificadas como externas foram conduzidas em coortes derivadas do mesmo sistema de sa&#250;de ou de sistemas intimamente relacionados, levantando preocupa&#231;&#245;es quanto &#224; independ&#234;ncia dessas valida&#231;&#245;es.</p>
      </sec>
      <sec>
        <title>Doen&#231;a cardiovascular ateroscler&#243;tica</title>
        <p>Os modelos de IA aplicados ao ECG tamb&#233;m se mostraram eficazes na predi&#231;&#227;o de DCVA. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> relataram um &#237;ndice-C de 0,696 utilizando ECGs de 12 deriva&#231;&#245;es e deriva&#231;&#227;o I, integrados a dados cl&#237;nicos e gen&#233;ticos do UK Biobank. Em Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> o modelo alcan&#231;ou uma AUC de 0,67 para ECG de 12 deriva&#231;&#245;es e 0,63 para Deriva&#231;&#227;o I. No entanto, a predi&#231;&#227;o de eventos de DCVA apresentou ampla varia&#231;&#227;o entre os estudos em termos de tamanho amostral e defini&#231;&#245;es de evento (por exemplo, desfechos compostos incluindo IAM, AVC e morte s&#250;bita versus defini&#231;&#245;es mais restritas). Quanto &#224; valida&#231;&#227;o, a maioria ocorreu dentro de institui&#231;&#245;es ou bases de dados relacionadas, e apenas Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> realizaram valida&#231;&#227;o externa verdadeiramente independente.</p>
      </sec>
      <sec>
        <title>Infarto do mioc&#225;rdio</title>
        <p>Os modelos de IA aplicados ao ECG demonstraram desempenho promissor na predi&#231;&#227;o de risco de infarto agudo do mioc&#225;rdio (IAM). Hughes et al.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> relataram um AUROC de 0,85 para predi&#231;&#227;o de IAM utilizando o escore SEER, al&#233;m de um HR de 2,0. De forma semelhante, Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> demonstraram um HR de 1,87. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> tamb&#233;m relataram risco elevado, com HR de 1,44 para IAM incidente. Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> igualmente demonstraram alta acur&#225;cia, com AUROC de 0,85.</p>
        <p>Os estudos apresentaram heterogeneidade em seu delineamento, com diferen&#231;as nas defini&#231;&#245;es de evento, taxas de eventos e tamanhos amostrais variando de milhares a milh&#245;es de participantes. Em rela&#231;&#227;o &#224; valida&#231;&#227;o, a maioria permaneceu dentro dos mesmos biobancos ou institui&#231;&#245;es, e apenas Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> testaram seus modelos em coortes geograficamente distintas.</p>
      </sec>
      <sec>
        <title>Insufici&#234;ncia card&#237;aca</title>
        <p>Os modelos de IA aplicados ao ECG demonstraram desempenho robusto na predi&#231;&#227;o de IC, embora os estudos apresentem varia&#231;&#245;es substanciais em metodologia. O melhor desempenho foi observado em Lin et al.,<sup><xref ref-type="bibr" rid="B12">12</xref></sup> que relataram um AUROC de 0,90 para predi&#231;&#227;o de IC. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> encontraram HRs de 6,51 na coorte YNHHS, 18,33 no UK Biobank e 32,06 na coorte ELSA-Brasil para IC incidente, demonstrando forte capacidade progn&#243;stica em diferentes popula&#231;&#245;es. Em Sau et al.,<sup><xref ref-type="bibr" rid="B6">6</xref></sup> foi alcan&#231;ado um &#237;ndice-C de 0,787 utilizando ECGs de 12 deriva&#231;&#245;es e Deriva&#231;&#227;o I, destacando a confiabilidade do modelo.</p>
        <p>O modelo de IA-ECG desenvolvido por Butler et al.<sup><xref ref-type="bibr" rid="B13">13</xref></sup> alcan&#231;ou uma AUC de 0,76 na coorte ARIC e 0,77 na coorte MESA utilizando apenas dados de ECG, com melhora significativa quando vari&#225;veis cl&#237;nicas foram incorporadas. Quando combinado com 12 fatores de risco, o modelo AI-ECG-Cox atingiu uma AUC de 0,82 no estudo ARIC e 0,84 no MESA, demonstrando maior valor preditivo por meio da integra&#231;&#227;o de dados cl&#237;nicos. Em Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> o modelo apresentou sensibilidade de 76% e especificidade de 82% para predi&#231;&#227;o de IC, refor&#231;ando suas capacidades diagn&#243;sticas. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> relataram um HR de 2,26. Por fim, Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> relataram acur&#225;cia de 75% na Coorte A e 71,8% na Coorte B para insufici&#234;ncia card&#237;aca com fra&#231;&#227;o de eje&#231;&#227;o preservada, com sensibilidade (71,7%) e especificidade (71,9%) equilibradas.</p>
        <p>A maioria das valida&#231;&#245;es foi conduzida em bases de dados relacionadas; apenas um pequeno n&#250;mero de estudos<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup> validou seus modelos em popula&#231;&#245;es demograficamente distintas.</p>
      </sec>
      <sec>
        <title>Acidente vascular cerebral (AVC)</title>
        <p>Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> demonstraram o potencial dos modelos de IA aplicados ao ECG na predi&#231;&#227;o de risco de AVC, alcan&#231;ando um AUROC de 0,76. Em Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> indiv&#237;duos da coorte de Stanford com os maiores escores SEER apresentaram um aumento de 1,6 vez no HR para AVC incidente, ajustado por idade e sexo, destacando a capacidade do modelo de identificar precocemente indiv&#237;duos com risco cerebrovascular elevado. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> relataram um HR de 2,30 no UK Biobank, e Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> relataram um HR semelhante de 2,30.</p>
        <p>Os estudos variaram quanto &#224; inclus&#227;o apenas de AVC isqu&#234;mico ou de todos os subtipos de AVC, o que contribuiu para diferen&#231;as nas taxas de eventos. A valida&#231;&#227;o foi frequentemente limitada a replica&#231;&#245;es internas ou a bases de dados de institui&#231;&#245;es relacionadas, e apenas Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> realizaram valida&#231;&#227;o externa verdadeiramente independente.</p>
      </sec>
      <sec>
        <title>Correla&#231;&#227;o com escores tradicionais</title>
        <p>Conforme apresentado na <xref ref-type="table" rid="t3">Tabela 3</xref>, Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup>demonstraram discrimina&#231;&#227;o superior em compara&#231;&#227;o ao Framingham Risk Score (FRS) em m&#250;ltiplos desfechos. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> mostraram que seu modelo de IA-ECG superou a <italic>Pooled Cohort Equation</italic> para insufici&#234;ncia card&#237;aca (PCE-HF). Na coorte YNHHS, o modelo de IA-ECG alcan&#231;ou um &#237;ndice-C de 0,718, em compara&#231;&#227;o com 0,601 para o PCE-HF. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> relataram que seu modelo de IA-ECG superou o PCE-HF na predi&#231;&#227;o de DCVA, com um &#237;ndice-C de 0,696 versus 0,547 para o modelo SEER.</p>
        <p>
          <table-wrap id="t3">
            <label>Tabela 3</label>
            <caption>
              <title>&#8211; Resumo das associa&#231;&#245;es com escores cl&#237;nicos tradicionais</title>
            </caption>
            <table frame="hsides" rules="groups">
              <colgroup>
                <col/>
                <col/>
                <col/>
              </colgroup>
              <thead>
                <tr>
                  <th align="left">Autores</th>
                  <th>Estudo</th>
                  <th>Correla&#231;&#227;o com Escores Tradicionais</th>
                </tr>
              </thead>
              <tbody>
                <tr>
                  <td>Sau et al.<sup>6</sup></td>
                  <td align="center">Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study</td>
                  <td align="center">A IA-ECG (&#237;ndice-C 0,696) superou o SEER (&#237;ndice-C 0,547) e apresentou melhor discrimina&#231;&#227;o em compara&#231;&#227;o com modelos tradicionais baseados em risco.</td>
                </tr>
                <tr>
                  <td>Hughes et al.<sup>7</sup></td>
                  <td align="center">A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease</td>
                  <td align="center">Correla&#231;&#227;o com o PCE (Pearson r = 0,218); NRI: 17,8%; reclassificou 16% dos pacientes de baixo para moderado risco de DCVA (definido pelo PCE). Melhor discrimina&#231;&#227;o para mortalidade cardiovascular em compara&#231;&#227;o ao PCE, por&#233;m menor discrimina&#231;&#227;o para desfechos de DCVA</td>
                </tr>
                <tr>
                  <td>Al-Alusi et al.<sup>8</sup></td>
                  <td align="center">A deep learning digital biomarker to detect hypertension and stratify cardiovascular risk from the electrocardiogram</td>
                  <td align="center">Discrimina&#231;&#227;o compar&#225;vel ao PCP-HF para predi&#231;&#227;o de insufici&#234;ncia card&#237;aca (&#237;ndice-C 0,804 vs. 0,806). Discrimina&#231;&#227;o superior ao PCE para IAM (0,733 vs 0,681) e AVC (0,704 vs 0,683)</td>
                </tr>
                <tr>
                  <td>Sau et al.<sup>9</sup></td>
                  <td align="center">Artificial intelligence-enhanced electrocardiography for the identification of a sex-related cardiovascular risk continuum: a retrospective cohort study</td>
                  <td align="center">NR</td>
                </tr>
                <tr>
                  <td>Gao et al.<sup>16</sup></td>
                  <td align="center">Electrocardiograph analysis for risk assessment of hear t failure with preserved ejection fraction: A deep learning model</td>
                  <td align="center">NR</td>
                </tr>
                <tr>
                  <td>Raghunath et al.<sup>10</sup></td>
                  <td align="center">Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network</td>
                  <td align="center">O modelo IA-ECG (AUC 0,876) superou o FRS (AUC 0,648) e o CCI (AUC 0,816)</td>
                </tr>
                <tr>
                  <td>Dhingra et al.<sup>11</sup></td>
                  <td align="center">Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study</td>
                  <td align="center">YNHHS: &#237;ndice-C 0,718 (IA-ECG) vs. 0,601 (PCP-HF); melhor discrimina&#231;&#227;o e reclassifica&#231;&#227;o (NRI 21,9%)</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>12</sup></td>
                  <td align="center">A multitask deep learning model utilizing electrocardiograms for major cardiovascular adverse events prediction</td>
                  <td align="center">O modelo IA-ECG demonstrou discrimina&#231;&#227;o superior em compara&#231;&#227;o ao FRS em todos os desfechos (AUROC em 1 ano: IC 0,90 vs 0,64; IAM 0,85 vs 0,69; AVC 0,76 vs 0,61; mortalidade por todas as causas 0,89 vs 0,52)</td>
                </tr>
                <tr>
                  <td>Butler et al.<sup>13</sup></td>
                  <td align="center">A generalizable electrocardiogram-based artificial intelligence model for 10-year heart failure risk prediction</td>
                  <td align="center">O modelo IA-ECG foi compar&#225;vel ao FHS-HF (AUC 0,77 vs 0,74), com desempenho aprimorado quando combinado a vari&#225;veis cl&#237;nicas (AUC 0,84)</td>
                </tr>
                <tr>
                  <td>Sun et al.<sup>14</sup></td>
                  <td align="center">Towards artificial intelligence-based learning health system for population-level mortality prediction using electrocardiograms</td>
                  <td align="center">NR</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>15</sup></td>
                  <td align="center">ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram</td>
                  <td align="center">O IA-ECG superou modelos de Cox baseados em Framingham para morte cardiovascular (&#237;ndice-C 0,891 vs 0,734)</td>
                </tr>
              </tbody>
            </table>
            <table-wrap-foot>
              <fn id="TFN3">
                <p>IA: intelig&#234;ncia artificial; DCVA: doen&#231;a cardiovascular ateroscler&#243;tica; AUC/AUROC: &#225;rea sob a curva ROC; CCI: &#205;ndice de Comorbidades de Charlson; &#237;ndice-C: &#237;ndice de concord&#226;ncia; ECG: eletrocardiograma; FRS: Framingham Risk Score (Escore de Risco de Framingham); FRS-HF: Framingham Risk Score para Insufici&#234;ncia Card&#237;aca; HR: hazard ratio (raz&#227;o de risco); IAM: infarto do mioc&#225;rdio NRI: melhoria l&#237;quida de reclassifica&#231;&#227;o; NR: n&#227;o reportado; PCE: pooled cohort equations (equa&#231;&#245;es de coorte agrupadas); PCE-HF: pooled cohort equations para preven&#231;&#227;o de insufici&#234;ncia card&#237;aca; SEER: Surveillance, Epidemiology, and End Results (base de dados dos EUA).</p>
              </fn>
            </table-wrap-foot>
          </table-wrap>
        </p>
        <p>Raghunath et al.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> demonstraram desempenho superior na predi&#231;&#227;o de mortalidade por todas as causas em 1 ano, superando tanto o FRS (AUC 0,648) quanto o <italic>Charlson Comorbidity Index</italic> (CCI) (AUC 0,816), alcan&#231;ando uma AUC de 0,876 utilizando apenas dados de ECG. Em Butler et al.,<sup><xref ref-type="bibr" rid="B13">13</xref></sup> o FRS para insufici&#234;ncia card&#237;aca apresentou AUC de 0,78 na coorte ARIC e 0,74 na coorte MESA. Hughes et al.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> encontraram correla&#231;&#227;o significativa com o PCE-HF, relatando uma <italic>Net Reclassification Improvement</italic> (NRI) de 17,8%, reclassificando corretamente 16% dos pacientes inicialmente categorizados como baixo risco para a categoria de risco moderado.</p>
      </sec>
      <sec>
        <title>Qualidade dos estudos</title>
        <p>A qualidade dos estudos inclu&#237;dos foi avaliada utilizando a escala Newcastle&#8211;Ottawa (NOS), conforme apresentado na <xref ref-type="table" rid="t4">Tabela 4</xref>. A NOS avalia os estudos em tr&#234;s dom&#237;nios: sele&#231;&#227;o, comparabilidade e desfecho. Cada dom&#237;nio cont&#233;m crit&#233;rios espec&#237;ficos, com uma pontua&#231;&#227;o m&#225;xima poss&#237;vel de 9 pontos.</p>
        <p>
          <table-wrap id="t4">
            <label>Tabela 4</label>
            <caption>
              <title>&#8211; Avalia&#231;&#227;o da qualidade dos estudos pela Escala de Newcastle-Ottawa</title>
            </caption>
            <table frame="hsides" rules="groups">
              <colgroup>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
              </colgroup>
              <thead>
                <tr>
                  <th align="left" rowspan="2">Estudo</th>
                  <th colspan="4">Sele&#231;&#227;o</th>
                  <th colspan="2">Comparabilidade</th>
                  <th colspan="3">Desfecho</th>
                  <th>Escore total de qualidade</th>
                </tr>
                <tr>
                  <th>Representatividade da Coorte Exposta</th>
                  <th>Sele&#231;&#227;o da Coorte N&#227;o Exposta</th>
                  <th>Determina&#231;&#227;o da Exposi&#231;&#227;o</th>
                  <th>Desfecho Ausente no In&#237;cio do Estudo</th>
                  <th>Fator Principal</th>
                  <th>Fator adicional</th>
                  <th>Avalia&#231;&#227;o dos Desfechos</th>
                  <th>Dura&#231;&#227;o do Seguimento</th>
                  <th>Adequa&#231;&#227;o do Seguimento</th>
                  <th align="left" style="font-weight:normal"> </th>
                </tr>
              </thead>
              <tbody>
                <tr>
                  <td>Sau et al.<sup>6</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Hughes et al.<sup>7</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Al-Alusi et al.<sup>8</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Sau et al.<sup>9</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td> </td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">8</td>
                </tr>
                <tr>
                  <td>Gao et al.<sup>16</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td> </td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">8</td>
                </tr>
                <tr>
                  <td>Raghunath et al.<sup>10</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Dhingra et al.<sup>11</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>12</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Butler et al.<sup>13</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
                <tr>
                  <td>Sun et al.<sup>14</sup></td>
                  <td align="center">*</td>
                  <td> </td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">8</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>15</sup></td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">*</td>
                  <td align="center">9</td>
                </tr>
              </tbody>
            </table>
          </table-wrap>
        </p>
        <p>Oito estudos alcan&#231;aram a pontua&#231;&#227;o m&#225;xima de 9, indicando alta qualidade metodol&#243;gica. Esses estudos demonstraram forte sele&#231;&#227;o de coorte, comparabilidade adequada com base em fatores-chave e avalia&#231;&#227;o robusta dos desfechos, com per&#237;odos de seguimento apropriados. Tr&#234;s estudos obtiveram 8 pontos devido &#224; aus&#234;ncia de informa&#231;&#245;es sobre fatores de comparabilidade ou &#224; falta de confirma&#231;&#227;o de que o desfecho n&#227;o estava presente no in&#237;cio do estudo. No entanto, v&#225;rios estudos n&#227;o relataram caracter&#237;sticas demogr&#225;ficas essenciais, como idade, sexo ou etnia &#8211; fatores importantes para avaliar a validade externa e a equidade entre subgrupos. Embora essas omiss&#245;es n&#227;o sejam totalmente capturadas pela NOS, elas representam limita&#231;&#245;es relevantes ao considerar a generaliza&#231;&#227;o do desempenho de modelos de IA em popula&#231;&#245;es diversas. Apesar dessas quest&#245;es, a qualidade geral dos estudos foi alta, sustentando a confiabilidade dos achados da revis&#227;o.</p>
      </sec>
      <sec>
        <title>T&#233;cnicas de Interpretabilidade em Modelos de IA-ECG</title>
        <p>Entre os estudos inclu&#237;dos, t&#233;cnicas de interpretabilidade foram relatadas em nove dos onze artigos, conforme mostrado na <xref ref-type="table" rid="t5">Tabela 5</xref>. Os <italic>saliency maps</italic> foram os m&#233;todos mais frequentemente aplicados para destacar as regi&#245;es do tra&#231;ado de ECG mais relevantes para as predi&#231;&#245;es dos modelos (por exemplo, onda P, intervalo PR).<sup><xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref></sup> Outras estrat&#233;gias inclu&#237;ram <italic>SHapley Additive exPlanations</italic> (SHAP),<sup><xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref></sup> que forneceram atribui&#231;&#227;o de caracter&#237;sticas em modelos baseados em &#225;rvores ou CNN; abla&#231;&#227;o de neur&#244;nios e <italic>heatmaps</italic> de pesos de CNN,<sup><xref ref-type="bibr" rid="B16">16</xref></sup> que examinaram o comportamento interno do modelo por meio de perturba&#231;&#245;es; al&#233;m de autoencoders variacionais (VAEs), m&#233;dia de formas de onda e an&#225;lises de associa&#231;&#227;o biol&#243;gica (GWAS, PheWAS) em Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> e Sau et al.,<sup><xref ref-type="bibr" rid="B9">9</xref></sup>utilizadas para explorar plausibilidade biol&#243;gica e padr&#245;es de morfologia do ECG. Dois estudos n&#227;o relataram o uso de qualquer t&#233;cnica de interpretabilidade.<sup><xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup></p>
        <p>
          <table-wrap id="t5">
            <label>Tabela 5</label>
            <caption>
              <title>&#8211; T&#233;cnicas de interpretabilidade aplicadas nos estudos inclu&#237;dos</title>
            </caption>
            <table frame="hsides" rules="groups">
              <colgroup>
                <col/>
                <col/>
                <col/>
              </colgroup>
              <thead>
                <tr>
                  <th align="left">Estudo</th>
                  <th>T&#233;cnicas de Interpretabilidade</th>
                  <th>Descri&#231;&#227;o</th>
                </tr>
              </thead>
              <tbody>
                <tr>
                  <td>Sau et al.<sup>6</sup></td>
                  <td align="center">VAE, m&#233;dia de formas de onda, PheWAS, GWAS, modelos de Cox</td>
                  <td align="center">Vinculou as predi&#231;&#245;es da IA-ECG a caracter&#237;sticas biol&#243;gicas e mecanismos fisiopatol&#243;gicos plaus&#237;veis</td>
                </tr>
                <tr>
                  <td>Hughes et al.<sup>7</sup></td>
                  <td align="center">NR</td>
                  <td align="center">Comparou o desempenho com os escores de risco PCE e SEER, mas n&#227;o relatou m&#233;todos espec&#237;ficos de interpretabilidade</td>
                </tr>
                <tr>
                  <td>Al-Alusi et al.<sup>8</sup></td>
                  <td align="center">Mapas de sali&#234;ncia</td>
                  <td align="center">Utilizado para identificar segmentos da onda do ECG preditivos de hipertens&#227;o</td>
                </tr>
                <tr>
                  <td>Sau et al.<sup>9</sup></td>
                  <td align="center">VAE, m&#233;dia de formas de onda</td>
                  <td align="center">Visualizou a morfologia do ECG associada &#224; discord&#226;ncia de risco baseada em sexo; VAE utilizado para redu&#231;&#227;o de dimensionalidade</td>
                </tr>
                <tr>
                  <td>Gao et al.<sup>16</sup></td>
                  <td align="center">Abla&#231;&#227;o de neur&#244;nios, mapas de calor de pesos da CNN</td>
                  <td align="center">Perturbou neur&#244;nios nas camadas da CNN e visualizou os pesos associados; abordagem semelhante &#224; an&#225;lise de sensibilidade por oclus&#227;o</td>
                </tr>
                <tr>
                  <td>Raghunath et al.<sup>10</sup></td>
                  <td align="center">Mapas de sali&#234;ncia, feedback de especialistas</td>
                  <td align="center">Forneceu explica&#231;&#245;es visuais para ECGs normais sinalizados pela IA; incluiu feedback de cardiologistas para avaliar interpretabilidade</td>
                </tr>
                <tr>
                  <td>Dhingra et al.<sup>11</sup></td>
                  <td align="center">NR</td>
                  <td align="center">Utilizou IA baseada em imagens de ECG para predi&#231;&#227;o de risco; n&#227;o relatou m&#233;todos espec&#237;ficos de interpretabilidade</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>12</sup></td>
                  <td align="center">Mapas de sali&#234;ncia</td>
                  <td align="center">Identificou a onda P e o intervalo PR como principais preditores de mortalidade usando visualiza&#231;&#227;o de sali&#234;ncia</td>
                </tr>
                <tr>
                  <td>Butler et al.<sup>13</sup></td>
                  <td align="center">SHAP</td>
                  <td align="center">Aplicou SHAP a caracter&#237;sticas derivadas do ECG para estratifica&#231;&#227;o de risco de insufici&#234;ncia card&#237;aca; identificou os tra&#231;os de ECG mais relevantes</td>
                </tr>
                <tr>
                  <td>Sun et al.<sup>14</sup></td>
                  <td align="center">SHAP, mapas de sali&#234;ncia</td>
                  <td align="center">SHAP utilizado no XGB para atribui&#231;&#227;o de caracter&#237;sticas; Grad-CAM aplicado &#224; CNN para destacar segmentos-chave do ECG (PR, QRS, ST-T)</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>15</sup></td>
                  <td align="center">mapas de sali&#234;ncia</td>
                  <td align="center">Aplicou mapas de sali&#234;ncia visual para interpreta&#231;&#227;o do modelo</td>
                </tr>
              </tbody>
            </table>
            <table-wrap-foot>
              <fn id="TFN4">
                <p>CNN: rede neural convolucional; ECG: eletrocardiograma; Grad-CAM: Gradient-weighted Class Activation Mapping (mapeamento de ativa&#231;&#227;o de classe ponderado por gradiente); GWAS: Genome-Wide Association Study (estudo de associa&#231;&#227;o gen&#244;mica ampla); IA: intelig&#234;ncia artificial; NR: n&#227;o reportado; PCE: Pooled Cohort Equation (equa&#231;&#227;o de coorte agrupada); PheWAS: Phenome-Wide Association Study (estudo de associa&#231;&#227;o fenot&#237;pica ampla); PR: intervalo PR; QRS: complexo QRS; SEER: Surveillance, Epidemiology, and End Results (base de dados dos EUA); SHAP: SHapley Additive exPlanations; ST-T: segmento ST e onda T; VAE: Variational Autoencoder (autoencoder variacional); XGB: XGBoost.</p>
              </fn>
            </table-wrap-foot>
          </table-wrap>
        </p>
      </sec>
      <sec>
        <title>Aplica&#231;&#227;o cl&#237;nica e uso prospectivo da IA-ECG</title>
        <p>Quanto &#224; aplica&#231;&#227;o cl&#237;nica, a maioria dos estudos explorou o potencial dos modelos de IA-ECG para estratifica&#231;&#227;o precoce de risco, monitoramento de pacientes ou triagem em n&#237;vel populacional; entretanto, essas aplica&#231;&#245;es permaneceram retrospectivas ou te&#243;ricas. Notavelmente, apenas Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> conduziram uma aplica&#231;&#227;o prospectiva de seu modelo de DL: o algoritmo foi aplicado no momento da admiss&#227;o de novos pacientes com suspeita de insufici&#234;ncia card&#237;aca com fra&#231;&#227;o de eje&#231;&#227;o preservada, e as predi&#231;&#245;es foram comparadas com medidas invasivas de press&#227;o diast&#243;lica final do ventr&#237;culo esquerdo. Esse cen&#225;rio de valida&#231;&#227;o em tempo real representa o &#250;nico exemplo de uso do modelo em um contexto de tomada de decis&#227;o cl&#237;nica entre os estudos inclu&#237;dos. Embora constitua o &#250;nico cen&#225;rio de implementa&#231;&#227;o em tempo real, ele n&#227;o con valida&#231;&#227;o prospectiva para predi&#231;&#227;o de desfechos cl&#237;nicos.</p>
      </sec>
    </sec>
    <sec sec-type="discussion">
      <title>Discuss&#227;o</title>
      <p>Os estudos revisados demonstram avan&#231;os substanciais na aplica&#231;&#227;o de IA &#224; an&#225;lise de ECG para estratifica&#231;&#227;o de risco cardiovascular. T&#233;cnicas de DL, particularmente CNNs, mostraram capacidade de detectar altera&#231;&#245;es sutis no ECG associadas &#224; hipertens&#227;o, disfun&#231;&#227;o ventricular e fibrila&#231;&#227;o atrial &#8212; mesmo em tra&#231;ados considerados normais por especialistas. Esses achados refor&#231;am o potencial cl&#237;nico da IA-ECG para a detec&#231;&#227;o precoce de condi&#231;&#245;es assintom&#225;ticas ou subcl&#237;nicas, com a possibilidade de transformar a pr&#225;tica cl&#237;nica ao permitir, a partir de um &#250;nico exame, a identifica&#231;&#227;o de indiv&#237;duos de alto risco e a implementa&#231;&#227;o de interven&#231;&#245;es precoces.</p>
      <p>Paralelamente, outras linhas importantes de investiga&#231;&#227;o t&#234;m explorado biomarcadores derivados de ECG por IA, como a idade eletrocardiogr&#225;fica, que demonstraram fortes associa&#231;&#245;es com mortalidade e desfechos cardiovasculares em grandes coortes populacionais. Por exemplo, estudos anteriores mostraram que discrep&#226;ncias entre a idade estimada pelo ECG por IA e a idade cronol&#243;gica est&#227;o associadas a maior risco de mortalidade e eventos cardiovasculares incidentes, incluindo achados provenientes de coortes comunit&#225;rias.<sup><xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref></sup> Essas abordagens destacam o potencial mais amplo da IA-ECG para capturar sinais latentes de envelhecimento fisiol&#243;gico e biol&#243;gico. Embora tais estudos n&#227;o tenham sido inclu&#237;dos na presente revis&#227;o devido ao nosso foco pr&#233;-definido em modelos que predizem diretamente desfechos cl&#237;nicos a partir exclusivamente dos sinais de ECG, eles fornecem evid&#234;ncias complementares importantes que sustentam o valor progn&#243;stico da IA-ECG.</p>
      <p>Apesar desses avan&#231;os, importantes limita&#231;&#245;es metodol&#243;gicas comprometem compara&#231;&#245;es diretas entre os estudos. A heterogeneidade dos modelos de IA utilizados, a diversidade de m&#233;tricas de desempenho empregadas e a varia&#231;&#227;o nos desfechos cl&#237;nicos analisados dificultam a consolida&#231;&#227;o dos resultados. Uma considera&#231;&#227;o crucial diz respeito &#224; distin&#231;&#227;o entre validade estat&#237;stica e validade cl&#237;nica: embora m&#233;tricas como AUROC e HR indiquem alto desempenho, permanece essencial determinar se as anormalidades identificadas pelos modelos correspondem a fen&#243;tipos fisiologicamente significativos, com implica&#231;&#245;es pr&#225;ticas para o manejo cl&#237;nico. Caso contr&#225;rio, existe o risco de que os algoritmos detectem apenas assinaturas estat&#237;sticas sem relev&#226;ncia cl&#237;nica real.</p>
      <p>A an&#225;lise dos resultados mostra que, para mortalidade por todas as causas, os estudos utilizaram diferentes m&#233;tricas &#8211; como AUROC e HR &#8211; al&#233;m de per&#237;odos de seguimento variados (por exemplo, mortalidade em 1 ano versus 5 anos), o que dificulta compara&#231;&#245;es diretas. Ainda assim, nos estudos que utilizaram as mesmas m&#233;tricas, os modelos demonstraram desempenho consistente em todos os horizontes temporais. Limita&#231;&#227;o semelhante foi observada para mortalidade cardiovascular, embora todos os modelos tenham apresentado alta acur&#225;cia preditiva. Apenas dois estudos avaliaram DCVA, cada um empregando m&#233;tricas distintas, o que limitou compara&#231;&#245;es diretas; apesar de discrimina&#231;&#227;o ligeiramente menor em compara&#231;&#227;o com mortalidade por todas as causas, os resultados permaneceram satisfat&#243;rios. Para infarto agudo do mioc&#225;rdio, quatro estudos relataram desfechos: tr&#234;s utilizando valores de HR variando de 1,44 a 3,53 em diferentes coortes, e dois relatando AUROC em torno de 0,85, indicando desempenho preditivo consistente. Insufici&#234;ncia card&#237;aca esteve entre os desfechos com maior desempenho relatado, alcan&#231;ando AUROC de 0,90 e HR de 32,06. Por fim, alguns estudos tamb&#233;m avaliaram acidente vascular cerebral, relatando HR entre 1,6 e 2,3, refor&#231;ando a capacidade dos modelos de IA-ECG de capturar risco em m&#250;ltiplos desfechos cardiovasculares.</p>
      <p>Modelos de IA-ECG tendem a apresentar desempenho mais robusto para mortalidade por todas as causas em compara&#231;&#227;o com desfechos cardiovasculares mais espec&#237;ficos, como DCVA. Essa diferen&#231;a provavelmente reflete o fato de que a mortalidade por todas as causas abrange um espectro mais amplo de anormalidades sist&#234;micas e card&#237;acas que podem ser detect&#225;veis nos sinais de ECG, enquanto a doen&#231;a ateroscler&#243;tica nem sempre produz altera&#231;&#245;es el&#233;tricas ou estruturais mensur&#225;veis antes da ocorr&#234;ncia de eventos cl&#237;nicos. Como resultado, a aterosclerose subcl&#237;nica pode permanecer indetect&#225;vel por modelos baseados em ECG, limitando seu desempenho preditivo para DCVA. Al&#233;m disso, a predi&#231;&#227;o de DCVA apresentou heterogeneidade substancial entre os estudos, incluindo diferen&#231;as no tamanho das amostras e nas defini&#231;&#245;es de desfecho, o que restringe ainda mais a comparabilidade direta e pode contribuir para a variabilidade dos resultados relatados.</p>
      <p>Compara&#231;&#245;es entre modelos de IA-ECG e escores tradicionais de risco devem ser interpretadas com cautela, pois muitas ferramentas convencionais n&#227;o foram originalmente desenvolvidas para prever desfechos como mortalidade por todas as causas. Diferen&#231;as nos alvos de predi&#231;&#227;o podem explicar parcialmente as varia&#231;&#245;es nas m&#233;tricas de desempenho relatadas e limitar compara&#231;&#245;es diretas. Do ponto de vista cl&#237;nico, esses achados sugerem que modelos de IA-ECG e escores tradicionais de risco podem desempenhar pap&#233;is complementares: enquanto ferramentas convencionais permanecem mais adequadas para estimar risco espec&#237;fico de doen&#231;a (por exemplo, DCVA), a IA-ECG pode oferecer uma estratifica&#231;&#227;o de risco mais ampla ao capturar sinais fisiol&#243;gicos globais, apoiando uma avalia&#231;&#227;o mais integrada do risco do paciente.</p>
      <p>Apesar disso, persistem lacunas metodol&#243;gicas. Nem todos os estudos relataram o n&#250;mero de eventos cl&#237;nicos, alguns omitiram m&#233;tricas de calibra&#231;&#227;o e muitos concentraram-se em desfechos de curto prazo, o que limita a extrapola&#231;&#227;o para risco de longo prazo. Outro aspecto pouco explorado &#233; a generaliza&#231;&#227;o temporal: a maioria dos modelos foi desenvolvida em coortes est&#225;ticas, sem avaliar se o desempenho permanece est&#225;vel ao longo do tempo em popula&#231;&#245;es sujeitas a mudan&#231;as epidemiol&#243;gicas, terap&#234;uticas ou de estilo de vida. A atualiza&#231;&#227;o cont&#237;nua dos modelos e o monitoramento prospectivo ser&#227;o necess&#225;rios para garantir utilidade cl&#237;nica duradoura da IA-ECG.</p>
      <p>Al&#233;m disso, nossa sele&#231;&#227;o de estudos foi restrita a modelos que utilizam sinais de ECG como &#250;nica vari&#225;vel de entrada, o que aumenta a consist&#234;ncia, mas n&#227;o reflete completamente a pr&#225;tica cl&#237;nica real, na qual m&#250;ltiplos fatores do paciente s&#227;o considerados. Portanto, quando combinada com vari&#225;veis cl&#237;nicas tradicionais, a IA-ECG tem potencial para funcionar como um biomarcador digital, aprimorando a identifica&#231;&#227;o de indiv&#237;duos de alto risco e reduzindo o subdiagn&#243;stico. Pesquisas futuras devem avan&#231;ar integrando IA-ECG com dados laboratoriais, exames de imagem e at&#233; informa&#231;&#245;es provenientes de dispositivos vest&#237;veis, caminhando em dire&#231;&#227;o a uma abordagem multimodal capaz de personalizar a estratifica&#231;&#227;o de risco cardiovascular.</p>
      <p>A escalabilidade da IA-ECG representa uma grande vantagem, especialmente em contextos com recursos limitados, dada a acessibilidade e a simplicidade da aquisi&#231;&#227;o do ECG. No entanto, a maioria dos modelos foi desenvolvida e validada utilizando dados de pa&#237;ses de alta renda, predominantemente dos Estados Unidos, o que limita sua generaliza&#231;&#227;o. Em muitos casos, as valida&#231;&#245;es consistiram em replica&#231;&#245;es dentro de sistemas de sa&#250;de relacionados, biobancos ou popula&#231;&#245;es com perfis demogr&#225;ficos e cl&#237;nicos semelhantes, em vez de avalia&#231;&#245;es em coortes distintas e heterog&#234;neas. Distinguir entre replica&#231;&#227;o interna e valida&#231;&#227;o externa genu&#237;na &#233; essencial, pois esta &#250;ltima fornece o teste mais rigoroso da robustez e da transportabilidade dos modelos de IA-ECG.</p>
      <p>O estudo de Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> foi um dos poucos a realizar valida&#231;&#227;o externa em uma popula&#231;&#227;o geogr&#225;fica ou demograficamente distinta &#8211; especificamente no Brasil &#8211; fornecendo evid&#234;ncias mais fortes de generaliza&#231;&#227;o. Um segundo estudo, de Dhingra et al.,<sup><xref ref-type="bibr" rid="B11">11</xref></sup> tamb&#233;m incluiu uma coorte brasileira, confirmando que, mesmo em popula&#231;&#245;es com caracter&#237;sticas sociodemogr&#225;ficas e cl&#237;nicas diferentes, os modelos mantiveram bom desempenho. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> conduziram valida&#231;&#245;es interna e externa em popula&#231;&#245;es semelhantes, com pequenas diferen&#231;as, como maior propor&#231;&#227;o de mulheres e indiv&#237;duos negros na coorte externa. Ainda assim, os perfis de comorbidades eram amplamente compar&#225;veis entre os grupos. Por outro lado, Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> n&#227;o relataram caracter&#237;sticas detalhadas da popula&#231;&#227;o, limitando as compara&#231;&#245;es entre coortes. Da mesma forma, nos estudos de Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> e Lin et al.,<sup><xref ref-type="bibr" rid="B15">15</xref></sup> a valida&#231;&#227;o externa foi limitada devido &#224; aus&#234;ncia de informa&#231;&#245;es cl&#237;nicas, permitindo apenas a avalia&#231;&#227;o da mortalidade em um ano como desfecho.</p>
      <p>O n&#250;mero reduzido de valida&#231;&#245;es verdadeiramente independentes, combinado com dados demogr&#225;ficos incompletos, permanece uma barreira importante para a ampla ado&#231;&#227;o cl&#237;nica dessas ferramentas, pois dificulta a avalia&#231;&#227;o da diversidade populacional e reduz tanto a comparabilidade quanto a interpretabilidade dos resultados.</p>
      <p>Entre os pontos fortes dos estudos inclu&#237;dos est&#227;o o uso de grandes bases de dados, metodologias avan&#231;adas de IA &#8211; especialmente CNNs &#8211; e per&#237;odos de seguimento prolongados. As CNNs demonstraram not&#225;vel efic&#225;cia na an&#225;lise de sinais brutos de ECG, capturando caracter&#237;sticas temporais e espaciais com m&#237;nima necessidade de pr&#233;-processamento. Em contraste, estudos classificados como DNNs ou DL geral frequentemente careciam de especificidade arquitetural e ofereciam interpretabilidade limitada.</p>
      <p>Nesse contexto, a maioria dos estudos analisados empregou t&#233;cnicas de interpretabilidade para aumentar a transpar&#234;ncia dos modelos de IA aplicados ao ECG, refletindo a crescente demanda por sistemas explic&#225;veis. O m&#233;todo mais utilizado foi o de <italic>saliency maps</italic>, que gera <italic>heatmaps</italic> sobre o tra&#231;ado do ECG para destacar regi&#245;es presumivelmente relevantes para as predi&#231;&#245;es, como a onda P, o complexo QRS ou o intervalo PR. Apesar de sua popularidade, evid&#234;ncias da literatura em imagem m&#233;dica indicam que esses mapas podem ser inst&#225;veis, pouco reprodut&#237;veis e, em alguns casos, fornecer explica&#231;&#245;es enganosas que parecem plaus&#237;veis, mas n&#227;o refletem o verdadeiro racioc&#237;nio do modelo.<sup><xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref></sup></p>
      <p>O SHAP tamb&#233;m foi utilizado em v&#225;rios estudos. Esse m&#233;todo, fundamentado na teoria dos jogos, atribui valores quantitativos &#224; contribui&#231;&#227;o de cada vari&#225;vel para a sa&#237;da do modelo, permitindo quantificar a influ&#234;ncia de cada caracter&#237;stica. Diferentemente dos m&#233;todos puramente baseados em gradiente, o SHAP fornece uma estrutura de atribui&#231;&#227;o aditiva que integra interpreta&#231;&#245;es locais e globais, oferecendo maior consist&#234;ncia entre ambas. Sua capacidade de gerar visualiza&#231;&#245;es intuitivas &#8212; como gr&#225;ficos coloridos que destacam pixels importantes em imagens ou segmentos temporais relevantes em sinais de ECG &#8212; tornou-o particularmente valioso em aplica&#231;&#245;es biom&#233;dicas. Entre suas limita&#231;&#245;es est&#227;o o alto custo computacional, desafios relacionados a vari&#225;veis temporais altamente correlacionadas e poss&#237;veis inconsist&#234;ncias na captura de depend&#234;ncias entre caracter&#237;sticas.<sup><xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref></sup></p>
      <p>Outras abordagens de interpretabilidade, incluindo abla&#231;&#227;o de neur&#244;nios combinada com <italic>heatmaps</italic> de pesos de CNN, VAEs e an&#225;lises de associa&#231;&#227;o biol&#243;gica (por exemplo, GWAS, PheWAS), tamb&#233;m foram exploradas para fornecer insights adicionais sobre o comportamento dos modelos. Essas t&#233;cnicas podem esclarecer representa&#231;&#245;es latentes ou a plausibilidade biol&#243;gica das predi&#231;&#245;es, mas muitas vezes s&#227;o limitadas por complexidade t&#233;cnica, restri&#231;&#245;es de acesso ao modelo e vieses presentes nos dados.<sup><xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref></sup></p>
      <p>De modo geral, o cen&#225;rio atual indica que, embora m&#250;ltiplas t&#233;cnicas de interpretabilidade estejam dispon&#237;veis, sua integra&#231;&#227;o na pr&#225;tica cl&#237;nica ainda &#233; limitada. Como observado por Yanagawa e Sato,<sup><xref ref-type="bibr" rid="B22">22</xref></sup> esfor&#231;os futuros devem priorizar m&#233;todos que n&#227;o apenas esclare&#231;am as predi&#231;&#245;es dos modelos, mas tamb&#233;m aproximem a compreens&#227;o cl&#237;nica dos padr&#245;es do ECG, potencialmente revelando marcadores fisiol&#243;gicos sutis ou subtipos de doen&#231;a que n&#227;o s&#227;o facilmente identific&#225;veis por meio da an&#225;lise convencional. A combina&#231;&#227;o de interpretabilidade com valida&#231;&#227;o sistem&#225;tica e avalia&#231;&#227;o cl&#237;nica prospectiva pode transformar os modelos de IA de ferramentas puramente preditivas em instrumentos que apoiam ativamente o racioc&#237;nio cl&#237;nico e a gera&#231;&#227;o de hip&#243;teses.</p>
      <p>Notavelmente, apenas Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> testaram prospectivamente a IA-ECG em um cen&#225;rio do mundo real, aplicando o modelo no momento da admiss&#227;o para estimar o risco de ICFEp em compara&#231;&#227;o com medi&#231;&#245;es invasivas. Embora modelos como o AIRE<sup><xref ref-type="bibr" rid="B6">6</xref></sup> e o ECG-surv15<sup><xref ref-type="bibr" rid="B15">15</xref></sup> tenham demonstrado fortes associa&#231;&#245;es com caracter&#237;sticas fisiol&#243;gicas e sobrevida, a implementa&#231;&#227;o cl&#237;nica em larga escala ainda &#233; incipiente.</p>
      <p>Para que a IA-ECG fa&#231;a a transi&#231;&#227;o do uso experimental para a pr&#225;tica cl&#237;nica, &#233; essencial sua integra&#231;&#227;o com os sistemas de sa&#250;de e os prontu&#225;rios eletr&#244;nicos. Os marcos regulat&#243;rios precisam evoluir para avaliar os modelos n&#227;o apenas quanto &#224; acur&#225;cia, mas tamb&#233;m quanto &#224; equidade, transpar&#234;ncia e utilidade cl&#237;nica. Investimentos em infraestrutura, interoperabilidade e capacita&#231;&#227;o de profissionais de sa&#250;de s&#227;o necess&#225;rios para garantir o uso seguro e eficaz dessas tecnologias. Ensaios cl&#237;nicos prospectivos tamb&#233;m s&#227;o fundamentais, especialmente em contextos de preven&#231;&#227;o prim&#225;ria.</p>
      <p>Por fim, estrat&#233;gias de sa&#250;de p&#250;blica podem se beneficiar da IA-ECG ao possibilitar a detec&#231;&#227;o precoce de indiv&#237;duos de alto risco, potencialmente reduzindo custos de longo prazo e melhorando desfechos. No entanto, considera&#231;&#245;es &#233;ticas &#8211; como transpar&#234;ncia algor&#237;tmica, privacidade de dados, responsabilidade e acesso equitativo &#8211; precisam ser abordadas. Modelos treinados em popula&#231;&#245;es homog&#234;neas e privilegiadas correm o risco de perpetuar disparidades em sa&#250;de, a menos que sejam adaptados a contextos cl&#237;nicos e demogr&#225;ficos diversos.</p>
      <p>Em resumo, o uso efetivo da IA-ECG exige uma abordagem transdisciplinar que conecte inova&#231;&#227;o tecnol&#243;gica, pr&#225;tica cl&#237;nica, sa&#250;de p&#250;blica e &#233;tica. Somente sob essa perspectiva a IA-ECG pode evoluir de uma ferramenta promissora para um instrumento transformador na redu&#231;&#227;o de iniquidades em sa&#250;de, na otimiza&#231;&#227;o do cuidado e no avan&#231;o da sa&#250;de populacional.</p>
    </sec>
    <sec sec-type="conclusions">
      <title>Conclus&#227;o</title>
      <p>A aplica&#231;&#227;o de IA &#224; an&#225;lise de ECG representa um avan&#231;o promissor na avalia&#231;&#227;o personalizada do risco cardiovascular. No entanto, mais pesquisas s&#227;o necess&#225;rias para garantir a seguran&#231;a, a efic&#225;cia e a integra&#231;&#227;o cl&#237;nica equitativa dessas tecnologias. As evid&#234;ncias atuais, embora convincentes, destacam a necessidade de popula&#231;&#245;es mais diversas nos conjuntos de treinamento e valida&#231;&#227;o, maior interpretabilidade dos modelos e ensaios cl&#237;nicos prospectivos para avaliar plenamente o impacto no mundo real e a generaliza&#231;&#227;o dos modelos de IA-ECG. Somente por meio de valida&#231;&#227;o rigorosa e integra&#231;&#227;o cuidadosa a IA-ECG poder&#225; fazer a transi&#231;&#227;o de uma fronteira de pesquisa para uma ferramenta transformadora na medicina cardiovascular.</p>
    </sec>
  </body>
  <back>
    <ack>
      <title>Agradecimentos</title>
      <p>Os autores utilizaram o Grammarly, uma ferramenta de edi&#231;&#227;o de texto baseada em IA, para auxiliar na revis&#227;o do ingl&#234;s e aprimorar a clareza e a flu&#234;ncia do manuscrito. Todo o conte&#250;do foi revisado e aprovado pelos autores.</p>
    </ack>
    <ref-list>
      <title>Refer&#234;ncias</title>
      <ref id="B1">
        <label>1</label>
        <mixed-citation>1. World Health Organization. Noncommunicable diseases [Internet]. Geneva: World Health Organization; 2025 [cited 2026 Jun 08]. Available from: <ext-link ext-link-type="uri" xlink:href="https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases">https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases</ext-link>.</mixed-citation>
        <element-citation publication-type="webpage">
          <person-group person-group-type="author">
            <collab>World Health Organization</collab>
          </person-group>
          <source>Noncommunicable diseases</source>
          <comment>Internet</comment>
          <publisher-loc>Geneva</publisher-loc>
          <publisher-name>World Health Organization</publisher-name>
          <year>2025</year>
          <date-in-citation content-type="cited-date">cited 2026 Jun 08</date-in-citation>
          <ext-link ext-link-type="uri" xlink:href="https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases">https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases</ext-link>
        </element-citation>
      </ref>
      <ref id="B2">
        <label>2</label>
        <mixed-citation>2. L&#252;scher TF, Wenzl FA, D'Ascenzo F, Friedman PA, Antoniades C. Artificial Intelligence in Cardiovascular Medicine: Clinical Applications. Eur Heart J. 2024;45(40):4291-304. doi: 10.1093/eurheartj/ehae465.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>L&#252;scher</surname>
              <given-names>TF</given-names>
            </name>
            <name>
              <surname>Wenzl</surname>
              <given-names>FA</given-names>
            </name>
            <name>
              <surname>D'Ascenzo</surname>
              <given-names>F</given-names>
            </name>
            <name>
              <surname>Friedman</surname>
              <given-names>PA</given-names>
            </name>
            <name>
              <surname>Antoniades</surname>
              <given-names>C</given-names>
            </name>
          </person-group>
          <article-title>Artificial Intelligence in Cardiovascular Medicine: Clinical Applications</article-title>
          <source>Eur Heart J</source>
          <year>2024</year>
          <volume>45</volume>
          <issue>40</issue>
          <fpage>4291</fpage>
          <lpage>4304</lpage>
          <pub-id pub-id-type="doi">10.1093/eurheartj/ehae465</pub-id>
        </element-citation>
      </ref>
      <ref id="B3">
        <label>3</label>
        <mixed-citation>3. Arnett DK, Blumenthal RS, Albert MA, Buroker AB, Goldberger ZD, Hahn EJ, et al. 2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2019;140(11):e596-e646. doi: 10.1161/CIR.0000000000000678.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Arnett</surname>
              <given-names>DK</given-names>
            </name>
            <name>
              <surname>Blumenthal</surname>
              <given-names>RS</given-names>
            </name>
            <name>
              <surname>Albert</surname>
              <given-names>MA</given-names>
            </name>
            <name>
              <surname>Buroker</surname>
              <given-names>AB</given-names>
            </name>
            <name>
              <surname>Goldberger</surname>
              <given-names>ZD</given-names>
            </name>
            <name>
              <surname>Hahn</surname>
              <given-names>EJ</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines</article-title>
          <source>Circulation</source>
          <year>2019</year>
          <volume>140</volume>
          <issue>11</issue>
          <fpage>e596</fpage>
          <lpage>e646</lpage>
          <pub-id pub-id-type="doi">10.1161/CIR.0000000000000678</pub-id>
        </element-citation>
      </ref>
      <ref id="B4">
        <label>4</label>
        <mixed-citation>4. Talha I, Elkhoudri N, Hilali A. Major Limitations of Cardiovascular Risk Scores. Cardiovasc Ther. 2024;2024:4133365. doi: 10.1155/2024/4133365.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Talha</surname>
              <given-names>I</given-names>
            </name>
            <name>
              <surname>Elkhoudri</surname>
              <given-names>N</given-names>
            </name>
            <name>
              <surname>Hilali</surname>
              <given-names>A</given-names>
            </name>
          </person-group>
          <article-title>Major Limitations of Cardiovascular Risk Scores</article-title>
          <source>Cardiovasc Ther</source>
          <year>2024</year>
          <volume>2024</volume>
          <size units="pages">4133365</size>
          <pub-id pub-id-type="doi">10.1155/2024/4133365</pub-id>
        </element-citation>
      </ref>
      <ref id="B5">
        <label>5</label>
        <mixed-citation>5. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. BMJ. 2021;372:n71. doi: 10.1136/bmj.n71.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Page</surname>
              <given-names>MJ</given-names>
            </name>
            <name>
              <surname>McKenzie</surname>
              <given-names>JE</given-names>
            </name>
            <name>
              <surname>Bossuyt</surname>
              <given-names>PM</given-names>
            </name>
            <name>
              <surname>Boutron</surname>
              <given-names>I</given-names>
            </name>
            <name>
              <surname>Hoffmann</surname>
              <given-names>TC</given-names>
            </name>
            <name>
              <surname>Mulrow</surname>
              <given-names>CD</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews</article-title>
          <source>BMJ</source>
          <year>2021</year>
          <volume>372</volume>
          <size units="pages">n71</size>
          <pub-id pub-id-type="doi">10.1136/bmj.n71</pub-id>
        </element-citation>
      </ref>
      <ref id="B6">
        <label>6</label>
        <mixed-citation>6. Sau A, Pastika L, Sieliwonczyk E, Patlatzoglou K, Ribeiro AH, McGurk KA, et al. Artificial Intelligence-Enabled Electrocardiogram for Mortality and Cardiovascular Risk Estimation: A Model Development and Validation Study. Lancet Digit Health. 2024;6(11):e791-e802. doi: 10.1016/S2589-7500(24)00172-9.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Sau</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Pastika</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>Sieliwonczyk</surname>
              <given-names>E</given-names>
            </name>
            <name>
              <surname>Patlatzoglou</surname>
              <given-names>K</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>AH</given-names>
            </name>
            <name>
              <surname>McGurk</surname>
              <given-names>KA</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Artificial Intelligence-Enabled Electrocardiogram for Mortality and Cardiovascular Risk Estimation: A Model Development and Validation Study</article-title>
          <source>Lancet Digit Health</source>
          <year>2024</year>
          <volume>6</volume>
          <issue>11</issue>
          <fpage>e791</fpage>
          <lpage>e802</lpage>
          <pub-id pub-id-type="doi">10.1016/S2589-7500(24)00172-9</pub-id>
        </element-citation>
      </ref>
      <ref id="B7">
        <label>7</label>
        <mixed-citation>7. Hughes JW, Tooley J, Soto JT, Ostropolets A, Poterucha T, Christensen MK, et al. A Deep Learning-Based Electrocardiogram Risk Score for Long Term Cardiovascular Death and Disease. NPJ Digit Med. 2023;6(1):169. doi: 10.1038/s41746-023-00916-6.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Hughes</surname>
              <given-names>JW</given-names>
            </name>
            <name>
              <surname>Tooley</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Soto</surname>
              <given-names>JT</given-names>
            </name>
            <name>
              <surname>Ostropolets</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Poterucha</surname>
              <given-names>T</given-names>
            </name>
            <name>
              <surname>Christensen</surname>
              <given-names>MK</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>A Deep Learning-Based Electrocardiogram Risk Score for Long Term Cardiovascular Death and Disease</article-title>
          <source>NPJ Digit Med</source>
          <year>2023</year>
          <volume>6</volume>
          <issue>1</issue>
          <size units="pages">169</size>
          <pub-id pub-id-type="doi">10.1038/s41746-023-00916-6</pub-id>
        </element-citation>
      </ref>
      <ref id="B8">
        <label>8</label>
        <mixed-citation>8. Al-Alusi MA, Friedman SF, Kany S, R&#228;m&#246; JT, Pipilas D, Singh P, et al. A Deep Learning Digital Biomarker to Detect Hypertension and Stratify Cardiovascular Risk from the Electrocardiogram. NPJ Digit Med. 2025;8(1):120. doi: 10.1038/s41746-025-01491-8.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Al-Alusi</surname>
              <given-names>MA</given-names>
            </name>
            <name>
              <surname>Friedman</surname>
              <given-names>SF</given-names>
            </name>
            <name>
              <surname>Kany</surname>
              <given-names>S</given-names>
            </name>
            <name>
              <surname>R&#228;m&#246;</surname>
              <given-names>JT</given-names>
            </name>
            <name>
              <surname>Pipilas</surname>
              <given-names>D</given-names>
            </name>
            <name>
              <surname>Singh</surname>
              <given-names>P</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>A Deep Learning Digital Biomarker to Detect Hypertension and Stratify Cardiovascular Risk from the Electrocardiogram</article-title>
          <source>NPJ Digit Med</source>
          <year>2025</year>
          <volume>8</volume>
          <issue>1</issue>
          <size units="pages">120</size>
          <pub-id pub-id-type="doi">10.1038/s41746-025-01491-8</pub-id>
        </element-citation>
      </ref>
      <ref id="B9">
        <label>9</label>
        <mixed-citation>9. Sau A, Sieliwonczyk E, Patlatzoglou K, Pastika L, McGurk KA, Ribeiro AH, et al. Artificial Intelligence-Enhanced Electrocardiography for the Identification of a Sex-Related Cardiovascular Risk Continuum: A Retrospective Cohort Study. Lancet Digit Health. 2025;7(3):e184-94. doi: 10.1016/j.landig.2024.12.003.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Sau</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Sieliwonczyk</surname>
              <given-names>E</given-names>
            </name>
            <name>
              <surname>Patlatzoglou</surname>
              <given-names>K</given-names>
            </name>
            <name>
              <surname>Pastika</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>McGurk</surname>
              <given-names>KA</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>AH</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Artificial Intelligence-Enhanced Electrocardiography for the Identification of a Sex-Related Cardiovascular Risk Continuum: A Retrospective Cohort Study</article-title>
          <source>Lancet Digit Health</source>
          <year>2025</year>
          <volume>7</volume>
          <issue>3</issue>
          <fpage>e184</fpage>
          <lpage>e194</lpage>
          <pub-id pub-id-type="doi">10.1016/j.landig.2024.12.003</pub-id>
        </element-citation>
      </ref>
      <ref id="B10">
        <label>10</label>
        <mixed-citation>10. Raghunath S, Cerna AEU, Jing L, vanMaanen DP, Stough J, Hartzel DN, et al. Prediction of Mortality from 12-Lead Electrocardiogram Voltage Data Using a Deep Neural Network. Nat Med. 2020;26(6):886-91. doi: 10.1038/s41591-020-0870-z.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Raghunath</surname>
              <given-names>S</given-names>
            </name>
            <name>
              <surname>Cerna</surname>
              <given-names>AEU</given-names>
            </name>
            <name>
              <surname>Jing</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>vanMaanen</surname>
              <given-names>DP</given-names>
            </name>
            <name>
              <surname>Stough</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Hartzel</surname>
              <given-names>DN</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Prediction of Mortality from 12-Lead Electrocardiogram Voltage Data Using a Deep Neural Network</article-title>
          <source>Nat Med</source>
          <year>2020</year>
          <volume>26</volume>
          <issue>6</issue>
          <fpage>886</fpage>
          <lpage>891</lpage>
          <pub-id pub-id-type="doi">10.1038/s41591-020-0870-z</pub-id>
        </element-citation>
      </ref>
      <ref id="B11">
        <label>11</label>
        <mixed-citation>11. Dhingra LS, Aminorroaya A, Sangha V, Pedroso AF, Asselbergs FW, Brant LCC, et al. Heart Failure Risk Stratification Using Artificial Intelligence Applied to Electrocardiogram Images: A Multinational Study. Eur Heart J. 2025;46(11):1044-53. doi: 10.1093/eurheartj/ehae914.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Dhingra</surname>
              <given-names>LS</given-names>
            </name>
            <name>
              <surname>Aminorroaya</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Sangha</surname>
              <given-names>V</given-names>
            </name>
            <name>
              <surname>Pedroso</surname>
              <given-names>AF</given-names>
            </name>
            <name>
              <surname>Asselbergs</surname>
              <given-names>FW</given-names>
            </name>
            <name>
              <surname>Brant</surname>
              <given-names>LCC</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Heart Failure Risk Stratification Using Artificial Intelligence Applied to Electrocardiogram Images: A Multinational Study</article-title>
          <source>Eur Heart J</source>
          <year>2025</year>
          <volume>46</volume>
          <issue>11</issue>
          <fpage>1044</fpage>
          <lpage>1053</lpage>
          <pub-id pub-id-type="doi">10.1093/eurheartj/ehae914</pub-id>
        </element-citation>
      </ref>
      <ref id="B12">
        <label>12</label>
        <mixed-citation>12. Lin CH, Liu ZY, Chu PH, Chen JS, Wu HH, Wen MS, et al. A Multitask Deep Learning Model Utilizing Electrocardiograms for Major Cardiovascular Adverse Events Prediction. NPJ Digit Med. 2025;8(1):1. doi: 10.1038/s41746-024-01410-3.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Lin</surname>
              <given-names>CH</given-names>
            </name>
            <name>
              <surname>Liu</surname>
              <given-names>ZY</given-names>
            </name>
            <name>
              <surname>Chu</surname>
              <given-names>PH</given-names>
            </name>
            <name>
              <surname>Chen</surname>
              <given-names>JS</given-names>
            </name>
            <name>
              <surname>Wu</surname>
              <given-names>HH</given-names>
            </name>
            <name>
              <surname>Wen</surname>
              <given-names>MS</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>A Multitask Deep Learning Model Utilizing Electrocardiograms for Major Cardiovascular Adverse Events Prediction</article-title>
          <source>NPJ Digit Med</source>
          <year>2025</year>
          <volume>8</volume>
          <issue>1</issue>
          <size units="pages">1</size>
          <pub-id pub-id-type="doi">10.1038/s41746-024-01410-3</pub-id>
        </element-citation>
      </ref>
      <ref id="B13">
        <label>13</label>
        <mixed-citation>13. Butler L, Karabayir I, Kitzman DW, Alonso A, Tison GH, Chen LY, et al. A Generalizable Electrocardiogram-Based Artificial Intelligence Model for 10-Year Heart Failure Risk Prediction. Cardiovasc Digit Health J. 2023;4(6):183-90. doi: 10.1016/j.cvdhj.2023.11.003.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Butler</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>Karabayir</surname>
              <given-names>I</given-names>
            </name>
            <name>
              <surname>Kitzman</surname>
              <given-names>DW</given-names>
            </name>
            <name>
              <surname>Alonso</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Tison</surname>
              <given-names>GH</given-names>
            </name>
            <name>
              <surname>Chen</surname>
              <given-names>LY</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>A Generalizable Electrocardiogram-Based Artificial Intelligence Model for 10-Year Heart Failure Risk Prediction</article-title>
          <source>Cardiovasc Digit Health J</source>
          <year>2023</year>
          <volume>4</volume>
          <issue>6</issue>
          <fpage>183</fpage>
          <lpage>190</lpage>
          <pub-id pub-id-type="doi">10.1016/j.cvdhj.2023.11.003</pub-id>
        </element-citation>
      </ref>
      <ref id="B14">
        <label>14</label>
        <mixed-citation>14. Sun W, Kalmady SV, Sepehrvand N, Salimi A, Nademi Y, Bainey K, et al. Towards Artificial Intelligence-Based Learning Health System for Population-Level Mortality Prediction Using Electrocardiograms. NPJ Digit Med. 2023;6(1):21. doi: 10.1038/s41746-023-00765-3.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Sun</surname>
              <given-names>W</given-names>
            </name>
            <name>
              <surname>Kalmady</surname>
              <given-names>SV</given-names>
            </name>
            <name>
              <surname>Sepehrvand</surname>
              <given-names>N</given-names>
            </name>
            <name>
              <surname>Salimi</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Nademi</surname>
              <given-names>Y</given-names>
            </name>
            <name>
              <surname>Bainey</surname>
              <given-names>K</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Towards Artificial Intelligence-Based Learning Health System for Population-Level Mortality Prediction Using Electrocardiograms</article-title>
          <source>NPJ Digit Med</source>
          <year>2023</year>
          <volume>6</volume>
          <issue>1</issue>
          <size units="pages">21</size>
          <pub-id pub-id-type="doi">10.1038/s41746-023-00765-3</pub-id>
        </element-citation>
      </ref>
      <ref id="B15">
        <label>15</label>
        <mixed-citation>15. Lin CH, Liu ZY, Chen JS, Fann YC, Wen MS, Kuo CF. ECG-Surv: A Deep Learning-Based Model to Predict Time to 1-Year Mortality from 12-Lead Electrocardiogram. Biomed J. 2025;48(1):100732. doi: 10.1016/j.bj.2024.100732.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Lin</surname>
              <given-names>CH</given-names>
            </name>
            <name>
              <surname>Liu</surname>
              <given-names>ZY</given-names>
            </name>
            <name>
              <surname>Chen</surname>
              <given-names>JS</given-names>
            </name>
            <name>
              <surname>Fann</surname>
              <given-names>YC</given-names>
            </name>
            <name>
              <surname>Wen</surname>
              <given-names>MS</given-names>
            </name>
            <name>
              <surname>Kuo</surname>
              <given-names>CF</given-names>
            </name>
          </person-group>
          <article-title>ECG-Surv: A Deep Learning-Based Model to Predict Time to 1-Year Mortality from 12-Lead Electrocardiogram</article-title>
          <source>Biomed J</source>
          <year>2025</year>
          <volume>48</volume>
          <issue>1</issue>
          <size units="pages">100732</size>
          <pub-id pub-id-type="doi">10.1016/j.bj.2024.100732</pub-id>
        </element-citation>
      </ref>
      <ref id="B16">
        <label>16</label>
        <mixed-citation>16. Gao Z, Yang Y, Yang Z, Zhang X, Liu C. Electrocardiograph Analysis for Risk Assessment of Heart Failure with Preserved Ejection Fraction: A Deep Learning Model. ESC Heart Fail. 2025;12(1):631-9. doi: 10.1002/ehf2.15120</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Gao</surname>
              <given-names>Z</given-names>
            </name>
            <name>
              <surname>Yang</surname>
              <given-names>Y</given-names>
            </name>
            <name>
              <surname>Yang</surname>
              <given-names>Z</given-names>
            </name>
            <name>
              <surname>Zhang</surname>
              <given-names>X</given-names>
            </name>
            <name>
              <surname>Liu</surname>
              <given-names>C</given-names>
            </name>
          </person-group>
          <article-title>Electrocardiograph Analysis for Risk Assessment of Heart Failure with Preserved Ejection Fraction: A Deep Learning Model</article-title>
          <source>ESC Heart Fail</source>
          <year>2025</year>
          <volume>12</volume>
          <issue>1</issue>
          <fpage>631</fpage>
          <lpage>639</lpage>
          <pub-id pub-id-type="doi">10.1002/ehf2.15120</pub-id>
        </element-citation>
      </ref>
      <ref id="B17">
        <label>17</label>
        <mixed-citation>17. Lima EM, Ribeiro AH, Paix&#227;o GMM, Ribeiro MH, Pinto-Filho MM, Gomes PR, et al. Deep Neural Network-Estimated Electrocardiographic Age as a Mortality Predictor. Nat Commun. 2021;12(1):5117. doi: 10.1038/s41467-021-25351-7.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Lima</surname>
              <given-names>EM</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>AH</given-names>
            </name>
            <name>
              <surname>Paix&#227;o</surname>
              <given-names>GMM</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>MH</given-names>
            </name>
            <name>
              <surname>Pinto-Filho</surname>
              <given-names>MM</given-names>
            </name>
            <name>
              <surname>Gomes</surname>
              <given-names>PR</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Deep Neural Network-Estimated Electrocardiographic Age as a Mortality Predictor</article-title>
          <source>Nat Commun</source>
          <year>2021</year>
          <volume>12</volume>
          <issue>1</issue>
          <size units="pages">5117</size>
          <pub-id pub-id-type="doi">10.1038/s41467-021-25351-7</pub-id>
        </element-citation>
      </ref>
      <ref id="B18">
        <label>18</label>
        <mixed-citation>18. Brant LCC, Ribeiro AH, Pinto-Filho MM, Kornej J, Preis SR, Fetterman JL, et al. Association between Electrocardiographic Age and Cardiovascular Events in Community Settings: The Framingham Heart Study. Circ Cardiovasc Qual Outcomes. 2023;16(7):e009821. doi: 10.1161/CIRCOUTCOMES.122.009821.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Brant</surname>
              <given-names>LCC</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>AH</given-names>
            </name>
            <name>
              <surname>Pinto-Filho</surname>
              <given-names>MM</given-names>
            </name>
            <name>
              <surname>Kornej</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Preis</surname>
              <given-names>SR</given-names>
            </name>
            <name>
              <surname>Fetterman</surname>
              <given-names>JL</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Association between Electrocardiographic Age and Cardiovascular Events in Community Settings: The Framingham Heart Study</article-title>
          <source>Circ Cardiovasc Qual Outcomes</source>
          <year>2023</year>
          <volume>16</volume>
          <issue>7</issue>
          <elocation-id>e009821</elocation-id>
          <pub-id pub-id-type="doi">10.1161/CIRCOUTCOMES.122.009821</pub-id>
        </element-citation>
      </ref>
      <ref id="B19">
        <label>19</label>
        <mixed-citation>19. Baek YS, Lee DH, Jo Y, Lee SC, Choi W, Kim DH. Artificial Intelligence-Estimated Biological Heart Age Using a 12-Lead Electrocardiogram Predicts Mortality and Cardiovascular Outcomes. Front Cardiovasc Med. 2023;10:1137892. doi: 10.3389/fcvm.2023.1137892.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Baek</surname>
              <given-names>YS</given-names>
            </name>
            <name>
              <surname>Lee</surname>
              <given-names>DH</given-names>
            </name>
            <name>
              <surname>Jo</surname>
              <given-names>Y</given-names>
            </name>
            <name>
              <surname>Lee</surname>
              <given-names>SC</given-names>
            </name>
            <name>
              <surname>Choi</surname>
              <given-names>W</given-names>
            </name>
            <name>
              <surname>Kim</surname>
              <given-names>DH</given-names>
            </name>
          </person-group>
          <article-title>Artificial Intelligence-Estimated Biological Heart Age Using a 12-Lead Electrocardiogram Predicts Mortality and Cardiovascular Outcomes</article-title>
          <source>Front Cardiovasc Med</source>
          <year>2023</year>
          <volume>10</volume>
          <size units="pages">1137892</size>
          <pub-id pub-id-type="doi">10.3389/fcvm.2023.1137892</pub-id>
        </element-citation>
      </ref>
      <ref id="B20">
        <label>20</label>
        <mixed-citation>20. Bozzi ICRS, Lima MCAG, Ribeiro ALP, Paix&#227;o GMM. Artificial Intelligence-Derived ECG-Age as a Predictor of Mortality and Cardiovascular Events: A Systematic Review and Meta-Analysis. Arq Bras Cardiol. 2026;123(4):e20250650. doi: 10.36660/abc.20250650.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Bozzi</surname>
              <given-names>ICRS</given-names>
            </name>
            <name>
              <surname>Lima</surname>
              <given-names>MCAG</given-names>
            </name>
            <name>
              <surname>Ribeiro</surname>
              <given-names>ALP</given-names>
            </name>
            <name>
              <surname>Paix&#227;o</surname>
              <given-names>GMM</given-names>
            </name>
          </person-group>
          <article-title>Artificial Intelligence-Derived ECG-Age as a Predictor of Mortality and Cardiovascular Events: A Systematic Review and Meta-Analysis</article-title>
          <source>Arq Bras Cardiol</source>
          <year>2026</year>
          <volume>123</volume>
          <issue>4</issue>
          <elocation-id>e20250650</elocation-id>
          <pub-id pub-id-type="doi">10.36660/abc.20250650</pub-id>
        </element-citation>
      </ref>
      <ref id="B21">
        <label>21</label>
        <mixed-citation>21. Zhang J, Chao H, Dasegowda G, Wang G, Kalra MK, Yan P. Revisiting the Trustworthiness of Saliency Methods in Radiology AI. Radiol Artif Intell. 2024;6(1):e220221. doi: 10.1148/ryai.220221.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Zhang</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Chao</surname>
              <given-names>H</given-names>
            </name>
            <name>
              <surname>Dasegowda</surname>
              <given-names>G</given-names>
            </name>
            <name>
              <surname>Wang</surname>
              <given-names>G</given-names>
            </name>
            <name>
              <surname>Kalra</surname>
              <given-names>MK</given-names>
            </name>
            <name>
              <surname>Yan</surname>
              <given-names>P</given-names>
            </name>
          </person-group>
          <article-title>Revisiting the Trustworthiness of Saliency Methods in Radiology AI</article-title>
          <source>Radiol Artif Intell</source>
          <year>2024</year>
          <volume>6</volume>
          <issue>1</issue>
          <elocation-id>e220221</elocation-id>
          <pub-id pub-id-type="doi">10.1148/ryai.220221</pub-id>
        </element-citation>
      </ref>
      <ref id="B22">
        <label>22</label>
        <mixed-citation>22. Yanagawa M, Sato J. Seeing is Not Always Believing: Discrepancies in Saliency Maps. Radiol Artif Intell. 2024;6(1):e230488. doi: 10.1148/ryai.230488.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Yanagawa</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Sato</surname>
              <given-names>J</given-names>
            </name>
          </person-group>
          <article-title>Seeing is Not Always Believing: Discrepancies in Saliency Maps</article-title>
          <source>Radiol Artif Intell</source>
          <year>2024</year>
          <volume>6</volume>
          <issue>1</issue>
          <elocation-id>e230488</elocation-id>
          <pub-id pub-id-type="doi">10.1148/ryai.230488</pub-id>
        </element-citation>
      </ref>
      <ref id="B23">
        <label>23</label>
        <mixed-citation>23. Band SS, Yarahmadi A, Hsu CC, Biyari M, Sookhak M, Ameri R, et al. Application of Explainable Artificial Intelligence in Medical Health: A Systematic Review of Interpretability Methods. Inform Med Unlocked. 2023;40(3):101286. doi: 10.1016/j.imu.2023.101286.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Band</surname>
              <given-names>SS</given-names>
            </name>
            <name>
              <surname>Yarahmadi</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Hsu</surname>
              <given-names>CC</given-names>
            </name>
            <name>
              <surname>Biyari</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Sookhak</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Ameri</surname>
              <given-names>R</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Application of Explainable Artificial Intelligence in Medical Health: A Systematic Review of Interpretability Methods</article-title>
          <source>Inform Med Unlocked</source>
          <year>2023</year>
          <volume>40</volume>
          <issue>3</issue>
          <size units="pages">101286</size>
          <pub-id pub-id-type="doi">10.1016/j.imu.2023.101286</pub-id>
        </element-citation>
      </ref>
      <ref id="B24">
        <label>24</label>
        <mixed-citation>24. Sun Q, Akman A, Schuller BW. Explainable Artificial Intelligence for Medical Applications: A Review. arXiv:2412.01829. doi: 10.48550/arXiv.2412.01829.</mixed-citation>
        <element-citation publication-type="other">
          <person-group person-group-type="author">
            <name>
              <surname>Sun</surname>
              <given-names>Q</given-names>
            </name>
            <name>
              <surname>Akman</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Schuller</surname>
              <given-names>BW</given-names>
            </name>
          </person-group>
          <source>Explainable Artificial Intelligence for Medical Applications: A Review</source>
          <comment>arXiv:2412.01829</comment>
          <pub-id pub-id-type="doi">10.48550/arXiv.2412.01829</pub-id>
        </element-citation>
      </ref>
    </ref-list>
    <fn-group>
      <fn fn-type="other">
        <label>Vincula&#231;&#227;o Acad&#234;mica:</label>
        <p> N&#227;o h&#225; vincula&#231;&#227;o deste estudo a programas de p&#243;s-gradua&#231;&#227;o.</p>
      </fn>
      <fn fn-type="other">
        <label>Aprova&#231;&#227;o &#201;tica e Consentimento Informado:</label>
        <p> Este artigo n&#227;o cont&#233;m estudos com humanos ou animais realizados por nenhum dos autores.</p>
      </fn>
      <fn fn-type="other">
        <label>Uso de Intelig&#234;ncia Artificial:</label>
        <p> Durante a prepara&#231;&#227;o deste trabalho, o(s) autor(es) usaram Grammarly para aprimorar a l&#237;ngua inglesa, a gram&#225;tica, a ortografia e a clareza geral do manuscrito. Ap&#243;s o uso desta ferramenta/servi&#231;o, o(s) autor(es) revisaram e editaram o conte&#250;do conforme necess&#225;rio e assumem total responsabilidade pelo conte&#250;do do artigo publicado.</p>
      </fn>
      <fn fn-type="data-availability" specific-use="data-available-upon-request">
        <label>Disponibilidade de Dados:</label>
        <p> Todo o conjunto de dados que d&#225; suporte aos resultados deste estudo est&#225; dispon&#237;vel mediante solicita&#231;&#227;o ao autor correspondente.</p>
      </fn>
      <fn fn-type="financial-disclosure">
        <label>Fontes de Financiamento:</label>
        <p> O presente estudo n&#227;o teve fontes de financiamento externas.</p>
      </fn>
    </fn-group>
  </back>
  <sub-article article-type="translation" id="TRen" xml:lang="en">
    <front-stub>
      <article-id pub-id-type="doi">10.36660/abc.20250790i</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Review Article</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Artificial Intelligence Applied to Electrocardiogram-Based Cardiovascular Risk Assessment: A Systematic Review</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0009-0006-8808-1934</contrib-id>
          <name>
            <surname>Pinheiro</surname>
            <given-names>Maria Clara Mantoan</given-names>
          </name>
          <role>Conception and design of the research and Writing of the manuscript</role>
          <role>Acquisition of data</role>
          <role>Analysis and interpretation of the data and Critical revision of the manuscript for intellectual content</role>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-0269-569X</contrib-id>
          <name>
            <surname>Felberg</surname>
            <given-names>L&#237;via</given-names>
          </name>
          <role>Conception and design of the research and Writing of the manuscript</role>
          <role>Acquisition of data</role>
          <role>Analysis and interpretation of the data and Critical revision of the manuscript for intellectual content</role>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-3800-7107</contrib-id>
          <name>
            <surname>Bozzi</surname>
            <given-names>Isadora Cristine Reis Sguizzato</given-names>
          </name>
          <role>Acquisition of data</role>
          <role>Analysis and interpretation of the data and Critical revision of the manuscript for intellectual content</role>
          <role>Statistical analysis</role>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-2740-0042</contrib-id>
          <name>
            <surname>Ribeiro</surname>
            <given-names>Ant&#244;nio Luiz Pinho</given-names>
          </name>
          <role>Analysis and interpretation of the data and Critical revision of the manuscript for intellectual content</role>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0003-1349-1745</contrib-id>
          <name>
            <surname>Paix&#227;o</surname>
            <given-names>Gabriela Miana de Mattos</given-names>
          </name>
          <role>Conception and design of the research and Writing of the manuscript</role>
          <role>Acquisition of data</role>
          <role>Analysis and interpretation of the data and Critical revision of the manuscript for intellectual content</role>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff1001">
        <label>1</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">Universidade Federal de Minas Gerais, Belo Horizonte, MG &#8211; Brazil</institution>
      </aff>
      <author-notes>
        <corresp id="c01001"><label>Mailing Address:</label> Gabriela Miana de Mattos Paix&#227;o Universidade Federal de Minas Gerais - Rua Alfredo Balena, 190. Postal Code 30130-100, Santa Efig&#234;nia, Belo Horizonte, MG &#8211; Brazil E-mail: gabimiana@gmail.com </corresp>
        <fn fn-type="coi-statement">
          <label>Potential Conflict of Interest:</label>
          <p>No potential conflict of interest relevant to this article was reported.</p>
        </fn>
        <fn fn-type="edited-by">
          <label>Editor responsible for the review:</label>
          <p> Marcio Bittencourt</p>
        </fn>
      </author-notes>
      <abstract>
        <title>Abstract</title>
        <p>Cardiovascular diseases (CVDs) are the leading cause of mortality worldwide, underscoring the need for effective risk prediction and early detection. Although the electrocardiogram (ECG) is a widely available and low-cost diagnostic tool, its traditional interpretation is limited by subjectivity. Artificial intelligence (AI) has emerged as a promising approach, capable of extracting hidden prognostic information from ECG signals. This systematic review aimed to assess original studies applying AI techniques to ECGs for cardiovascular risk prediction and mortality. Original studies that used ECG signals as the sole input variable for AI models, focusing on cardiovascular risk outcomes, were included. A systematic search was conducted in different databases, and data were synthesized narratively. Eleven studies were included, predominantly retrospective cohorts applying convolutional neural networks (CNNs) to predict cardiovascular risk or mortality. The sample primarily consisted of adult populations in high-income countries. Primary outcomes included all-cause mortality, cardiovascular death, and major adverse cardiovascular events (MACE). Reported AUROC values ranged from 0.63 to 0.961 in training sets, with some models outperforming traditional risk scores. AI-ECG models demonstrated the potential to detect subclinical disease, enabling early risk stratification even in normal ECGs. However, challenges remain regarding population diversity, model interpretability, and prospective validation. The application of AI to ECG analysis represents a promising advancement in personalized cardiovascular risk assessment. Nonetheless, further research is needed to ensure the safety, effectiveness, and equitable clinical integration of these technologies.</p>
      </abstract>
      <abstract abstract-type="graphical">
        <p>
          <fig id="f01001">
            <label>Central Illustration</label>
            <caption>
              <title>: Artificial Intelligence Applied to Electrocardiogram-Based Cardiovascular Risk Assessment: A Systematic Review</title>
            </caption>
            <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf01-en.tif"/></fig>
        </p>
      </abstract>
      <kwd-group xml:lang="en">
        <title>Keywords:</title>
        <kwd>Artificial Intelligence</kwd>
        <kwd>Heart Disease Risk Factors</kwd>
        <kwd>Electrocardiography</kwd>
      </kwd-group>
    </front-stub>
    <body>
      <sec sec-type="intro">
        <title>Introduction</title>
        <p>Cardiovascular diseases (CVDs) are the leading cause of mortality worldwide, encompassing both communicable and noncommunicable conditions.<sup><xref ref-type="bibr" rid="B1">1</xref></sup> Therefore, cardiovascular risk stratification and the early detection of these diseases are essential. In this context, the electrocardiogram (ECG) stands out as a well-established tool due to its high availability, low cost, and operational simplicity. However, ECG interpretation has an important limitation: its subjectivity, as it relies on visual criteria that depend on the examiner.</p>
        <p>In recent years, artificial intelligence (AI) has gained significant prominence, particularly in the field of medicine, opening new frontiers in data analysis. Machine learning (ML) and deep learning (DL) approaches have demonstrated the ability to recognize complex and subtle patterns that may go unnoticed by human observers. This has led to the belief that integrating AI with ECG analysis may enable the extraction of prognostic information and cardiovascular risk stratification insights that were previously unidentified.<sup><xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref></sup></p>
        <p>Current models for estimating cardiovascular risk have notable limitations, as they do not always accurately reflect an individual patient&#8217;s reality and often do not incorporate ECG data. The use of additional methods, such as coronary calcium scoring, has shown good performance for cardiovascular risk re-stratification; however, cost remains a significant barrier, particularly in low- and middle-income countries.<sup><xref ref-type="bibr" rid="B4">4</xref></sup> As a result, a considerable number of patients are underserved, especially those from diverse populations whose social and racial characteristics are not adequately considered. This underscores the need for validation in heterogeneous populations &#8212; a practice still lacking in some existing risk stratification tools.<sup><xref ref-type="bibr" rid="B4">4</xref></sup></p>
        <p>Although most atherosclerotic events can be prevented through health-promotion strategies and the control of known risk factors, individual variables persist, including low treatment adherence and social inequality.<sup><xref ref-type="bibr" rid="B2">2</xref></sup> In this context, applying AI to ECG analysis may offer an alternative to support the development of more sensitive, personalized, and integrated prediction models.<sup><xref ref-type="bibr" rid="B3">3</xref></sup></p>
        <p>The purpose of this article is to provide an up-to-date and critical systematic review of the primary applications of AI in ECG interpretation, with a focus on predicting major adverse cardiovascular events (MACE). It examines recent advances, the challenges associated with large-scale implementation, and the prospects for adoption in clinical practice. The main findings of this systematic review are summarized in the <xref ref-type="fig" rid="f01001">Central Illustration</xref>.</p>
      </sec>
      <sec sec-type="methods">
        <title>Methods</title>
        <p>This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.<sup><xref ref-type="bibr" rid="B5">5</xref></sup> The protocol was registered in PROSPERO (1039916).</p>
        <p>The PICOTS framework (Population, Intervention, Comparator, Outcome, Timing, and Study design) guided the methodology of this systematic review. Studies involving adult populations that applied AI models using only ECG signals as input, without requiring an explicit comparator group, were included. The primary outcomes were all-cause mortality, cardiovascular death, heart failure (HF), myocardial infarction (MI), stroke, and atherosclerotic cardiovascular disease (ASCVD). Follow-up periods varied across studies, with a predominance of retrospective cohort designs.</p>
        <sec>
          <title>Eligibility criteria</title>
          <p>We included original research articles published in English within the past five years that directly evaluated the use of AI &#8211; specifically convolutional neural networks (CNNs) or deep neural networks (DNNs) &#8211; applied exclusively to ECGs for predicting cardiovascular risk in adult populations (aged &#8805;18 years). Eligible studies assessed the risk or prediction of all-cause mortality, ASCVD, and major adverse cardiovascular events (MACE), including MI, HF, stroke, and cardiovascular death.</p>
          <p>We excluded studies that: (1) integrated ECG data with other clinical, laboratory, or imaging information; (2) focused exclusively on arrhythmia diagnosis; (3) were limited to specific patient populations; or (4) were narrative reviews, case reports, or did not provide access to the full text.</p>
        </sec>
        <sec>
          <title>Information sources and search strategy</title>
          <p>The literature search was conducted in PubMed, Cochrane Library, Scopus, and ScienceDirect. Searches were limited to articles published between 2020 and 2025, and the final search was performed in May 2025. The following English-language terms and their synonyms were used as search descriptors: &#8220;Artificial Intelligence,&#8221; &#8220;Machine Learning,&#8221; &#8220;Deep Learning,&#8221; &#8220;Electrocardiography,&#8221; &#8220;Electrocardiogram,&#8221; &#8220;ECG,&#8221; &#8220;Cardiovascular Risk Factors,&#8221; &#8220;Risk Assessment,&#8221; &#8220;Cardiovascular Risk,&#8221; and &#8220;Heart Disease Risk.&#8221; Boolean operators &#8220;AND&#8221; and &#8220;OR&#8221; were applied to combine terms.</p>
        </sec>
        <sec>
          <title>Selection process</title>
          <p>The selection process consisted of four stages: identification, screening, eligibility assessment, and inclusion. The initial search identified 1,118 records, of which 1,056 were excluded after title screening because they did not address the core focus of the review. Of the remaining 62 records, 45 were excluded following abstract assessment and removal of duplicates. Subsequently, six additional records were excluded after full-text evaluation for not meeting the inclusion criteria. Ultimately, eleven studies fulfilled all eligibility criteria and were included in the review.</p>
          <p>Two reviewers independently conducted all phases of the selection process. Any discrepancies were resolved through discussion with a third reviewer. The complete selection workflow is presented in the PRISMA flow diagram (<xref ref-type="fig" rid="f02001">Figure 1</xref>).</p>
          <p>
            <fig id="f02001">
              <label>Figure 1</label>
              <caption>
                <title>&#8211; PRISMA Flow Diagram: study selection process.</title>
              </caption>
              <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf02-en.tif"/></fig>
          </p>
        </sec>
        <sec>
          <title>Data collection process and data Items</title>
          <p>Data extraction was performed independently by two reviewers using standardized data extraction forms created in Microsoft Excel. Extracted data included study design, population characteristics, AI model type (CNN or DNN), ECG acquisition details, and the cardiovascular outcomes assessed. Additional variables collected included study setting, sample size, the training and validation procedures of the AI models, and whether external validation was conducted.</p>
        </sec>
      </sec>
      <sec sec-type="results">
        <title>Results</title>
        <p>A total of eleven original studies were included in this systematic review. Their characteristics are summarized in <xref ref-type="table" rid="t1001">Table 1</xref>. Ten studies were retrospective cohorts<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B15">15</xref></sup> and one study had a prospective design.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> The studies analyzed a wide range of ECG signals, predominantly 12-lead ECGs, with some also evaluating single-lead (Lead I) recordings.<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref></sup></p>
        <p>
          <table-wrap id="t1001">
            <label>Table 1</label>
            <caption>
              <title>&#8211; Characteristics of Included AI-ECG Studies</title>
            </caption>
            <table frame="hsides" rules="groups">
              <colgroup>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
              </colgroup>
              <thead>
                <tr>
                  <th align="left">Authors</th>
                  <th>Year of publication</th>
                  <th>Article Title</th>
                  <th>Study Type</th>
                  <th>AI Techniques</th>
                  <th>Signal Input Type</th>
                  <th>Training Population</th>
                  <th>Number of ECGs</th>
                  <th>Mean Age (years)</th>
                  <th>Male (%)</th>
                  <th>Follow-up (Mean, years)</th>
                  <th>External Validation</th>
                  <th>Validation Population</th>
                </tr>
              </thead>
              <tbody>
                <tr>
                  <td>Sau et al.<sup>6</sup></td>
                  <td align="center">2024</td>
                  <td align="center">Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead and single-lead (Lead I) ECG</td>
                  <td align="center">BIDMC (USA)</td>
                  <td align="center">1,163,401</td>
                  <td align="center">NR</td>
                  <td align="center">NR</td>
                  <td align="center">5.46</td>
                  <td align="center">Yes</td>
                  <td align="center">CODE (Brazil) SaMi-Trop (Brazil) ELSA-Brazil (Brazil) UK Biobank (UK)</td>
                </tr>
                <tr>
                  <td>Hughes et al.<sup>7</sup></td>
                  <td align="center">2023</td>
                  <td align="center">A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead and single-lead (Lead I) ECG</td>
                  <td align="center">Stanford Health Care (USA)</td>
                  <td align="center">500,000</td>
                  <td align="center">59</td>
                  <td align="center">54%</td>
                  <td align="center">5</td>
                  <td align="center">Yes</td>
                  <td align="center">Cedars-Sinai Medical Center (USA) Columbia University Irving Medical Center (USA)</td>
                </tr>
                <tr>
                  <td>Al-Alusi et al.<sup>8</sup></td>
                  <td align="center">2025</td>
                  <td align="center">A deep learning digital biomarker to detect hypertension and stratify cardiovascular risk from the electrocardiogram</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">Massachusetts General Hospital (USA)</td>
                  <td align="center">752,415</td>
                  <td align="center">57.3 &#177; 16.8</td>
                  <td align="center">51.6%</td>
                  <td align="center">5.6&#8211;7.8</td>
                  <td align="center">Yes</td>
                  <td align="center">Brigham and Women&#8217;s Hospital (USA)</td>
                </tr>
                <tr>
                  <td>Sau et al.<sup>9</sup></td>
                  <td align="center">2025</td>
                  <td align="center">Artificial intelligence-enhanced electrocardiography for the identification of a sex-related cardiovascular risk continuum: a retrospective cohort study</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">BIDMC (USA)</td>
                  <td align="center">1,163,401</td>
                  <td align="center">57.7 &#177; 18.69</td>
                  <td align="center">47.9%</td>
                  <td align="center">3.4&#8211;4.8</td>
                  <td align="center">Yes</td>
                  <td align="center">UK Biobank (UK)</td>
                </tr>
                <tr>
                  <td>Gao et al.<sup>16</sup></td>
                  <td align="center">2025</td>
                  <td align="center">Electrocardiograph analysis for risk assessment of heart failure with preserved ejection fraction: A deep learning model</td>
                  <td align="center">Prospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">First Hospital of Hebei Medical University (China)</td>
                  <td align="center">238</td>
                  <td align="center">NR</td>
                  <td align="center">50.8%</td>
                  <td align="center">2</td>
                  <td align="center">No</td>
                  <td align="center">Prospective internal validation</td>
                </tr>
                <tr>
                  <td>Raghunath et al.<sup>10</sup></td>
                  <td align="center">2020</td>
                  <td align="center">Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">Geisinger (USA)</td>
                  <td align="center">1,169,662</td>
                  <td align="center">58 &#177; 18</td>
                  <td align="center">47%</td>
                  <td align="center">34</td>
                  <td align="center">No</td>
                  <td align="center">Internal fivefold cross-validation</td>
                </tr>
                <tr>
                  <td>Dhingra et al.<sup>11</sup></td>
                  <td align="center">2025</td>
                  <td align="center">Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">YNHHS (USA)</td>
                  <td align="center">286,880</td>
                  <td align="center">57</td>
                  <td align="center">43.4%</td>
                  <td align="center">10</td>
                  <td align="center">Yes</td>
                  <td align="center">UK Biobank (UK) ELSA-Brazil (Brazil)</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>12</sup></td>
                  <td align="center">2025</td>
                  <td align="center">A multitask deep learning model utilizing electrocardiograms for major cardiovascular adverse events prediction</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">CGMH (Taiwan)</td>
                  <td align="center">2,821,889</td>
                  <td align="center">60 &#177; 14.5</td>
                  <td align="center">50.2%</td>
                  <td align="center">12</td>
                  <td align="center">Yes</td>
                  <td align="center">TSGH (Taiwan)</td>
                </tr>
                <tr>
                  <td>Butler et al.<sup>13</sup></td>
                  <td align="center">2023</td>
                  <td align="center">A generalizable electrocardiogram-based artificial intelligence model for 10-year heart failure risk prediction</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">ARIC (USA)</td>
                  <td align="center">14,613</td>
                  <td align="center">54.2</td>
                  <td align="center">44.7%</td>
                  <td align="center">10</td>
                  <td align="center">Yes</td>
                  <td align="center">MESA (USA)</td>
                </tr>
                <tr>
                  <td>Sun et al.<sup>14</sup></td>
                  <td align="center">2023</td>
                  <td align="center">Towards artificial intelligence-based learning health system for population-level mortality prediction using electrocardiograms</td>
                  <td align="center">Retrospective</td>
                  <td align="center">DL</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">Alberta (Canada)</td>
                  <td align="center">1,605,268</td>
                  <td align="center">65.80 &#177; 17.25</td>
                  <td align="center">56.81%</td>
                  <td align="center">13</td>
                  <td align="center">No</td>
                  <td align="center">NR</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>15</sup></td>
                  <td align="center">2025</td>
                  <td align="center">ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram</td>
                  <td align="center">Retrospective</td>
                  <td align="center">CNN</td>
                  <td align="center">12-lead ECG</td>
                  <td align="center">CGMH (Taiwan)</td>
                  <td align="center">4,932,573</td>
                  <td align="center">60.5 &#177; 16.8</td>
                  <td align="center">52.5%</td>
                  <td align="center">12</td>
                  <td align="center">Yes</td>
                  <td align="center">Tri-Service General Hospital (Taiwan)</td>
                </tr>
              </tbody>
            </table>
            <table-wrap-foot>
              <fn id="TFN1001">
                <p>AI: Artificial Intelligence; ARIC: Atherosclerosis Risk in Communities Study; BIDMC: Beth Israel Deaconess Medical Center; BWH: Brigham and Women&#8217;s Hospital; CGMH: Chang Gung Memorial Hospital; CODE: Clinical Outcomes in Digital Electrocardiology; DL: Deep Learning; DNN: Deep Neural Network; ECG: Electrocardiogram; ELSA-Brasil: Brazilian Longitudinal Study of Adult Health; HECIUH: Health Examination Center at Inha University Hospital; HHMU: Hospital of Hebei Medical University; Lead I: Lead I (single-lead electrocardiogram); MESA: Multi-Ethnic Study of Atherosclerosis; MGH: Massachusetts General Hospital; NR: Not reported; SaMi-Trop: Minas Gerais State Health Department &#8211; Trypanosoma cruzi (Brazilian cohort of Chagas disease patients); TSGH: Tri-Service General Hospital; UK Biobank: United Kingdom Biobank; YNHHS: Yale New Haven Health System.</p>
              </fn>
            </table-wrap-foot>
          </table-wrap>
        </p>
        <p>Across the included studies, the majority (90.9%) employed CNNs as the primary architecture for ECG signal analysis.<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref></sup> Specifically, CNNs were used either alone or in combination with other methods, such as Long Short-Term Memory (LSTM) layers, Residual Networks (ResNet), or Variational Autoencoders (VAEs). One study utilized broader DL approaches without explicitly using CNN architectures, falling under the category of DNNs or general DL.<sup><xref ref-type="bibr" rid="B14">14</xref></sup></p>
        <p>Sample sizes varied substantially, ranging from 238 patients to more than 4 million ECGs. Study populations primarily consisted of adult patients from hospital systems associated with large biobanks, with mean ages typically ranging from 54 to 65 years. However, some studies did not provide detailed demographic information.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> Regarding follow-up duration, the longest reported period was 34 years,<sup><xref ref-type="bibr" rid="B10">10</xref></sup> whereas the shortest was two years.<sup><xref ref-type="bibr" rid="B16">16</xref></sup></p>
        <p>As shown in <xref ref-type="fig" rid="f03001">Figure 2</xref>, most cohorts used to train the AI models were based in the United States of America (USA).<sup><xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B12">12</xref></sup> Two studies used populations from Taiwan,<sup><xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B17">17</xref></sup> while one study each included populations from China<sup><xref ref-type="bibr" rid="B10">10</xref></sup> and Canada.<sup><xref ref-type="bibr" rid="B14">14</xref></sup></p>
        <p>
          <fig id="f03001">
            <label>Figure 2</label>
            <caption>
              <title>&#8211; Geographic distribution of AI-ECG study cohorts. The map highlights the locations of datasets used for training (blue) and external validation (red) of AI-ECG models. Most training cohorts originated from the USA, while validation datasets were concentrated in the UK, Brazil, Taiwan, and the USA.</title>
            </caption>
            <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf03-en.tif"/></fig>
        </p>
        <p>Studies that employed external validation also used participants from high-income countries, such as the United Kingdom (UK)<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup> and the USA.<sup><xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B13">13</xref></sup> Only two studies used a cohort from Brazil for external validation,<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B12">12</xref></sup> and two studies included populations from Taiwan.<sup><xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref></sup></p>
        <p>All outcomes relevant to this review are presented in <xref ref-type="table" rid="t2001">Table 2</xref>. Additionally, Figures 3 and 4 display forest plots summarizing the predictive performance of AI-ECG models across the included studies. <xref ref-type="fig" rid="f04001">Figure 3</xref> presents discrimination metrics, including the concordance index (C-index) and the area under the receiver operating characteristic curve (AUC/AUROC), while <xref ref-type="fig" rid="f05001">Figure 4</xref> shows effect measures expressed as hazard ratios (HR) for various cardiovascular outcomes, including 95% confidence intervals when available.</p>
        <p>
          <table-wrap id="t2001">
            <label>Table 2</label>
            <caption>
              <title>&#8211; Summary of AI-ECG Performance Across Clinical Outcomes</title>
            </caption>
            <table frame="hsides" rules="groups">
              <colgroup>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
                <col/>
              </colgroup>
              <thead>
                <tr>
                  <th align="left">Authors</th>
                  <th>All-cause mortality</th>
                  <th>Cardiovascular death</th>
                  <th>ASCVD</th>
                  <th>Heart Failure</th>
                  <th>HFpEF</th>
                  <th>AMI</th>
                  <th>Stroke</th>
                  <th>MACE</th>
                </tr>
              </thead>
              <tbody>
                <tr>
                  <td>Sau et al.<sup>6</sup></td>
                  <td align="center">C-index 0.775 (0.773&#8211;0.776) Event rate: NR</td>
                  <td align="center">C-index 0.832 (0.831&#8211;0.834) Event rate: NR</td>
                  <td align="center">C-index 0.696 (0.694&#8211;0.698) SEER model C-index 0.547 (0.527&#8211;0.567) Event rate: NR</td>
                  <td align="center">C-index 0.787 (0.785&#8211;0.789) Event rate: NR</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Hughes et al.<sup>7</sup></td>
                  <td align="center">-</td>
                  <td align="center">Stanford (12-lead) AUC 0.832 (0.810&#8211;0.854), C-index 0.815 (0.798&#8211;0.826); Cedars-Sinai (12-lead) AUC 0,781 (0.767&#8211;0.795), C-index 0.777 (0.768&#8211;0.783); Columbia (12-lead) AUC 0,825 (0.821&#8211;0.829), C-index 0.808 (0.806&#8211;0.810) Stanford (1-lead) AUC 0.797 (0.773&#8211;0.824), C-index 0.781 (0.760&#8211;0.793); Cedars-Sinai (1-lead) AUC 0,778 (0.765&#8211;0.792), C-index 0.771 (0.764&#8211;0.778); Columbia (1-lead) AUC 0,761 (0.757&#8211;0.765), C-index 0.755 (0.753&#8211;0.757) Event rate: NR</td>
                  <td align="center">Stanford (12-lead) AUC 0.668 (0.647&#8211;0.687), C-index 0.661 (0.652&#8211;0.669); Cedars-Sinai (12-lead) AUC 0,632 (0.591&#8211;0.674), C-index 0.635 (0.619&#8211;0.652) Event rate: NR</td>
                  <td align="center">Stanford HR 3.2 (2.9&#8211;3.5) Event rate: NR</td>
                  <td align="center">-</td>
                  <td align="center">Stanford HR 2.0 (1.7&#8211;2.5) Event rate: NR</td>
                  <td align="center">Stanford HR 1.6 (1.4&#8211;1.9) Event rate: NR</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Al-Alusi et al.<sup>8</sup></td>
                  <td align="center">HR 1.47 (1.36&#8211;1.60), C-index 0.804 (0.796&#8211;0.813) Event rate: 32,7% deaths in the high risk group and 13% in the low risk group</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">HR 2.26 (1.90&#8211;2.69), C-index 0.804 (0.796&#8211;0.813) Event rate: 5.9%</td>
                  <td align="center">-</td>
                  <td align="center">HR 1.87 (1.69&#8211;2.07), C-index 0.733 (0.725&#8211;0.742) Event rate: 14.1%</td>
                  <td align="center">HR 1.30 (1.18&#8211;1.44) Event rate: 6.5%</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Sau et al.<sup>9</sup></td>
                  <td align="center">Male HR 1.22 (1.19&#8211;1.25), Female HR 1.17 (1.19&#8211;1.31) Event rate: 18.4%</td>
                  <td align="center">Male HR 1.00 (0.63&#8211;1.58), Female HR 1.78 (1.18&#8211;2.70) Event rate: 3.7%</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Gao et al.<sup>16</sup></td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">ACC: 75% (Cohort A), 71.8% (Cohort B) Event rate: NR</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Raghunath et al.<sup>10</sup></td>
                  <td align="center">ECG AUC 0.855 Events: 14,207</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Dhingra et al.<sup>11</sup></td>
                  <td align="center">YNHHS HR 1.19 (1.15&#8211;1.24) UKB HR 2.13 (1.41&#8211;3.24) ELSA HR 3.64 (2.27&#8211;5.83) Event rate: YNHHS 7.5%, UKB 0.8%, ELSA 1.7%</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">YNHHS HR 6.51 (6.11&#8211;6.93), C-index 0.718 (0.697&#8211;0.738); UKB HR 18.33 (9.90&#8211;33.97), C-index 0.769 (0.670&#8211;0.867); ELSA HR 32.06 (15.36&#8211;66.92), C-index 0.810 (0.714&#8211;0.907) Event rate: YNHHS 4.2%</td>
                  <td align="center">-</td>
                  <td align="center">YNHHS HR 1.44 (1.04&#8211;2.00) UKB HR 3.16 (1.98&#8211;5.02) ELSA HR 3.53 (1.4&#8211;8.85) Event rate: YNHHS 0.1%, UKB 0.5%, ELSA 0.4%</td>
                  <td align="center">UKB HR 2.30 (1.36&#8211;3.9) ELSA HR 5.74 (2.59&#8211;12.72) Event rate: YNHHS 1.6%, UKB 0.5%, ELSA 0.4%</td>
                  <td align="center">YNHHS HR 2.10 (2.04&#8211;2.17) UKB HR 2.79 (2.17&#8211;3.6) ELSA HR 4.04 (2.77&#8211;5.89) Event rate: YNHHS 10.4%, UKB 1.8%, ELSA 2.5%</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>12</sup></td>
                  <td align="center">AUROC 0.89, Sensitivity 0.81, Specificity 0.83, AUROC (External) 0.83 Event rate: 6.3% (3 mo), 8.5% (6 mo), 10.0% (9 mo), 11.4% (1 year)</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">AUROC 0.90 Sensitivity 0.83 Specificity 0.81 Event rate: 0.6% (3 mo), 0.8% (6 mo), 1.1% (9 mo), 1.2% (1 year)</td>
                  <td align="center">-</td>
                  <td align="center">AUROC 0.85 Sensitivity 0.76 Specificity 0.79 Event rate: 0.3% (3 mo), 0.4% (6 mo), 0.5% (9 mo), 0.5% (1 year)</td>
                  <td align="center">AUROC 0.76 Sensitivity 0.75 Specificity 0.63 Event rate: 0.3% (3 mo), 0.4% (6 mo), 0.6% (9 mo), 0.7% (1 year)</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Butler et al.<sup>13</sup></td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">ARIC AUC 0.76 (0.72&#8211;0.83) MESA AUC 0.77 (0.74&#8211;0.79) Event rate: 3.6% (MESA), 5.5% (ARIC)</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Sun et al.<sup>14</sup></td>
                  <td align="center">30 days: AUROC 0.843 (0.832&#8211;0.858), Brier Score 5.97 (5.83&#8211;6.12); 1 year: AUROC 0.812 (0.807&#8211;0.81.6), Brier Score 11.44 (11.3-11.6); 5 years: AUROC 0.798 (0.791&#8211;0.802), Brier Score 17.52 (17.29&#8211;17.83) Event rate: 7.6% (30 days) 17.3%, (1 year) 32.9%, (5 years)</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
                <tr>
                  <td>Lin et al.<sup>15</sup></td>
                  <td align="center">CGMH: C-index 0.860 (0.859&#8211;0.861) TSGH: C-index 0.813 (0.807&#8211;0.814) Event rate: CGMH 5.9% (3 mo), 7.8% (6 mo), 9.3% (9 mo), 10.5% (1 year); TSGH 6.8% (3 mo), 8.5% (6 mo), 9.5% (9 mo), 10.5% (1 year)</td>
                  <td align="center">CGMH: C-index 0.891 (0.890&#8211;0.893) Event rate: NR</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                  <td align="center">-</td>
                </tr>
              </tbody>
            </table>
            <table-wrap-foot>
              <fn id="TFN2001">
                <p>Values are presented as point estimates with 95% confidence intervals in parentheses, unless otherwise specified. CC: accuracy; ARIC: Atherosclerosis Risk in Communities Study; ASCVD: atherosclerotic cardiovascular disease; AUC/AUROC: area under the receiver operating characteristic curve; CGMH: Chang Gung Memorial Hospital; CI: confidence interval; ECG: electrocardiogram; ELSA-Brasil: Brazilian Longitudinal Study of Adult Health; HFpEF: heart failure with preserved ejection fraction; HR: hazard ratio; MACE: major adverse cardiovascular events; MESA: Multi-Ethnic Study of Atherosclerosis; mo: months; NR: not reported; SEER: Surveillance, Epidemiology, and End Results (US database); TSGH: Tri-Service General Hospital; UKB: United Kingdom Biobank; YNHHS: Yale New Haven Health System.</p>
              </fn>
            </table-wrap-foot>
          </table-wrap>
        </p>
        <p>
          <fig id="f04001">
            <label>Figure 3</label>
            <caption>
              <title>&#8211; Discrimination metrics (AUC and C-index) for AI-ECG models. Forest plot summarizing the predictive performance of ECG-based AI models across different studies and time horizons. Blue markers represent the C-index (Harrell&#8217;s concordance index), and orange markers represent the Area Under the Receiver Operating Characteristic curve (AUC/AUROC). Data are presented as point estimates with 95% Confidence Intervals (95% CI). The dashed vertical line at 0.5 indicates the threshold for non-discriminatory (random) prediction.</title>
            </caption>
            <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf04-en.tif"/></fig>
        </p>
        <p>
          <fig id="f05001">
            <label>Figure 4</label>
            <caption>
              <title>&#8211; Hazard Ratios for clinical outcomes associated with AI-ECG models. Forest plot showing the association between AI-derived risk categories and clinical events (Mortality, Heart Failure, MACE, etc.). Points represent Hazard Ratios (HR), and horizontal bars indicate the 95% Confidence Intervals (95% CI). The dashed vertical line at 1.0 represents the null hypothesis (no difference in risk).</title>
            </caption>
            <graphic xlink:href="1678-4170-abc-123-07-e20250790-50790-gf05-en.tif"/></fig>
        </p>
        <sec>
          <title>All-cause mortality</title>
          <p>Eight studies reported promising results in predicting all-cause mortality using various AI-ECG models. Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> demonstrated high accuracy, with an AUROC of 0.89 for 1-year mortality and 0.83 for 1-year all-cause mortality in external validation. Sun et al.<sup><xref ref-type="bibr" rid="B14">14</xref></sup> found AUROCs of 0.843 for 30-day mortality, 0.812 for one-year mortality, and 0.798 for 5-year mortality, indicating strong performance across multiple timeframes. Raghunath et al.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> reported an AUC of 0.855 using ECG-only data and 0.876 when age and sex were incorporated. In the Surv-ECG study, Lin et al.<sup><xref ref-type="bibr" rid="B15">15</xref></sup> achieved a C-index of 0.860 for all-cause mortality. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> reported a C-index of 0.775, further supporting the predictive potential of AI-based models for mortality risk.</p>
          <p>Sau et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> identified HR of 1.22 for males and 1.17 for females, suggesting that sex discordance may serve as a novel marker of unrecognized cardiovascular risk. Supporting these findings, Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> reported an HR of 1.19 for all-cause mortality using an AI model applied to ECG images in a multinational study that also demonstrated predictive value for other cardiovascular outcomes, including MI and HF. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> showed that individuals classified as high-risk by the HTN-AI model had a significantly higher 10-year mortality rate (21.0%) compared with the low-risk group (5.4%), underscoring the value of AI in identifying high-risk individuals.</p>
          <p>There was marked heterogeneity, with populations ranging from 286,880 to almost five million ECGs and a mean follow-up from 3 to 34 years. While several reports claimed external validation, in most cases, these were replications within the same healthcare system or biobank. Only a minority of studies tested their models in truly independent populations with distinct demographic and clinical characteristics.<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup></p>
        </sec>
        <sec>
          <title>Cardiovascular death</title>
          <p>The ability of AI-ECG models to predict cardiovascular mortality was also validated. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> reported a C-index of 0.832 for cardiovascular death prediction. In Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> the model showed an AUC of 0.83 and a C-index of 0.82 for 5-year cardiovascular death using the full 12-lead ECG, with an AUC of 0.80 when using Lead I alone. Sau et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> reported a HR of 1.78 for females compared with 1.00 for males among individuals with a normal ECG, reinforcing the role of sex discordance as a potential independent predictor of cardiovascular mortality, particularly in women. Lin et al.,<sup><xref ref-type="bibr" rid="B15">15</xref></sup> in the Surv-ECG study, further supported this evidence by reporting a C-index of 0.891 for cardiovascular death prediction.</p>
          <p>Studies reporting cardiovascular death also exhibited significant heterogeneity. Populations differed in size and follow-up duration. Event rates were not specified in most studies, and many validations labeled as external were conducted in cohorts derived from the same or closely related healthcare systems, raising concerns about the independence of these validations.</p>
        </sec>
        <sec>
          <title>Atherosclerotic cardiovascular disease</title>
          <p>AI-ECG models proved effective in predicting ASCVD. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> reported a C-index of 0.696 using 12-lead and Lead I ECGs, integrated with clinical and genetic data from the UK Biobank. In Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> the model achieved an AUC of 0.67 for 12-lead ECG and 0.63 with Lead I. However, prediction of ASCVD events demonstrated wide variation across studies in terms of sample sizes and event definitions (e.g., composite endpoints including MI, stroke, and sudden cardiac death vs. narrower definitions). Regarding validation, most validations occurred within related institutions or datasets, and only Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> performed truly independent external validation.</p>
        </sec>
        <sec>
          <title>Myocardial infarction</title>
          <p>AI-ECG models have demonstrated promising performance in predicting the risk of acute myocardial infarction (AMI). Hughes et al.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> reported an AUROC of 0.85 for AMI prediction using the SEER score, along with a HR of 2.0. Similarly, Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> reported an HR of 1.87. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> also found an elevated risk, with an HR of 1.44 for incident AMI. Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> likewise demonstrated high accuracy, reporting an AUROC of 0.85.</p>
          <p>The studies were heterogeneous in design, with differences in event definitions, event rates, and sample sizes ranging from thousands to millions. Regarding validation, most studies conducted validation within the same or closely related biobanks or institutions, and only Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> evaluated their models in geographically distinct cohorts.</p>
        </sec>
        <sec>
          <title>Heart failure</title>
          <p>AI-ECG models have demonstrated strong performance in predicting heart failure (HF), although studies varied substantially in methodology. The highest performance was observed in Lin et al.,<sup><xref ref-type="bibr" rid="B12">12</xref></sup> who reported an AUROC of 0.90 for HF prediction. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> found HRs of 6.51 in the YNHHS cohort, 18.33 in the UK Biobank, and 32.06 in the ELSA-Brazil cohort for incident HF, demonstrating substantial prognostic strength across diverse populations. In Sau et al.,<sup><xref ref-type="bibr" rid="B6">6</xref></sup> a C-index of 0.787 was achieved using 12-lead and Lead I ECGs, highlighting the model&#8217;s reliability.</p>
          <p>The AI-ECG model developed by Butler et al.<sup><xref ref-type="bibr" rid="B13">13</xref></sup> achieved an AUC of 0.76 in the ARIC cohort and 0.77 in the MESA cohort using ECG data alone, with notable improvement when clinical variables were incorporated. When combined with 12 risk factors, the AI-ECG-Cox model achieved an AUC of 0.82 in ARIC and 0.84 in MESA, demonstrating enhanced predictive value through integration of clinical data. In Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> the model showed 76% sensitivity and 82% specificity for HF prediction, further supporting its diagnostic capabilities. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> reported an HR of 2.26. Finally, Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> reported 75% accuracy in Cohort A and 71.8% accuracy in Cohort B for heart failure with preserved ejection fraction, with balanced sensitivity (71.7%) and specificity (71.9%).</p>
          <p>Most validations were conducted in related datasets; only a small number of studies<sup><xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup> validated their models in demographically distinct populations.</p>
        </sec>
        <sec>
          <title>Stroke</title>
          <p>Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> demonstrated the potential of AI-ECG models in predicting stroke risk, achieving an AUROC of 0.76. In Hughes et al.,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> individuals in the Stanford cohort with the highest SEER scores exhibited a 1.6-fold increased HR for incident stroke, adjusted for age and sex, highlighting the model&#8217;s ability to identify individuals at elevated cerebrovascular risk early. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> reported an HR of 2.30 in the UK Biobank, and Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> similarly reported an HR of 2.30.</p>
          <p>Studies varied in whether they included only ischemic stroke or all stroke subtypes, which contributed to differences in event rates. Validation was often limited to internal replications or datasets from related institutions, with only Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> conducting truly independent external validation.</p>
        </sec>
        <sec>
          <title>Correlation with traditional scores</title>
          <p>As presented in <xref ref-type="table" rid="t3001">Table 3</xref>, Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> demonstrated superior discrimination compared with the Framingham Risk Score (FRS) across multiple outcomes. Dhingra et al.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> showed that their AI-ECG model outperformed the Pooled Cohort Equation for heart failure (PCE-HF). In the YNHHS cohort, the AI-ECG model achieved a C-index of 0.718, compared with 0.601 for the PCE-HF. Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> reported that their AI-ECG model outperformed the PCE-HF in predicting ASCVD, with a C-index of 0.696 versus 0.547 for the SEER model.</p>
          <p>
            <table-wrap id="t3001">
              <label>Table 3</label>
              <caption>
                <title>&#8211; Summary of Associations With Traditional Clinical Scores</title>
              </caption>
              <table frame="hsides" rules="groups">
                <colgroup>
                  <col/>
                  <col/>
                  <col/>
                </colgroup>
                <thead>
                  <tr>
                    <th align="left">Authors</th>
                    <th>Study</th>
                    <th>Correlation with Traditional Scores</th>
                  </tr>
                </thead>
                <tbody>
                  <tr>
                    <td>Sau et al.<sup>6</sup></td>
                    <td align="center">Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study</td>
                    <td align="center">AI-ECG (C-index 0.696) outperformed SEER (C-index 0.547) and showed improved discrimination versus traditional risk-based models</td>
                  </tr>
                  <tr>
                    <td>Hughes et al.<sup>7</sup></td>
                    <td align="center">A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease</td>
                    <td align="center">Correlation with PCE (Pearson r = 0.218); NRI: 17.8%; reclassified 16% of patients from low to moderate ASCVD risk (PCE-defined). Improved discrimination for cardiovascular mortality compared with PCE, but lower discrimination for ASCVD outcomes</td>
                  </tr>
                  <tr>
                    <td>Al-Alusi et al.<sup>8</sup></td>
                    <td align="center">A deep learning digital biomarker to detect hypertension and stratify cardiovascular risk from the electrocardiogram</td>
                    <td align="center">Comparable discrimination to PCP-HF for HF prediction (C-index 0.804 vs 0.806). Superior discrimination vs PCE for MI (0.733 vs 0.681) and stroke (0.704 vs 0.683)</td>
                  </tr>
                  <tr>
                    <td>Sau et al.<sup>9</sup></td>
                    <td align="center">Artificial intelligence-enhanced electrocardiography for the identification of a sex-related cardiovascular risk continuum: a retrospective cohort study</td>
                    <td align="center">NR</td>
                  </tr>
                  <tr>
                    <td>Gao et al.<sup>16</sup></td>
                    <td align="center">Electrocardiograph analysis for risk assessment of heart failure with preserved ejection fraction: A deep learning model</td>
                    <td align="center">NR</td>
                  </tr>
                  <tr>
                    <td>Raghunath et al.<sup>10</sup></td>
                    <td align="center">Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network</td>
                    <td align="center">AI-ECG model (AUC 0.876) outperformed FRS (AUC 0.648) and CCI (AUC 0.816)</td>
                  </tr>
                  <tr>
                    <td>Dhingra et al.<sup>11</sup></td>
                    <td align="center">Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study</td>
                    <td align="center">YNHHS C-index 0.718 (AI-ECG) vs 0.601 (PCP-HF); improved discrimination and reclassification (NRI 21.9%)</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>12</sup></td>
                    <td align="center">A multitask deep learning model utilizing electrocardiograms for major cardiovascular adverse events prediction</td>
                    <td align="center">AI-ECG model demonstrated superior discrimination compared with FRS across outcomes (1-year AUROC: HF 0.90 vs 0.64; MI 0.85 vs 0.69; stroke 0.76 vs 0.61; all-cause mortality 0.89 vs 0.52)</td>
                  </tr>
                  <tr>
                    <td>Butler et al.<sup>13</sup></td>
                    <td align="center">A generalizable electrocardiogram-based artificial intelligence model for 10-year heart failure risk prediction</td>
                    <td align="center">AI-ECG model was comparable to FHS-HF (AUC 0.77 vs 0.74), with improved performance when combined with clinical variables (AUC 0.84)</td>
                  </tr>
                  <tr>
                    <td>Sun et al.<sup>14</sup></td>
                    <td align="center">Towards artificial intelligence-based learning health system for population-level mortality prediction using electrocardiograms</td>
                    <td align="center">NR</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>15</sup></td>
                    <td align="center">ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram</td>
                    <td align="center">AI-ECG outperformed Framingham-based Cox models for cardiovascular death (C-index 0.891 vs 0.734)</td>
                  </tr>
                </tbody>
              </table>
              <table-wrap-foot>
                <fn id="TFN3001">
                  <p>AI: artificial intelligence; ASCVD: atherosclerotic cardiovascular disease; AUC/AUROC: area under the receiver operating characteristic curve; CCI: Charlson Comorbidity Index; C-index: concordance index; ECG: electrocardiogram; FRS: Framingham Risk Score; FRS-HF: Framingham Risk Score for Heart Failure; HF: heart failure; HR: hazard ratio; NRI: net reclassification improvement; NR: not reported; PCE: pooled cohort equations; PCE-HF: pooled cohort equations to prevent heart failure; SEER: Surveillance, Epidemiology, and End Results (US database).</p>
                </fn>
              </table-wrap-foot>
            </table-wrap>
          </p>
          <p>Raghunath et al.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> demonstrated superior performance in predicting one-year all-cause mortality, surpassing both the FRS (AUC 0.648) and the Charlson Comorbidity Index (CCI) (AUC 0.816), achieving an AUC of 0.876 using ECG-only data. In Butler et al.,<sup><xref ref-type="bibr" rid="B13">13</xref></sup> the FRS for HF achieved an AUC of 0.78 in the ARIC cohort and 0.74 in the MESA cohort. Hughes et al.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> found a significant correlation with the PCE-HF, reporting a Net Reclassification Improvement (NRI) of 17.8%, correctly reclassifying 16% of patients initially categorized as low-risk into a moderate-risk category.</p>
        </sec>
        <sec>
          <title>Quality of the studies</title>
          <p>The quality of the included studies was assessed using the Newcastle&#8211;Ottawa Scale (NOS), as shown in <xref ref-type="table" rid="t4001">Table 4</xref>. The NOS evaluates studies across three domains: selection, comparability, and outcome. Each domain contains specific criteria, with a maximum possible score of 9 points.</p>
          <p>
            <table-wrap id="t4001">
              <label>Table 4</label>
              <caption>
                <title>&#8211; Quality Assessment of the Studies by the Newcastle-Ottawa Scale</title>
              </caption>
              <table frame="hsides" rules="groups">
                <colgroup>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                  <col/>
                </colgroup>
                <thead>
                  <tr>
                    <th align="left" rowspan="2">Study</th>
                    <th colspan="4">Selection</th>
                    <th colspan="2">Comparability</th>
                    <th colspan="3">Outcome</th>
                    <th>Total Quality Score</th>
                  </tr>
                  <tr>
                    <th>Representativeness of Exposed Cohort</th>
                    <th>Selection of Non-Exposed Cohort</th>
                    <th>Ascertainment of Exposure</th>
                    <th>Outcome Not Present at Start</th>
                    <th>Main Factor</th>
                    <th>Additional Factor</th>
                    <th>Assessment of Outcomes</th>
                    <th>Length of Follow-up</th>
                    <th>Adequacy of Follow-up</th>
                    <th align="left" style="font-weight:normal"> </th>
                  </tr>
                </thead>
                <tbody>
                  <tr>
                    <td>Sau et al.<sup>6</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Hughes et al.<sup>7</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Al-Alusi et al.<sup>8</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Sau et al.<sup>9</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td> </td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">8</td>
                  </tr>
                  <tr>
                    <td>Gao et al.<sup>16</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td> </td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">8</td>
                  </tr>
                  <tr>
                    <td>Raghunath et al.<sup>10</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Dhingra et al.<sup>11</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>12</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Butler et al.<sup>13</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                  <tr>
                    <td>Sun et al.<sup>14</sup></td>
                    <td align="center">*</td>
                    <td> </td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">8</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>15</sup></td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">*</td>
                    <td align="center">9</td>
                  </tr>
                </tbody>
              </table>
            </table-wrap>
          </p>
          <p>Eight studies achieved the maximum score of 9, indicating high methodological quality. These studies demonstrated strong cohort selection, appropriate comparability based on key factors, and robust outcome assessment with adequate follow-up periods. Three studies scored 8 points due to missing information regarding comparability factors or the absence of confirmation that the outcome was not present at baseline. However, several studies did not report essential demographic characteristics such as age, sex, or ethnicity &#8211; factors that are important for assessing external validity and subgroup fairness. Although these omissions are not fully captured by the NOS, they represent relevant limitations when considering the generalizability of AI model performance across diverse populations. Despite these issues, the overall quality of the studies was high, supporting the reliability of the review&#8217;s findings.</p>
        </sec>
        <sec>
          <title>Interpretability Techniques in AI-ECG Models</title>
          <p>Among the included studies, interpretability techniques were reported in nine out of eleven articles, as presented in <xref ref-type="table" rid="t5001">Table 5</xref>. Saliency maps were the most frequently applied to highlight ECG waveform regions most relevant to the models predictions (e.g., P wave, PR interval).<sup><xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref></sup> Other strategies included SHapley Additive exPlanations (SHAP),<sup><xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref></sup> which provided feature attribution in tree-based or CNN models; neuron ablation and CNN weight heatmaps,<sup><xref ref-type="bibr" rid="B16">16</xref></sup> which examined internal model behavior through perturbation; and variational autoencoders (VAEs), waveform averaging, and biological association analyses (GWAS, PheWAS) in Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> and Sau et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> to explore biological plausibility and ECG morphology patterns. Two studies did not report the use of any interpretability technique.<sup><xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref></sup></p>
          <p>
            <table-wrap id="t5001">
              <label>Table 5</label>
              <caption>
                <title>&#8211; Interpretability Techniques Applied in the Included Studies</title>
              </caption>
              <table frame="hsides" rules="groups">
                <colgroup>
                  <col/>
                  <col/>
                  <col/>
                </colgroup>
                <thead>
                  <tr>
                    <th align="left">Study</th>
                    <th>Interpretability Techniques</th>
                    <th>Description</th>
                  </tr>
                </thead>
                <tbody>
                  <tr>
                    <td>Sau et al.<sup>6</sup></td>
                    <td align="center">VAE, waveform averaging, PheWAS, GWAS, Cox models</td>
                    <td align="center">Linked AI-ECG predictions to biological traits and plausible pathophysiological mechanisms</td>
                  </tr>
                  <tr>
                    <td>Hughes et al.<sup>7</sup></td>
                    <td align="center">NR</td>
                    <td align="center">Compared performance with PCE and SEER risk scores, but did not report specific interpretability methods</td>
                  </tr>
                  <tr>
                    <td>Al-Alusi et al.<sup>8</sup></td>
                    <td align="center">Saliency maps</td>
                    <td align="center">Used to identify waveform segments predictive of hypertension</td>
                  </tr>
                  <tr>
                    <td>Sau et al.<sup>9</sup></td>
                    <td align="center">VAE, waveform averaging</td>
                    <td align="center">Visualized ECG morphology associated with sex-based risk discordance; VAE used for dimensionality reduction</td>
                  </tr>
                  <tr>
                    <td>Gao et al.<sup>16</sup></td>
                    <td align="center">Neuron ablation, CNN weight heatmaps</td>
                    <td align="center">Perturbed neurons in CNN layers and visualized associated weights; approach similar to occlusion sensitivity analysis</td>
                  </tr>
                  <tr>
                    <td>Raghunath et al.<sup>10</sup></td>
                    <td align="center">Saliency maps, expert feedback</td>
                    <td align="center">Provided visual explanations for normal ECGs flagged by AI; included cardiologist feedback to assess interpretability</td>
                  </tr>
                  <tr>
                    <td>Dhingra et al.<sup>11</sup></td>
                    <td align="center">NR</td>
                    <td align="center">Used image-based AI-ECG for risk prediction; did not report specific interpretability methods</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>12</sup></td>
                    <td align="center">Saliency maps</td>
                    <td align="center">Identified the P wave and PR interval as key mortality predictors using saliency visualization</td>
                  </tr>
                  <tr>
                    <td>Butler et al.<sup>13</sup></td>
                    <td align="center">SHAP</td>
                    <td align="center">Applied SHAP to ECG-derived features for heart failure risk stratification; identified most relevant ECG traits</td>
                  </tr>
                  <tr>
                    <td>Sun et al.<sup>14</sup></td>
                    <td align="center">SHAP, saliency maps</td>
                    <td align="center">SHAP used in XGB for feature attribution; Grad-CAM applied to CNN to highlight key ECG segments (PR, QRS, ST-T)</td>
                  </tr>
                  <tr>
                    <td>Lin et al.<sup>15</sup></td>
                    <td align="center">Saliency maps</td>
                    <td align="center">Applied visual saliency maps for model interpretation</td>
                  </tr>
                </tbody>
              </table>
              <table-wrap-foot>
                <fn id="TFN4001">
                  <p>AI-ECG: Artificial Intelligence-enabled Electrocardiogram; CNN: Convolutional Neural Network; ECG: Electrocardiogram; Grad-CAM: Gradient-weighted Class Activation Mapping; GWAS: Genome-Wide Association Study; NR: Not reported; PCE: Pooled Cohort Equation; PheWAS: Phenome-Wide Association Study; PR: PR interval; QRS: QRS complex; SEER: Surveillance, Epidemiology, and End Results (U.S. database); SHAP: SHapley Additive exPlanations; ST-T: ST segment and T wave; VAE: Variational Autoencoder; XGB: XGBoost.</p>
                </fn>
              </table-wrap-foot>
            </table-wrap>
          </p>
        </sec>
        <sec>
          <title>Clinical application and prospective use of AI-ECG</title>
          <p>Regarding clinical application, most studies explored the potential utility of AI-ECG models for early risk stratification, patient monitoring, or population-level screening; however, these remained retrospective or theoretical. Notably, only Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup>conducted a prospective application of their DL model: the algorithm was applied at the time of admission in newly recruited patients with suspected heart failure with preserved ejection fraction (HFpEF), and predictions were compared against invasive left ventricular pressure (LVEDP) measurements. This real-time validation scenario represents the only example of model use in a clinical decision-making context within the included studies. Although it represents the only real-time implementation scenario among the included studies, it does not constitute prospective validation for clinical outcome prediction.</p>
        </sec>
      </sec>
      <sec sec-type="discussion">
        <title>Discussion</title>
        <p>The reviewed studies demonstrate substantial advances in the application of AI to ECG analysis for cardiovascular risk stratification. DL techniques, particularly CNNs, have shown the ability to detect subtle ECG alterations associated with hypertension, ventricular dysfunction, and atrial fibrillation&#8212;even in tracings considered normal by specialists. These findings reinforce the clinical potential of AI-ECG for the early detection of asymptomatic or subclinical conditions, with the possibility of transforming clinical practice by enabling the identification of high-risk individuals and the implementation of early interventions from a single examination.</p>
        <p>In parallel, other important lines of investigation have explored AI-derived ECG biomarkers, such as electrocardiographic age, which have demonstrated strong associations with mortality and cardiovascular outcomes in large population-based cohorts. For example, previous studies have shown that discrepancies between AI-estimated ECG age and chronological age are linked to increased mortality risk and incident cardiovascular events, including findings from community-based cohorts.<sup><xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref></sup> These approaches highlight the broader potential of AI-ECG to capture latent physiological and biological aging signals. Although such studies were not included in the present review due to our predefined focus on models directly predicting clinical outcomes from ECG signals alone, they provide important complementary evidence supporting the prognostic value of AI-ECG.</p>
        <p>Despite this progress, important methodological limitations undermine direct comparisons across studies. The heterogeneity of AI models employed, the diversity of performance metrics used, and the variation in clinical outcomes analyzed hinder the consolidation of results. A crucial consideration concerns the distinction between statistical validity and clinical validity: although metrics such as AUROC and HR indicate high performance, it remains essential to determine whether the abnormalities identified by the models correspond to physiologically meaningful phenotypes with practical implications for clinical management. Otherwise, there is a risk that algorithms may detect only statistical signatures without genuine clinical relevance.</p>
        <p>An analysis of the results shows that, for all-cause mortality, the studies employed different metrics &#8211; such as AUROC and HR &#8211; as well as varying follow-up periods (e.g., 1-year vs. 5-year mortality), which complicates direct comparisons. Nevertheless, in studies using the same metrics, the models demonstrated consistent performance across all time horizons. A similar limitation was observed for cardiovascular mortality, although all models reported high predictive accuracy. Only two studies evaluated ASCVD, each using distinct metrics, which limited direct comparisons; despite slightly lower discrimination compared with all-cause mortality, results remained satisfactory. For acute myocardial infarction, four studies reported outcomes: three using HR values ranging from 1.44 to 3.53 across different cohorts, and two reporting AUROC values around 0.85, indicating consistent predictive performance. Heart failure was among the outcomes with the highest reported performance, achieving an AUROC of 0.90 and a HR of 32.06. Finally, some studies also assessed stroke, reporting HRs between 1.6 and 2.3, supporting the ability of AI-ECG models to capture risk across multiple cardiovascular outcomes.</p>
        <p>AI-ECG models tend to demonstrate stronger performance for all-cause mortality compared with more specific cardiovascular outcomes such as ASCVD. This difference likely reflects the fact that all-cause mortality encompasses a broader spectrum of systemic and cardiac abnormalities that may be detectable in ECG signals, whereas atherosclerotic disease does not consistently produce measurable electrical or structural changes before clinical events occur. As a result, subclinical atherosclerosis may remain undetected by ECG-based models, limiting their predictive performance for ASCVD. Additionally, ASCVD prediction showed substantial heterogeneity across studies, including differences in sample size and outcome definitions, which further restricts direct comparability and may contribute to variability in reported results.</p>
        <p>Comparisons between AI-ECG models and traditional risk scores should be interpreted with caution, as many conventional tools were not originally designed to predict outcomes such as all-cause mortality. Differences in intended prediction targets may partially explain variations in reported performance metrics and limit direct comparability. From a clinical perspective, these findings suggest that AI-ECG models and traditional risk scores may serve complementary roles: while conventional tools remain more appropriate for estimating disease-specific risk (e.g., ASCVD), AI-ECG may offer broader risk stratification by capturing global physiological signals, thereby supporting a more integrated assessment of patient risk.</p>
        <p>Nonetheless, methodological gaps remain. Not all studies reported the number of clinical events, some omitted calibration metrics, and many focused on short-term outcomes, limiting extrapolation to long-term risk. Another underexplored aspect is temporal generalization: most models were developed in static cohorts, without assessing whether performance remains stable over time in populations subject to epidemiological, therapeutic, or lifestyle changes. Continuous model updating and prospective monitoring will be necessary to ensure lasting clinical utility of AI-ECG.</p>
        <p>Additionally, our study selection was restricted to models using ECG signals as the sole input variable, enhancing consistency but not fully reflecting real-world clinical practice, where multiple patient factors are considered. Therefore, when combined with traditional clinical variables, AI-ECG has the potential to function as a digital biomarker, improving the identification of high-risk individuals and reducing underdiagnosis. Future research should advance further by integrating AI-ECG with laboratory data, imaging studies, and even information from wearable devices, moving toward a multimodal approach capable of personalizing cardiovascular risk stratification.</p>
        <p>The scalability of AI-ECG represents a major advantage, particularly in resource-limited settings, given the accessibility and simplicity of ECG acquisition. However, most models were developed and validated using data from high-income countries, predominantly the United States, which limits their generalizability. In many cases, validations consisted of replications within related healthcare systems, biobanks, or populations with similar demographic and clinical profiles, rather than assessments in distinct and heterogeneous cohorts. Distinguishing between internal replication and genuine external validation is essential, as the latter provides the most rigorous test of the robustness and transportability of AI-ECG models.</p>
        <p>The study by Sau et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> was among the few to perform external validation in a geographically or demographically distinct population&#8212;specifically in Brazil&#8212;providing stronger evidence of generalizability. A second study, by Dhingra et al.,<sup><xref ref-type="bibr" rid="B11">11</xref></sup> also included a Brazilian cohort, confirming that even in populations with different sociodemographic and clinical characteristics, the models maintained strong performance. Al-Alusi et al.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> conducted both internal and external validations in similar populations, with minor differences, such as a higher proportion of women and Black individuals in the external cohort. Nevertheless, comorbidity profiles were largely comparable across groups. Conversely, Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> did not report detailed population characteristics, limiting cohort comparisons. Similarly, in the studies by Lin et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> and Lin et al.,<sup><xref ref-type="bibr" rid="B15">15</xref></sup> external validation was limited due to the absence of clinical information, allowing only the assessment of one-year mortality as an outcome.</p>
        <p>The limited number of truly independent validations, combined with incomplete demographic data, remains a major barrier to the widespread clinical adoption of these tools, as it hampers the evaluation of population diversity and reduces both comparability and interpretability of results.</p>
        <p>Among the strengths of the included studies are the use of large datasets, advanced AI methodologies &#8211; particularly CNNs &#8211; and extended follow-up periods. CNNs demonstrated notable effectiveness in analyzing raw ECG signals, capturing temporal and spatial features with minimal preprocessing. In contrast, studies categorized as DNNs or general deep learning often lacked architectural specificity and provided limited interpretability.</p>
        <p>In this context, most of the analyzed studies employed interpretability techniques to enhance the transparency of AI models applied to ECG, reflecting the growing demand for explainable systems. The most commonly used method was saliency maps, which generate heatmaps over the ECG tracing to highlight regions presumed to influence predictions, such as the P wave, QRS complex, or PR interval. Despite their popularity, evidence from medical imaging research indicates that such maps may be unstable, poorly reproducible, and, in some cases, provide misleading explanations that appear plausible but do not reflect the model&#8217;s true reasoning.<sup><xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref></sup></p>
        <p>SHAP was also employed in several studies. This method, grounded in game theory, assigns quantitative values to the contribution of each variable to the model&#8217;s output, enabling the quantification of each feature&#8217;s influence. Unlike purely gradient-based methods, SHAP provides an additive attribution framework that integrates local and global interpretations, offering greater consistency between the two. Its ability to generate intuitive visualizations &#8211; such as color plots highlighting important pixels in images or temporal segments in ECG signals &#8211; has made it particularly valuable in biomedical applications. Limitations include high computational cost, challenges related to highly correlated temporal variables, and potential inconsistencies in capturing feature dependencies.<sup><xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref></sup></p>
        <p>Other interpretability approaches, including neuron ablation combined with CNN weight heatmaps, VAEs, and biological association analyses (e.g., GWAS, PheWAS), have also been explored to provide additional insights into model behavior. These techniques can shed light on latent representations or the biological plausibility of predictions, but they are often constrained by technical complexity, limited model accessibility, and underlying data biases.<sup><xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref></sup></p>
        <p>Overall, the current landscape indicates that although multiple interpretability techniques are available, their integration into clinical practice remains limited. As observed by Yanagawa and Sato,<sup><xref ref-type="bibr" rid="B22">22</xref></sup> future efforts should prioritize methods that not only clarify model predictions but also bring clinical understanding closer to ECG patterns, potentially uncovering subtle physiological markers or disease subtypes not easily identifiable through conventional analysis. Combining interpretability with systematic validation and prospective clinical assessment may transform AI models from purely predictive tools into instruments that actively support clinical reasoning and hypothesis generation.</p>
        <p>Notably, only Gao et al.<sup><xref ref-type="bibr" rid="B16">16</xref></sup> prospectively tested AI-ECG in a real-world setting, applying the model at patient admission to estimate the risk of HFpEF in comparison with invasive measurements. While models such as AIRE6 and ECG-surv<sup><xref ref-type="bibr" rid="B15">15</xref></sup> demonstrated strong associations with physiological characteristics and survival, broader clinical implementation remains incipient.</p>
        <p>For AI-ECG to transition from experimental use to clinical practice, integration with healthcare systems and electronic health records is essential. Regulatory frameworks must evolve to evaluate models not only on accuracy but also on fairness, transparency, and clinical utility. Investments in infrastructure, interoperability, and training of healthcare professionals are required to ensure the safe and effective use of these technologies. Prospective clinical trials are equally critical, particularly in primary prevention contexts.</p>
        <p>Finally, public health strategies may benefit from AI-ECG by enabling the early detection of high-risk individuals, potentially reducing long-term costs and improving outcomes. However, ethical considerations &#8211; such as algorithmic transparency, data privacy, accountability, and equitable access &#8211; must be addressed. Models trained on homogeneous, privileged populations risk perpetuating health disparities unless adapted to diverse clinical and demographic contexts.</p>
        <p>In summary, the effective use of AI-ECG demands a transdisciplinary approach that bridges technological innovation, clinical practice, public health, and ethics. Only through this lens can AI-ECG evolve from a promising tool into a transformative instrument for reducing health inequities, optimizing care, and advancing population health.</p>
      </sec>
      <sec sec-type="conclusions">
        <title>Conclusion</title>
        <p>The application of AI to ECG analysis represents a promising advancement in personalized cardiovascular risk assessment. Nonetheless, further research is needed to ensure the safety, effectiveness, and equitable clinical integration of these technologies. The current evidence, while compelling, underscores the need for more diverse populations in training and validation datasets, improved model interpretability, and prospective clinical trials to fully assess the real-world impact and generalizability of AI-ECG models. Only through rigorous validation and thoughtful implementation can AI-ECG transition from a research frontier to a transformative tool in cardiovascular medicine.</p>
      </sec>
    </body>
    <back>
      <ack>
        <title>Acknowledgement</title>
        <p>The authors used Grammarly, an AI-based language editing tool, to assist with English-language revision and to improve the clarity and fluency of the manuscript. All content was reviewed and approved by the authors.</p>
      </ack>
      <fn-group>
        <fn fn-type="other">
          <label>Study Association:</label>
          <p> This study is not associated with any thesis or dissertation work.</p>
        </fn>
        <fn fn-type="other">
          <label>Ethics Approval and Consent to Participate:</label>
          <p> This article does not contain any studies with human participants or animals performed by any of the authors.</p>
        </fn>
        <fn fn-type="other">
          <label>Use of Artificial Intelligence:</label>
          <p> During the preparation of this work, the author(s) used Grammarly to improve the English language, grammar, spelling, and overall clarity of the manuscript. After using this tool/service, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article.</p>
        </fn>
        <fn fn-type="data-availability" specific-use="data-available-upon-request">
          <label>Availability of Research Data:</label>
          <p> All datasets supporting the results of this study are available upon request from the corresponding author.</p>
        </fn>
        <fn fn-type="financial-disclosure">
          <label>Sources of Funding:</label>
          <p> There were no external funding sources for this study.</p>
        </fn>
      </fn-group>
    </back>
  </sub-article>
</article>
