<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.1 20151215//EN" "https://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="other" 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">ZXcY7GKRpnKdzQT46DqvF7z</article-id>
      <article-id specific-use="scielo-v2" pub-id-type="publisher-id">S0066-782X2020001100579</article-id>
      <article-id pub-id-type="publisher-id">abc.20200143</article-id>
      <article-id pub-id-type="doi">10.36660/abc.20200143</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Ponto de Vista</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>&#201;tica, Intelig&#234;ncia Artificial e Cardiologia</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-0381-3344</contrib-id>
          <name>
            <surname>Souza</surname>
            <given-names>Erito Marques de</given-names>
            <suffix>Filho</suffix>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff2">
            <sup>2</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Fernandes</surname>
            <given-names>Fernando de Amorim</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff3">
            <sup>3</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-7186-9876</contrib-id>
          <name>
            <surname>Pereira</surname>
            <given-names>Nikolas Cunha de Assis</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0002-1466-9413</contrib-id>
          <name>
            <surname>Mesquita</surname>
            <given-names>Claudio Tinoco</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff3">
            <sup>3</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Gismondi</surname>
            <given-names>Ronaldo Altenburg</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff5">
            <sup>5</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff1">
        <label>1</label>
        <institution content-type="orgname">Universidade Federal Fluminense</institution>
        <addr-line>
          <named-content content-type="city">Niter&#243;i</named-content>
          <named-content content-type="state">RJ</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">P&#243;s-gradua&#231;&#227;o em Ci&#234;ncias Cardiovasculares da Universidade Federal Fluminense,Niter&#243;i, RJ - Brasil</institution>
      </aff>
      <aff id="aff2">
        <label>2</label>
        <institution content-type="orgname">Universidade Federal Rural do Rio de Janeiro</institution>
        <institution content-type="orgdiv1">Departamento de Tecnologias e Linguagens</institution>
        <addr-line>
          <named-content content-type="city">Nova Igua&#231;u</named-content>
          <named-content content-type="state">RJ</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">Universidade Federal Rural do Rio de Janeiro - Departamento de Tecnologias e Linguagens,Nova Igua&#231;u, RJ &#8211; Brasil</institution>
      </aff>
      <aff id="aff3">
        <label>3</label>
        <institution content-type="orgdiv2">Setor de Medicina Nuclear do Servi&#231;o de Radiologia</institution>
        <institution content-type="orgdiv1">Hospital Universit&#225;rio Ant&#244;nio Pedro</institution>
        <institution content-type="orgname">UFF</institution>
        <addr-line>
          <named-content content-type="city">Niter&#243;i</named-content>
          <named-content content-type="state">RJ</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">Setor de Medicina Nuclear do Servi&#231;o de Radiologia do Hospital Universit&#225;rio Ant&#244;nio Pedro (EBESERH-HUAP-UFF),Niter&#243;i, RJ - Brasil</institution>
      </aff>
      <aff id="aff4">
        <label>4</label>
        <institution content-type="orgdiv1">Faculdade de Medicina</institution>
        <institution content-type="orgname">Universidade Federal Fluminense</institution>
        <addr-line>
          <named-content content-type="city">Niter&#243;i</named-content>
          <named-content content-type="state">RJ</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">Faculdade de Medicina da Universidade Federal Fluminense,Niter&#243;i, RJ - Brasil</institution>
      </aff>
      <aff id="aff5">
        <label>5</label>
        <institution content-type="orgname">Hospital Niter&#243;i D&#180;or</institution>
        <addr-line>
          <named-content content-type="city">Niter&#243;i</named-content>
          <named-content content-type="state">RJ</named-content>
        </addr-line>
        <country country="BR">Brasil</country>
        <institution content-type="original">Hospital Niter&#243;i D&#180;or,Niter&#243;i, RJ &#8211; Brasil</institution>
      </aff>
      <author-notes>
        <corresp id="c01"><bold>Correspond&#234;ncia: Erito Marques de Marques</bold> &#8226; Universidade Federal Fluminense - Departamento de Medicina Cl&#237;nica &#8211; Av. Marques do Paran&#225;, 303. CEP 24033-900, Niter&#243;i, RJ &#8211; Brasil. E-mail: <email>mederitomarques@gmail.com</email> </corresp>
        <fn fn-type="con">
          <p>Contribui&#231;&#227;o dos autores</p>
          <p>Concep&#231;&#227;o e desenho da pesquisa e Reda&#231;&#227;o do manuscrito: Souza Filho EM, Fernandes FA, Pereira NCA, Mesquita CT, Gismondi RA; Obten&#231;&#227;o de dados: Souza Filho EM, Pereira NCA, Mesquita CT, Gismondi RA; An&#225;lise e interpreta&#231;&#227;o dos dados: Souza Filho EM, Fernandes FA, Mesquita CT, Gismondi RA; Revis&#227;o cr&#237;tica do manuscrito quanto ao conte&#250;do intelectual importante: Souza Filho EM, Mesquita CT, Gismondi RA.</p>
        </fn>
        <fn fn-type="conflict">
          <p>Potencial conflito de interesses</p>
          <p>Declaro n&#227;o haver conflito de interesses pertinentes.</p>
        </fn>
      </author-notes>
      <pub-date date-type="pub" publication-format="electronic">
        <day>28</day>
        <month>09</month>
        <year>2020</year>
      </pub-date>
      <pub-date date-type="collection" publication-format="electronic">
        <month>09</month>
        <year>2020</year>
      </pub-date>
      <volume>115</volume>
      <issue>3</issue>
      <fpage>579</fpage>
      <lpage>583</lpage>
      <history>
        <date date-type="received">
          <day>20</day>
          <month>02</month>
          <year>2020</year>
        </date>
        <date date-type="rev-recd">
          <day>04</day>
          <month>04</month>
          <year>2020</year>
        </date>
        <date date-type="accepted">
          <day>30</day>
          <month>04</month>
          <year>2020</year>
        </date>
      </history>
      <permissions>
        <license license-type="open-access" xlink:href="https://creativecommons.org/licenses/by/4.0/" xml:lang="en">
          <license-p> This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. </license-p>
        </license>
      </permissions>
      <kwd-group xml:lang="pt">
        <kwd>Intelig&#234;ncia Artificial</kwd>
        <kwd>Programas Inform&#225;ticos</kwd>
        <kwd>Tomada de decis&#245;es assistida por computador</kwd>
        <kwd>Gest&#227;o de Informa&#231;&#245;es</kwd>
      </kwd-group>
      <counts>
        <fig-count count="2"/>
        <table-count count="0"/>
        <equation-count count="0"/>
        <ref-count count="18"/>
        <page-count count="5"/>
      </counts>
    </article-meta>
  </front>
  <body>
    <disp-quote>
      <p>&#8220;De longe, o maior perigo da Intelig&#234;ncia Artificial &#233; que as pessoas concluam muito cedo que a entendem. &#8221;</p>
      <p>Eliezer Yudkowsky</p>
    </disp-quote>
    <sec sec-type="intro">
      <title>Introdu&#231;&#227;o</title>
      <p>Em um futuro n&#227;o muito distante, um programa de computador artificialmente inteligente provavelmente diagnosticar&#225; as doen&#231;as card&#237;acas com mais precis&#227;o que um cardiologista certificado. O conhecimento biom&#233;dico cresce significativamente, impossibilitando que os profissionais de sa&#250;de contempor&#226;neos se mantenham atualizados sobre todos os conte&#250;dos publicados em seu campo. Da mesma forma, informa&#231;&#245;es sobre os pacientes est&#227;o cada vez mais numerosas e acess&#237;veis, tornando impratic&#225;vel o gerenciamento, filtragem e sele&#231;&#227;o em tempo real. Nesse contexto, a Intelig&#234;ncia Artificial (IA) tem um papel relevante na tomada de decis&#245;es em sa&#250;de. &#201; resultado da combina&#231;&#227;o de modelos matem&#225;ticos sofisticados e da computa&#231;&#227;o, para produzir algoritmos refinados capazes de emular (ou imitar) a intelig&#234;ncia humana.<sup>1</sup> A IA permitiu aplica&#231;&#245;es interessantes em praticamente todos os campos da medicina e do conhecimento humano. Particularmente, na cardiologia, v&#225;rias aplica&#231;&#245;es se mostraram exitosas. Han et al.,<sup>2</sup> por exemplo, usaram o Machine Learning (ML), um subconjunto da IA, para analisar se essa ferramenta seria &#250;til para identificar pacientes com risco de r&#225;pida progress&#227;o de placa coronariana. Foram utilizadas caracter&#237;sticas epidemiol&#243;gicas cl&#237;nicas e informa&#231;&#245;es quantitativas e qualitativas da angiografia tomogr&#225;fica computadorizada das coron&#225;rias (todas obtidas a partir do estudo PARADIGM). No total, 1.083 pacientes foram inclu&#237;dos no estudo, sendo testados 10 modelos diferentes. O LogitBoost apresentou melhor desempenho. A &#225;rea sob a curva (AUC) <italic>receiver operating characteristic</italic> foi de 0,83 &#8212; melhor que o escore de risco de doen&#231;a cardiovascular ateroscler&#243;tica de 10 anos (escore de risco DCA), que foi de 0,59. Em outro estudo, Than et al.,<sup>3</sup> avaliaram se o <italic>Gradient Boosting</italic> (tamb&#233;m um algoritmo de ML) seria ben&#233;fico na previs&#227;o da probabilidade de infarto agudo do mioc&#225;rdio tipo 1. Foram considerados aspectos como sexo, idade, taxa de altera&#231;&#227;o da concentra&#231;&#227;o card&#237;aca de troponina I e troponina card&#237;aca pareada I de uma amostra com 11.011 pacientes. A AUC foi de 0,96 e o modelo de ML teve melhor desempenho do que a rota tradicional da Sociedade Europeia de Cardiologia de 0/3 horas. Hedman et al.,<sup>4</sup> desenvolveram um algoritmo de ML para descrever grupos de pacientes com insufici&#234;ncia card&#237;aca com fra&#231;&#227;o de eje&#231;&#227;o preservada com base em seu fen&#243;tipo. Eles usaram dados cl&#237;nicos e eletrocardiogr&#225;ficos. Foram identificados seis grupos diferentes, com diferentes n&#237;veis de prote&#237;nas inflamat&#243;rias e cardiovasculares e tamb&#233;m com diferentes desfechos. Diante disso, na cardiologia, o processo de incorpora&#231;&#227;o da IA na pr&#225;tica cl&#237;nica &#233; acelerado. O uso da IA na cardiologia est&#225; presente em nossa rotina di&#225;ria, como o reconhecimento de fen&#243;tipos de doen&#231;as, diagn&#243;stico, progn&#243;stico e algoritmos de tratamento. A IA tem um enorme potencial disruptivo e alguns defendem a possibilidade do surgimento de uma nova esp&#233;cie, o Homo incredibile<sup>1</sup>, que apoia suas decis&#245;es sobre dados e promove uma revolu&#231;&#227;o no ecossistema digital. No entanto, essa mudan&#231;a de paradigma trouxe in&#250;meros desafios, infelizmente. Quest&#245;es &#233;ticas s&#227;o uma grande preocupa&#231;&#227;o em rela&#231;&#227;o a essas novas tecnologias, e discutiremos algumas delas, bem como poss&#237;veis solu&#231;&#245;es e precau&#231;&#245;es.</p>
      <sec>
        <title>Aspectos &#201;ticos</title>
        <sec>
          <title>Discrimina&#231;&#227;o e Privacidade de Dados</title>
          <p>Esses algoritmos podem, por exemplo, ser usados para discriminar pessoas, dar vida a dispositivos que colocam outras vidas em risco ou at&#233; para produzir e divulgar not&#237;cias falsas &#8212; sem falar dos poss&#237;veis danos em caso de pol&#237;ticas inadequadas de seguran&#231;a da informa&#231;&#227;o.<sup>5-7</sup> O sequestro de arquivos, ocorrido em 2017, com mais de 300 milh&#245;es de computadores atacados pelo ransomware WannaCry em 150 pa&#237;ses e o vazamento de dados pela empresa Ashley Madison em 2015, s&#227;o exemplos do potencial destrutivo de a&#231;&#245;es de hackers. Isso exemplifica alguns obst&#225;culos a serem superados na inclus&#227;o de dispositivos m&#233;dicos port&#225;teis na pr&#225;tica cl&#237;nica e no uso de sistemas aut&#244;nomos para apoiar a tomada de decis&#245;es na &#225;rea da sa&#250;de.</p>
          <p>A obten&#231;&#227;o do consentimento informado &#233; uma preocupa&#231;&#227;o da maioria dos bioeticistas. Os modelos atuais de IA dependem muito das informa&#231;&#245;es contidas nos prontu&#225;rios m&#233;dicos. &#201; poss&#237;vel garantir que as informa&#231;&#245;es pessoais permane&#231;am confidenciais, mesmo com dados que circulam pela internet? O vazamento de informa&#231;&#245;es m&#233;dicas de pessoas famosas, como a ex-primeira-dama Marisa Silva e o atual Presidente, s&#227;o apenas alguns exemplos de problemas relacionados &#224; confidencialidade de dados.</p>
        </sec>
        <sec>
          <title>Transpar&#234;ncia e Seguran&#231;a</title>
          <p>Al&#233;m disso, quando se trata de ci&#234;ncias m&#233;dicas e da sa&#250;de, especificamente, outros riscos se destacam. Um deles &#233; a falta de transpar&#234;ncia na tomada de decis&#245;es ou a incapacidade de explicar o &#8220;racioc&#237;nio&#8221; por tr&#225;s do resultado final, representado pelas chamadas caixas pretas. Foram feitas pesquisas em busca de solu&#231;&#245;es. No entanto, a realidade atual &#233; que os respons&#225;veis pela maioria das grandes coisas feitas com o <italic>Deep Learning</italic> n&#227;o sabem explicar completamente o funcionamento de seus eficientes sistemas.<sup>8-10</sup> Por outro lado, nem sempre &#233; poss&#237;vel fornecer explica&#231;&#245;es detalhadas sobre a fisiopatologia de certas doen&#231;as ou sobre o mecanismo de a&#231;&#227;o de alguns medicamentos, mesmo que os ensaios cl&#237;nicos tenham mostrado benef&#237;cios para o paciente. Isso aumenta o desafio de garantir a reprodutibilidade e a replicabilidade dos algoritmos de IA. Como apontado por Beam et al., um estudo &#233; reproduz&#237;vel se, com base no acesso a dados e na an&#225;lise do c&#243;digo do algoritmo, um grupo independente puder obter os mesmos resultados observados no estudo original, enquanto a replicabilidade est&#225; associada ao fato de um grupo independente conseguir estudar o mesmo fen&#244;meno e obter as mesmas conclus&#245;es ap&#243;s a realiza&#231;&#227;o de um conjunto de experimentos ou an&#225;lises de um novo conjunto de dados.<sup>11</sup> Outras quest&#245;es relevantes s&#227;o a seguran&#231;a dos dados dos pacientes e a conscientiza&#231;&#227;o dos pacientes sobre o uso de seus dados. Uma parceria entre o Sistema Nacional de Sa&#250;de Brit&#226;nico (NHS) e uma subsidi&#225;ria de uma grande empresa privada de tecnologia em 2015, que incluiu a transfer&#234;ncia n&#227;o consentida de uma base de dados identific&#225;vel de mais de 1,6 milh&#227;o de habitantes, &#233; um dos casos mais famosos e controversos at&#233; o momento. Apesar das boas inten&#231;&#245;es de ambos os lados, ficou claro o quanto podemos estar expostos se n&#227;o discutirmos, agora, at&#233; que ponto os dados pertencem a um indiv&#237;duo. Al&#233;m disso, com esse tipo de contrato, grandes empresas de tecnologia tendem a aumentar ainda mais o oligop&#243;lio existente.</p>
        </sec>
        <sec>
          <title>Valores e Prefer&#234;ncias do Paciente, Julgamento Cl&#237;nico e Empatia</title>
          <p>O contato humano entre m&#233;dicos e pacientes &#233; um dos fundamentos da medicina desde Hip&#243;crates. Existem d&#250;vidas sobre se a IA &#233; capaz de levar em considera&#231;&#227;o o contexto social da pessoa, fatores ambientais, prefer&#234;ncias e valores morais no algoritmo de decis&#227;o do tratamento.</p>
          <p>Outro aspecto importante &#233; a representa&#231;&#227;o de minorias &#233;tnicas, sociais e culturais nos prontu&#225;rios m&#233;dicos que servem de base para o algoritmo da IA. Se esses dados n&#227;o forem muito representativos ou forem distorcidos, poder&#227;o ocorrer erros de interpreta&#231;&#227;o.</p>
        </sec>
      </sec>
    </sec>
    <sec>
      <title>Medidas a Serem Implementadas</title>
      <p>Conforme definido por Keskinbora,<sup>12</sup> para confiar na IA, precisamos do seguinte:</p>
      <list list-type="bullet">
        <list-item>
          <p>Transpar&#234;ncia de dados, opera&#231;&#227;o e algoritmos</p>
        </list-item>
        <list-item>
          <p>Credibilidade e auditabilidade, incluindo o relat&#243;rio de vieses e erros</p>
        </list-item>
        <list-item>
          <p>Confiabilidade, com IA clinicamente validada</p>
        </list-item>
        <list-item>
          <p>Recuperabilidade, permitindo o controle manual da opera&#231;&#227;o, se necess&#225;rio</p>
        </list-item>
      </list>
      <p>Esse cen&#225;rio traz consigo a necessidade de discuss&#227;o sobre o uso da IA na sa&#250;de, e seus limites, considerando os princ&#237;pios fundamentais da bio&#233;tica na sa&#250;de: justi&#231;a, n&#227;o malefic&#234;ncia, benefic&#234;ncia, equidade, igualdade, aceita&#231;&#227;o social e respeito &#224; autonomia do paciente.<sup>13</sup> A quest&#227;o que surge nesse contexto &#233;: como incorporar a IA &#224; pr&#225;tica biom&#233;dica, respeitando esses princ&#237;pios para gerar valor? Embora n&#227;o haja resposta definitiva para a pergunta, uma estrat&#233;gia promissora (<xref ref-type="fig" rid="f01">Figura 1</xref>) inclui:</p>
      <p>
        <fig id="f01">
          <label>Figura 1</label>
          <caption>
            <title>&#8211; Estrat&#233;gia proposta para implementa&#231;&#227;o da Intelig&#234;ncia Artificial na pr&#225;tica cl&#237;nica considerando a &#233;tica. IA: intelig&#234;ncia artificial.</title>
          </caption>
          <graphic xlink:href="1678-4170-abc-115-03-0579-00143-gf01.tif"/></fig>
      </p>
      <list list-type="alpha-lower">
        <list-item>
          <p><bold>Coopera&#231;&#227;o:</bold> Os modelos de IA tendem a ter desempenho melhor quando temos dados saud&#225;veis sobre o que queremos estudar. Assim, a colabora&#231;&#227;o interinstitucional exerce papel fundamental nesse processo, pois o compartilhamento desses dados favorece a obten&#231;&#227;o de m&#233;tricas de excel&#234;ncia.<sup>1</sup></p>
        </list-item>
        <list-item>
          <p><bold>Literacia em Sa&#250;de:</bold> refere-se ao n&#237;vel de informa&#231;&#245;es de sa&#250;de que cada indiv&#237;duo &#233; capaz de obter, gerenciar e entender para aplicar no processo de tomada de decis&#245;es em sa&#250;de.<sup>14</sup> Indiv&#237;duos com maior n&#237;vel de literacia tendem a tomar melhores decis&#245;es em sa&#250;de. Assim, &#224; medida que os modelos de IA s&#227;o incorporados &#224; pr&#225;tica cl&#237;nica, &#233; essencial que a literacia sobre eles tamb&#233;m seja ampliada. Isso inclui uma rela&#231;&#227;o m&#233;dico-paciente ampliada, preocupada em incluir o paciente no centro da tomada de decis&#227;o multifatorial e multiprofissional. Da mesma forma, a literacia dos profissionais de sa&#250;de em IA deve ser incentivada.</p>
        </list-item>
        <list-item>
          <p><bold>Seguran&#231;a e privacidade:</bold> os dados criptografados s&#227;o apenas o primeiro passo em medidas mais gerais para garantir a privacidade dos dados. O esc&#226;ndalo da Cambridge Analytica foi um grande aviso sobre os poss&#237;veis danos causados pelo mau uso dos BIG DATA. Nesse contexto, o estrito cumprimento do GDPR<sup>15</sup> deve ser visto como um direito fundamental de qualquer ser humano, para o qual nenhum esfor&#231;o deve ser feito para garantir. Essa &#233; uma das quest&#245;es mais centrais na refer&#234;ncia &#233;tica para a implementa&#231;&#227;o da IA e precisa estar bem estabelecida. Os prontu&#225;rios m&#233;dicos eletr&#244;nicos s&#227;o os dados mais valiosos administrados por m&#233;dicos e cardiologistas nos dias de hoje. Outra quest&#227;o importante &#233; a prote&#231;&#227;o das imagens fotogr&#225;ficas dos pacientes na medida em que se aplicam &#224; tecnologia de reconhecimento facial, o que pode amea&#231;ar o consentimento informado adequado e a seguran&#231;a dos pacientes.<sup>16</sup></p>
        </list-item>
        <list-item>
          <p><bold>Objetivo:</bold> A IA deve ser usada como uma ferramenta cujo objetivo &#233; promover a qualidade de vida, a sa&#250;de e o bem-estar dos seres humanos. Permitir que os interesses econ&#244;micos superem as reais necessidades humanas &#233; um erro grave que pode ter consequ&#234;ncias desastrosas. Keskinbora<sup>12</sup> sugere, por exemplo, o desenvolvimento de uma IA livre, criada com um objetivo comum e cujo sentimento se baseie na opera&#231;&#227;o de acordo com valores &#233;ticos.</p>
        </list-item>
        <list-item>
          <p><bold>Otimiza&#231;&#227;o do tempo:</bold> o modus operandi da for&#231;a de trabalho traz consigo tarefas que requerem in&#250;meras horas dos seres humanos em seu trabalho, realizando tarefas repetitivas. Muitas dessas tarefas podem ser substitu&#237;das por m&#225;quinas com desempenho melhor ou semelhantes aos humanos. Isso cria uma oportunidade na qual os seres humanos t&#234;m potencial para trabalhar menos, em empregos mais especializados. O tempo extra que eles t&#234;m deve ser investido em estudos adicionais, atividades de lazer, atividade f&#237;sica, dedica&#231;&#227;o &#224; fam&#237;lia etc. Isso certamente implica uma nova reformula&#231;&#227;o das pol&#237;ticas trabalhistas e educacionais.</p>
        </list-item>
        <list-item>
          <p><bold>Auditoria de erros e vigil&#226;ncia p&#250;blica:</bold> Os modelos de IA podem cometer erros e suas decis&#245;es nem sempre s&#227;o compreens&#237;veis para os seres humanos. Portanto, &#233; essencial que os algoritmos sejam auditados periodicamente e que suas m&#233;tricas de desempenho sejam informadas aos pacientes antes de tomar decis&#245;es sobre sua sa&#250;de. H&#225; um desafio importante aqui, que est&#225; relacionado ao desenvolvimento de um aparato jur&#237;dico espec&#237;fico sobre o assunto.</p>
        </list-item>
        <list-item>
          <p><bold>Educa&#231;&#227;o:</bold> ao longo de sua vida, diferentes conjuntos de conhecimentos humanos se tornam in&#250;teis. Em um mundo vol&#225;til, inst&#225;vel, complexo e amb&#237;guo, as tecnologias disruptivas podem tornar obsoletas as habilidades e os conhecimentos anteriores ao longo do tempo: &#233; a meia-vida do conhecimento biom&#233;dico. A solu&#231;&#227;o &#233; o processo de estudo cont&#237;nuo, onde os seres humanos estudar&#227;o para sempre! Outro ponto &#233; o qu&#234; estudar; certamente, um modelo centrado na mem&#243;ria deve ser substitu&#237;do por um modelo voltado para a solu&#231;&#227;o de problemas reais da sociedade. Para tanto, &#233; obrigat&#243;rio ampliar os estudos de matem&#225;tica, computa&#231;&#227;o e ci&#234;ncias b&#225;sicas nos cursos de gradua&#231;&#227;o e p&#243;s-gradua&#231;&#227;o em sa&#250;de com esse objetivo.<sup>1</sup></p>
        </list-item>
        <list-item>
          <p><bold>Aten&#231;&#227;o aos vieses:</bold> Os modelos de Machine Learning (um subconjunto da IA) produzem suas respostas de acordo com os dados que s&#227;o usados como entradas para os algoritmos. Assim, &#233; poss&#237;vel gerar comportamento discriminat&#243;rio em rela&#231;&#227;o a determinados grupos. Por exemplo, se n&#227;o houver dados suficientes para serem usados no treinamento de algoritmos. Um exemplo desse vi&#233;s foi o <italic>chatbot</italic> do programa Tay da Microsoft, que aprendeu linguagem racista e sexista e precisou ser removido no dia do seu lan&#231;amento.<sup>17,18</sup></p>
        </list-item>
      </list>
    </sec>
    <sec>
      <title>Exemplo de Caso</title>
      <p>Veja o seguinte exemplo hipot&#233;tico. Um homem de 70 anos sofre de insufici&#234;ncia card&#237;aca: dispneia ao esfor&#231;o, ortopneia, crepita&#231;&#245;es nas bases pulmonares e edema dos membros inferiores. Preocupado com sua situa&#231;&#227;o, acessou rapidamente um site de diagn&#243;sticos m&#233;dicos online, que mostrou 99% de probabilidade de insufici&#234;ncia card&#237;aca. Para economizar tempo, ele realizou um ecocardiograma sozinho. Um algoritmo de diagn&#243;stico de IA mostrou 97% de chances de cardiomiopatia dilatada idiop&#225;tica, com progn&#243;stico de 12 meses de sobrevida de apenas 13% e contraindica&#231;&#227;o ao transplante. E isso veio escrito no prontu&#225;rio autom&#225;tico gerado por um computador com IA.</p>
      <p>Chateado com a situa&#231;&#227;o, ele vendeu todos os seus pertences e reservou uma viagem pelos cinco continentes para o m&#234;s seguinte, mas a companhia a&#233;rea exigiu um atestado m&#233;dico, alegando que havia 80% de chance de complica&#231;&#245;es a bordo, al&#233;m de cobrar uma taxa extra de 30% sobre o valor final. A empresa de seguros de viagem n&#227;o quis oferecer uma ap&#243;lice de seguro ao paciente com base em seu perfil de risco e a companhia de seguros de sa&#250;de foi &#224; justi&#231;a para rescindir o contrato, pois seu rel&#243;gio de pulso indicava um per&#237;odo de arritmias ventriculares que o paciente negara ao assinar o contrato.</p>
      <p>No entanto, na consulta m&#233;dica, o m&#233;dico constatou que o paciente havia nascido em uma &#225;rea de risco para a doen&#231;a de Chagas, um dado que havia sido omitido. Como resultado, foi realizada a sorologia Elisa, que permitiu o in&#237;cio do tratamento e atrasou a progress&#227;o da doen&#231;a.</p>
      <p>Ao ler esse trecho, que poss&#237;veis usos indevidos da IA foram identificados?</p>
      <p>Privacidade de dados, respeito &#224; autonomia, erros de entrada de dados, vi&#233;s do algoritmo de diagn&#243;stico. Voc&#234; acha isso muito dif&#237;cil de acontecer? Algumas pessoas ainda pensam que a conex&#227;o de v&#237;deo e voz &#233; uma cena dos Jetsons!</p>
    </sec>
    <sec sec-type="conclusions">
      <title>Conclus&#245;es</title>
      <p>A IA certamente traz uma poss&#237;vel revolu&#231;&#227;o na &#225;rea da sa&#250;de. No entanto, seu uso inadequado pode ser uma fonte prejudicial para os pacientes. Os preceitos &#233;ticos devem, portanto, ser o pilar norteador de qualquer implementa&#231;&#227;o dessa tecnologia. Uma quest&#227;o importante que deve sempre ser fundamental na pr&#225;tica cl&#237;nica &#233; a empatia; a capacidade de entender ou sentir o que outra pessoa est&#225; passando a partir do ponto de vista dela. Os cardiologistas precisam usar suas habilidades cl&#237;nicas, sua sabedoria, empatia e princ&#237;pios &#233;ticos para usar ferramentas de assist&#234;ncia baseadas em intelig&#234;ncia artificial no melhor interesse de seus pacientes. Nesse contexto, o reconhecimento e a identifica&#231;&#227;o de vulnerabilidades e desafios associados ao tema devem fazer parte do cotidiano das institui&#231;&#245;es de sa&#250;de.</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <title>Refer&#234;ncias</title>
      <ref id="B1">
        <label>1</label>
        <mixed-citation>1. Souza Filho EM, Fernandes FA, Soares CLA.Artificial Intelligence in Cardiology: Concepts, Tools and Challenges - &#8220;The Horse is the One Who Runs, You Must Be the Jockey&#8221;. Arq Bras Cardiol. 2020;114(4):718-25.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Souza</surname>
              <given-names>EM</given-names>
              <suffix>Filho</suffix>
            </name>
            <name>
              <surname>Fernandes</surname>
              <given-names>FA</given-names>
            </name>
            <name>
              <surname>Soares</surname>
              <given-names>CLA</given-names>
            </name>
          </person-group>
          <article-title>Artificial Intelligence in Cardiology: Concepts, Tools and Challenges - &#8220;The Horse is the One Who Runs, You Must Be the Jockey&#8221;</article-title>
          <source>Arq Bras Cardiol</source>
          <year>2020</year>
          <volume>114</volume>
          <issue>4</issue>
          <fpage>718</fpage>
          <lpage>725</lpage>
        </element-citation>
      </ref>
      <ref id="B2">
        <label>2</label>
        <mixed-citation>2. Han D, Kolli KK, Al&#8217;Aref SJ, Baskaran L, van Rosendall A, Gransar H, Machine Learning Framework to Identify Individuals at Risk of Rapid Progression of Coronary Atherosclerosis: From the PARADIGM Registry. J Am Heart Assoc. 2020 Mar 3;9(5):e013958.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Han</surname>
              <given-names>D</given-names>
            </name>
            <name>
              <surname>Kolli</surname>
              <given-names>KK</given-names>
            </name>
            <name>
              <surname>Al&#8217;Aref</surname>
              <given-names>SJ</given-names>
            </name>
            <name>
              <surname>Baskaran</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>van Rosendall</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Gransar</surname>
              <given-names>H</given-names>
            </name>
          </person-group>
          <article-title>Machine Learning Framework to Identify Individuals at Risk of Rapid Progression of Coronary Atherosclerosis: From the PARADIGM Registry</article-title>
          <source>J Am Heart Assoc</source>
          <year>2020</year>
          <month>03</month>
          <day>3</day>
          <volume>9</volume>
          <issue>5</issue>
          <elocation-id>e013958</elocation-id>
        </element-citation>
      </ref>
      <ref id="B3">
        <label>3</label>
        <mixed-citation>3. Than MP, Pickering JW, Sandoval Y, Shah A, Tsanas A, Apple FS. Machine Learning to Predict the Likelihood of Acute Myocardial Infarction. Circulation. 2019 ;140(11):899-909.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Than</surname>
              <given-names>MP</given-names>
            </name>
            <name>
              <surname>Pickering</surname>
              <given-names>JW</given-names>
            </name>
            <name>
              <surname>Sandoval</surname>
              <given-names>Y</given-names>
            </name>
            <name>
              <surname>Shah</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Tsanas</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Apple</surname>
              <given-names>FS</given-names>
            </name>
          </person-group>
          <article-title>Machine Learning to Predict the Likelihood of Acute Myocardial Infarction</article-title>
          <source>Circulation</source>
          <year>2019</year>
          <volume>140</volume>
          <issue>11</issue>
          <fpage>899</fpage>
          <lpage>909</lpage>
        </element-citation>
      </ref>
      <ref id="B4">
        <label>4</label>
        <mixed-citation>4. Hedman &#197;K, Hage C, Sharma A, Brosnan MJ, Buckbinder L, Gan LM, et al. Identification of novel pheno-groups in heart failure with preserved ejection fraction using machine learning. Heart. 2020 Mar;106(5):342-9.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Hedman</surname>
              <given-names>&#197;K</given-names>
            </name>
            <name>
              <surname>Hage</surname>
              <given-names>C</given-names>
            </name>
            <name>
              <surname>Sharma</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>Brosnan</surname>
              <given-names>MJ</given-names>
            </name>
            <name>
              <surname>Buckbinder</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>Gan</surname>
              <given-names>LM</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <article-title>Identification of novel pheno-groups in heart failure with preserved ejection fraction using machine learning</article-title>
          <source>Heart</source>
          <year>2020</year>
          <month>03</month>
          <volume>106</volume>
          <issue>5</issue>
          <fpage>342</fpage>
          <lpage>349</lpage>
        </element-citation>
      </ref>
      <ref id="B5">
        <label>5</label>
        <mixed-citation>5. Wang Y, Kosinski M. Deep neural networks are more accurate than humans at detecting sexual orientation from facial images. J Pers Soc Psychol. 2018;114(2):246-57.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Wang</surname>
              <given-names>Y</given-names>
            </name>
            <name>
              <surname>Kosinski</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
          <article-title>Deep neural networks are more accurate than humans at detecting sexual orientation from facial images</article-title>
          <source>J Pers Soc Psychol</source>
          <year>2018</year>
          <volume>114</volume>
          <issue>2</issue>
          <fpage>246</fpage>
          <lpage>257</lpage>
        </element-citation>
      </ref>
      <ref id="B6">
        <label>6</label>
        <mixed-citation>6. Musk, Wozniak and Hawking urge ban on warfare AI and autonomous weapons. 27 Jul 2015.The Guardian. [Cited in 2020 May 20] Available from: <ext-link ext-link-type="uri" xlink:href="https://www.theguardian.com/technology/2015/jul/27/musk-hawking-ban-ai-autonomous-weapons">https://www.theguardian.com/technology/2015/jul/27/musk-hawking-ban-ai-autonomous-weapons</ext-link> </mixed-citation>
        <element-citation publication-type="webpage">
          <article-title>Musk, Wozniak and Hawking urge ban on warfare AI and autonomous weapons</article-title>
          <day>27</day>
          <month>07</month>
          <year>2015</year>
          <source>The Guardian</source>
          <date-in-citation content-type="cited-date">Cited in 2020 May 20</date-in-citation>
          <comment>
            <ext-link ext-link-type="uri" xlink:href="https://www.theguardian.com/technology/2015/jul/27/musk-hawking-ban-ai-autonomous-weapons">https://www.theguardian.com/technology/2015/jul/27/musk-hawking-ban-ai-autonomous-weapons</ext-link>
          </comment>
        </element-citation>
      </ref>
      <ref id="B7">
        <label>7</label>
        <mixed-citation>7. Brundage M, Avin S, Clark J, Toner H, Eckersley P, Garfinkel B, et al.The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation. 2018. doi.org/10.17863/CAM.22520</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Brundage</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Avin</surname>
              <given-names>S</given-names>
            </name>
            <name>
              <surname>Clark</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Toner</surname>
              <given-names>H</given-names>
            </name>
            <name>
              <surname>Eckersley</surname>
              <given-names>P</given-names>
            </name>
            <name>
              <surname>Garfinkel</surname>
              <given-names>B</given-names>
            </name>
            <etal>et al</etal>
          </person-group>
          <source>The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation</source>
          <year>2018</year>
          <pub-id pub-id-type="doi">10.17863/CAM.22520</pub-id>
        </element-citation>
      </ref>
      <ref id="B8">
        <label>8</label>
        <mixed-citation>8. Brounillette, M. Deep Learning Is a Black Box, but Health Care Won&#8217;t Mind, MIT Technology Review. 2017.</mixed-citation>
        <element-citation publication-type="book">
          <person-group person-group-type="author">
            <name>
              <surname>Brounillette</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
          <source>Deep Learning Is a Black Box, but Health Care Won&#8217;t Mind, MIT Technology Review</source>
          <year>2017</year>
        </element-citation>
      </ref>
      <ref id="B9">
        <label>9</label>
        <mixed-citation>9. Miotto, R, Li L, Brian A, Kidd &amp;Joel T.Dudley. Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the EHR. Sci Rep. 2016;6:26094.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Miotto</surname>
              <given-names>R</given-names>
            </name>
            <name>
              <surname>Li</surname>
              <given-names>L</given-names>
            </name>
            <name>
              <surname>Kidd</surname>
              <given-names>Brian A</given-names>
            </name>
            <name>
              <surname>T.Dudley</surname>
              <given-names>Joel</given-names>
            </name>
          </person-group>
          <article-title>Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the EHR</article-title>
          <source>Sci Rep</source>
          <year>2016</year>
          <volume>6</volume>
          <size units="pages">26094</size>
        </element-citation>
      </ref>
      <ref id="B10">
        <label>10</label>
        <mixed-citation>10. Topol, EJ. Deep Medicine: how artificial intelligence can make healthcare human again. Philadelphia;2019.ISBN-13: 9781541644649</mixed-citation>
        <element-citation publication-type="book">
          <person-group person-group-type="author">
            <name>
              <surname>Topol</surname>
              <given-names>EJ</given-names>
            </name>
          </person-group>
          <source>Deep Medicine: how artificial intelligence can make healthcare human again</source>
          <publisher-loc>Philadelphia</publisher-loc>
          <year>2019</year>
          <comment>ISBN-13: 9781541644649</comment>
        </element-citation>
      </ref>
      <ref id="B11">
        <label>11</label>
        <mixed-citation>11. Beam AL, Manrai AK, Ghassemi M. Challenges to the Reproducibility of Machine Learning Models in Health Care. JAMA. 2020; Jan 6. doi: 10.1001/jama.2019.20866. [Epub ahead of print]</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Beam</surname>
              <given-names>AL</given-names>
            </name>
            <name>
              <surname>Manrai</surname>
              <given-names>AK</given-names>
            </name>
            <name>
              <surname>Ghassemi</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
          <article-title>Challenges to the Reproducibility of Machine Learning Models in Health Care</article-title>
          <source>JAMA</source>
          <year>2020</year>
          <month>01</month>
          <day>6</day>
          <pub-id pub-id-type="doi">10.1001/jama.2019.20866</pub-id>
          <comment>Epub ahead of print</comment>
        </element-citation>
      </ref>
      <ref id="B12">
        <label>12</label>
        <mixed-citation>12. Keskinbora KH. Medical ethics considerations on artificial intelligence. J Clin Neurosci. 2019 Jun;64:277-82.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Keskinbora</surname>
              <given-names>KH</given-names>
            </name>
          </person-group>
          <article-title>Medical ethics considerations on artificial intelligence</article-title>
          <source>J Clin Neurosci</source>
          <year>2019</year>
          <month>06</month>
          <volume>64</volume>
          <fpage>277</fpage>
          <lpage>282</lpage>
        </element-citation>
      </ref>
      <ref id="B13">
        <label>13</label>
        <mixed-citation>13. Beauchamp T, Childress J. Principles of biomedical ethics. 7th Ed. USA: Oxford University Press; 2013.</mixed-citation>
        <element-citation publication-type="book">
          <person-group person-group-type="author">
            <name>
              <surname>Beauchamp</surname>
              <given-names>T</given-names>
            </name>
            <name>
              <surname>Childress</surname>
              <given-names>J</given-names>
            </name>
          </person-group>
          <source>Principles of biomedical ethics</source>
          <edition>7th</edition>
          <publisher-loc>USA</publisher-loc>
          <publisher-name>Oxford University Press</publisher-name>
          <year>2013</year>
        </element-citation>
      </ref>
      <ref id="B14">
        <label>14</label>
        <mixed-citation>14. Parker RM, Ratzan SC, Lurie N. Health literacy: a policy challenge for advancing high-quality health care. Health Aff (Millwood) . 2003;22(4):147-53.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Parker</surname>
              <given-names>RM</given-names>
            </name>
            <name>
              <surname>Ratzan</surname>
              <given-names>SC</given-names>
            </name>
            <name>
              <surname>Lurie</surname>
              <given-names>N</given-names>
            </name>
          </person-group>
          <article-title>Health literacy: a policy challenge for advancing high-quality health care</article-title>
          <source>Health Aff</source>
          <publisher-loc>Millwood</publisher-loc>
          <year>2003</year>
          <volume>22</volume>
          <issue>4</issue>
          <fpage>147</fpage>
          <lpage>153</lpage>
        </element-citation>
      </ref>
      <ref id="B15">
        <label>15</label>
        <mixed-citation>15. The General Data Protection Regulation applies in on Member States from 25 May 2018.[Cited in 2020 jan19]. Available in <ext-link ext-link-type="uri" xlink:href="https://eur-lex.europa.eu/content/news/general-data-protection-regulation-GDPR-applies-from-25-May-2018.html">https://eur-lex.europa.eu/content/news/general-data-protection-regulation-GDPR-applies-from-25-May-2018.html</ext-link>.</mixed-citation>
        <element-citation publication-type="webpage">
          <source>The General Data Protection Regulation applies in on Member States from 25 May 2018</source>
          <date-in-citation content-type="cited-date">Cited in 2020 jan19</date-in-citation>
          <comment>
            <ext-link ext-link-type="uri" xlink:href="https://eur-lex.europa.eu/content/news/general-data-protection-regulation-GDPR-applies-from-25-May-2018.html">https://eur-lex.europa.eu/content/news/general-data-protection-regulation-GDPR-applies-from-25-May-2018.html</ext-link>
          </comment>
        </element-citation>
      </ref>
      <ref id="B16">
        <label>16</label>
        <mixed-citation>16. Rigby MJ. Ethical Dimensions of Using Artificial Intelligence in Health Care AMA Journal of Ethics February 2019, Volume 21, Number 2: E121-124</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Rigby</surname>
              <given-names>MJ</given-names>
            </name>
          </person-group>
          <article-type>Ethical Dimensions of Using Artificial Intelligence in Health Care</article-type>
          <source>AMA Journal of Ethics February</source>
          <year>2019</year>
          <volume>21</volume>
          <issue>2</issue>
          <fpage>E121</fpage>
          <lpage>E124</lpage>
        </element-citation>
      </ref>
      <ref id="B17">
        <label>17</label>
        <mixed-citation>17. Sch&#246;nberger D. Artificial intelligence in healthcare: a critical analysis of the legal and ethical implications. International J Law Inform Technol. 2019;27:171-203.</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Sch&#246;nberger</surname>
              <given-names>D</given-names>
            </name>
          </person-group>
          <article-title>Artificial intelligence in healthcare: a critical analysis of the legal and ethical implications</article-title>
          <source>International J Law Inform Technol</source>
          <year>2019</year>
          <volume>27</volume>
          <fpage>171</fpage>
          <lpage>203</lpage>
        </element-citation>
      </ref>
      <ref id="B18">
        <label>18</label>
        <mixed-citation>18. Vincent J. Twitter taught Microsoft&#8217;s AI Chatbot to be a Racist Asshole in Less than a Day. The Verge, 24 May 2016. [Cited in 2020 May 12]. Available in: <ext-link ext-link-type="uri" xlink:href="https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist">https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist</ext-link> </mixed-citation>
        <element-citation publication-type="webpage">
          <person-group person-group-type="author">
            <name>
              <surname>Vincent</surname>
              <given-names>J</given-names>
            </name>
          </person-group>
          <source>Twitter taught Microsoft&#8217;s AI Chatbot to be a Racist Asshole in Less than a Day. The Verge, 24 May 2016</source>
          <date-in-citation content-type="cited-date">Cited in 2020 May 12</date-in-citation>
          <comment>
            <ext-link ext-link-type="uri" xlink:href="https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist">https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist</ext-link>
          </comment>
        </element-citation>
      </ref>
    </ref-list>
    <fn-group>
      <fn fn-type="other">
        <p>Vincula&#231;&#227;o acad&#234;mica</p>
        <p>Este artigo &#233; parte de tese de Doutorado de Erito Marques de Souza Filho pela Universidade Federal Fluminense.</p>
      </fn>
      <fn fn-type="other">
        <p>Aprova&#231;&#227;o &#233;tica e consentimento informado</p>
        <p>Este artigo n&#227;o cont&#233;m estudos com humanos ou animais realizados por nenhum dos autores.</p>
      </fn>
    </fn-group>
    <fn-group>
      <fn fn-type="supported-by">
        <p>Fontes de financiamento</p>
        <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-categories>
        <subj-group subj-group-type="heading">
          <subject>Viewpoint</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Ethics, Artificial Intelligence and Cardiology</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <name>
            <surname>Souza</surname>
            <given-names>Erito Marques de</given-names>
            <suffix>Filho</suffix>
          </name>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff2001">
            <sup>2</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Fernandes</surname>
            <given-names>Fernando de Amorim</given-names>
          </name>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff3001">
            <sup>3</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Pereira</surname>
            <given-names>Nikolas Cunha de Assis</given-names>
          </name>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Mesquita</surname>
            <given-names>Claudio Tinoco</given-names>
          </name>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff3001">
            <sup>3</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Gismondi</surname>
            <given-names>Ronaldo Altenburg</given-names>
          </name>
          <xref ref-type="aff" rid="aff1001">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff5001">
            <sup>5</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff1001">
        <label>1</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">P&#243;s-gradua&#231;&#227;o em Ci&#234;ncias Cardiovasculares da Universidade Federal Fluminense, Niter&#243;i, RJ - Brazil</institution>
      </aff>
      <aff id="aff2001">
        <label>2</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">Universidade Federal Rural do Rio de Janeiro - Departamento de Tecnologias e Linguagens, Nova Igua&#231;u, RJ &#8211; Brazil</institution>
      </aff>
      <aff id="aff3001">
        <label>3</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">Setor de Medicina Nuclear do Servi&#231;o de Radiologia do Hospital Universit&#225;rio Ant&#244;nio Pedro (EBESERH-HUAP-UFF), Niter&#243;i, RJ - Brazil</institution>
      </aff>
      <aff id="aff4001">
        <label>4</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">Faculdade de Medicina da Universidade Federal Fluminense, Niter&#243;i, RJ - Brazil</institution>
      </aff>
      <aff id="aff5001">
        <label>5</label>
        <country country="BR">Brazil</country>
        <institution content-type="original">Hospital Niter&#243;i D&#180;or, Niter&#243;i, RJ &#8211; Brazil</institution>
      </aff>
      <author-notes>
        <corresp id="c01001"><bold>Mailing Address: Erito Marques de Marques</bold> &#8226; Universidade Federal Fluminense - Departamento de Medicina Cl&#237;nica &#8211; Av. Marques do Paran&#225;, 303. Postal Code 24033-900, Niter&#243;i, RJ &#8211; Brazil. E-mail: mederitomarques@gmail.com </corresp>
        <fn fn-type="con">
          <p>Author contributions</p>
          <p>Conception and design of the research and Writing of the manuscript: Souza Filho EM, Fernandes FA, Pereira NCA, Mesquita CT, Gismondi RA; Acquisition of data: Souza Filho EM, Pereira NCA, Mesquita CT, Gismondi RA; Analysis and interpretation of the data: Souza Filho EM, Fernandes FA, Mesquita CT, Gismondi RA; Critical revision of the manuscript for intellectual content: Souza Filho EM, Mesquita CT, Gismondi RA.</p>
          <p>Potential Conflict of Interest</p>
        </fn>
        <fn fn-type="conflict">
          <p>No potential conflict of interest relevant to this article was reported.</p>
        </fn>
      </author-notes>
      <kwd-group xml:lang="en">
        <kwd>Artificial Intelligence</kwd>
        <kwd>Software</kwd>
        <kwd>Decision Making,Compter-Assisted</kwd>
        <kwd>Information Management</kwd>
      </kwd-group>
    </front-stub>
    <body>
      <disp-quote>
        <p>&#8220;By far, the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.&#8221;</p>
        <p>Eliezer Yudkowsky</p>
      </disp-quote>
      <sec sec-type="intro">
        <title>Introduction</title>
        <p>In the not too distant future, an artificially intelligent computer program will probably diagnose heart conditions more accurately than a board-certified cardiologist. Biomedical knowledge grows significantly, making it impossible for contemporary health professionals to be updated on all content published in their field. Similarly, the amount of information about the patient is increasingly larger and more accessible, making real-time management, filtering and selection impractical for an individual. In this context, Artificial Intelligence (AI) has a relevant role in health decision-making. It is the result of the combination of sophisticated mathematical models and computation to produce refined algorithms capable of emulating (or imitating) human intelligence.<sup><xref ref-type="bibr" rid="B1">1</xref></sup> It has allowed for interesting applications in virtually all fields of medicine and human knowledge. In particular, in cardiology, several applications have been shown to be successful. Han et al.,<sup><xref ref-type="bibr" rid="B2">2</xref></sup> for instance, used machine learning (ML), a subset of AI, to analyze if this tool would be helpful to identify patients at risk of future rapid coronary plaque progression. Clinical epidemiological features and quantitative and qualitative information from coronary computed tomographic angiography were used (all of which were obtained from the PARADIGM study). They included 1,083 patients in the study and tested 10 different models. LogitBoost performed better. The area under the receiver operating characteristic curve (AUC) was 0.83 &#8212; better than the 10-year atherosclerotic cardiovascular disease risk score (ASCVD risk score), which was 0.59. In another study, Than et al.<sup><xref ref-type="bibr" rid="B3">3</xref></sup> evaluated whether Gradient Boosting (also a ML algorithm) would be beneficial in predicting the likelihood of type 1 acute myocardial infarction. Features such as sex, age, rate of change of cardiac troponin I concentration and paired cardiac troponin I of a sample with 11,011 patients were considered. AUC was 0.96 and the ML model had better performance than the traditional European Society of Cardiology 0/3-hour pathway. Hedman et al.<sup><xref ref-type="bibr" rid="B4">4</xref></sup> developed a ML algorithm to describe heart failure with preserved ejection fraction groups of patients based on their phenotype. They used clinical and electrocardiogram data. Six different groups were identified, with different levels of inflammatory and cardiovascular proteins and also with different outcomes. In light of that, in cardiology, the process of incorporating AI into the clinical practice is accelerated. The use of AI in cardiology is present in our daily lives, such as recognition of disease phenotypes, diagnosis, prognosis, and in treatment algorithms. AI has a huge disruptive potential and some advocate the possibility of the emergence of a new species, Homo incredibile,<sup><xref ref-type="bibr" rid="B1">1</xref></sup> which supports its decisions on data and promotes a revolution in the digital ecosystem. However, this paradigm shift has unfortunately brought with it a myriad of challenges. Ethical issues are a great concern regarding these new technologies, and we will discuss some of them, as well as possible solutions and precautions.</p>
        <sec>
          <title>Ethical Concerns</title>
          <sec>
            <title>Discrimination and Data Privacy</title>
            <p>These algorithms can, for example, be used to discriminate against people, to give life to devices that put other lives at risk or even to produce and disseminate fake news &#8212; not to mention the potential damage in case of inadequate information security policies.<sup><xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref></sup>The hijacking of files that took place in 2017, with more than 300 million computers affected by the WannaCry ransomware in 150 countries and the data leak by the company Ashley Madison in 2015, are examples of the destructive potential of hacker actions. This exemplifies some obstacles to be overcome on the inclusion of wearable medical devices in clinical practice and on the use of autonomous systems to support decision-making in healthcare.</p>
            <p>Obtaining informed consent is a concern of most bioethicists. Current models of AI are very dependent on information from medical records. Is it possible to ensure that personal information remains confidential even with data traveling over the internet? Leakage of medical information of famous people such as former First Lady Marisa Silva and the current President are just a few examples of problems with data confidentiality.</p>
          </sec>
          <sec>
            <title>Transparency and Safety</title>
            <p>In addition, when it comes to the medical and health sciences, specifically, other risks stand out. One of them is the lack of transparency in decision-making or the inability to explain the &#8220;reasoning&#8221; to obtain the final result, represented by the so-called black boxes. Research has been carried out in search of solutions. However, the current reality is that those responsible for most of the great things done with Deep Learning do not know how to fully explain the functioning of their successful systems.<sup><xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref></sup> On the other hand, it is not always possible to provide detailed explanations about the pathophysiology of certain diseases or about the mechanism of action of some drugs, even though clinical trials have shown benefits for the patient. This adds to the challenge of ensuring the reproducibility and replicability of AI algorithms. As pointed out by Beam et al., a study is reproducible if, based on data access and analysis of the algorithm code, an independent group can obtain the same results observed by the original study while replicability is associated with the fact that an independent group can study the same phenomenon and obtain the same conclusions after carrying out a set of experiments or analyses from a new set of data.<sup><xref ref-type="bibr" rid="B11">11</xref></sup> Other relevant questions are how safe the patients&#8217; data are and how aware the patients are about the use of their data. A partnership between the British National Health System (NHS) and a subsidiary of a big private tech company in 2015, which included the unconsented transfer of an identifiable database of more than 1.6 million inhabitants, was one of the most famous and controversial cases to date. Despite the good intentions of both sides, it was clear how much we can be exposed if we do not discuss, right now, the extent to which the data is owned by an individual. Moreover, with this type of agreement, large technology companies tend to further increase the existing oligopoly.</p>
          </sec>
          <sec>
            <title>Patient Values and Preferences, Clinical Judgement and Empathy</title>
            <p>Human contact between doctors and patients is one of the foundations of medicine since Hippocrates. Doubts exist as to whether an AI is able to take into account the person&#8217;s social context, environmental factors, preferences and moral values in the treatment decision algorithm.</p>
            <p>Another important aspect is the representation of ethnic, social and cultural minorities in the medical records that serve as the basis for the AI algorithm. If these data are not very representative or skewed, errors of interpretation may occur.</p>
          </sec>
          <sec>
            <title>Measures to be Implemented</title>
            <p>As defined by Keskinbora,<sup><xref ref-type="bibr" rid="B12">12</xref></sup> to trust AI we need the following:</p>
            <list list-type="bullet">
              <list-item>
                <p>Transparency of data, operation and algorithms</p>
              </list-item>
              <list-item>
                <p>Credibility and auditability, including the report of bias and errors</p>
              </list-item>
              <list-item>
                <p>Reliability, with clinically validated AI</p>
              </list-item>
              <list-item>
                <p>Recoverability, allowing manual control of the operation if needed</p>
              </list-item>
            </list>
            <p>This scenario brings with it the need for a discussion on the use of AI in health and its limits considering the fundamental principles of bioethics in health: justice, non-maleficence, beneficence, fairness, equality, social acceptance and respect for patient autonomy.<sup><xref ref-type="bibr" rid="B13">13</xref></sup> The question that emerges in this context is: how to incorporate AI into biomedical practice while respecting these principles in order to generate value? Although there is no definitive answer to the question, a promising strategy (<xref ref-type="fig" rid="f01001">figure 1</xref>) includes:</p>
            <p>
              <fig id="f01001">
                <label>Figure 1</label>
                <caption>
                  <title>&#8211; Proposed strategy for Artificial Intelligence implementation in clinical practice considering ethics. AI: artificial intelligence.</title>
                </caption>
                <graphic xlink:href="1678-4170-abc-115-03-0579-00143-gf01-en.tif"/></fig>
            </p>
            <list list-type="alpha-lower">
              <list-item>
                <p><bold>Cooperation</bold>: AI models tend to perform better when we have healthy data about what we want to study. Thus, interinstitutional collaboration has a fundamental role in this process, as the sharing of this data favors the achievement of metrics of excellence.<sup><xref ref-type="bibr" rid="B1">1</xref></sup></p>
              </list-item>
              <list-item>
                <p><bold>Health Literacy</bold>: this relates to the level of health information that each individual is able to obtain, manage and understand to apply in the decision-making process of health.<sup><xref ref-type="bibr" rid="B14">14</xref></sup> Individuals with greater literacy tend to make better health decisions. Thus, as AI models are incorporated into clinical practice, it is essential that literacy about them also be expanded. This includes an expanded doctor-patient relationship, which is concerned with including the patient at the center of multifactorial and multi-professional decision-making. Similarly, health professional literacy in AI should be encouraged.</p>
              </list-item>
              <list-item>
                <p><bold>Security and privacy</bold>: encrypted data is just the first step in more general measures to ensure data privacy. The Cambridge Analytica scandal was a major warning about the potential harms due to the misuse of BIG DATA. In this context, strict compliance with GDPR<sup><xref ref-type="bibr" rid="B15">15</xref></sup> should be seen as a fundamental right of any human being for which no efforts should be made to ensure. This is one of the most central issues in the ethical reference for AI implementation and needs to be well secured. Medical electronic records are the most valuable data clinicians and cardiologists are managing today. Another important issue is the protection of patient photographic images as they apply to facial recognition technology, which could threaten proper informed consent and safety of patients.<sup><xref ref-type="bibr" rid="B16">16</xref></sup></p>
              </list-item>
              <list-item>
                <p><bold>Purpose</bold>: AI should be used as a tool whose objective is to promote quality of life, health and well-being of human beings. Letting economic interests outweigh real human needs is a serious mistake that can have disastrous consequences. Keskinbora<sup><xref ref-type="bibr" rid="B12">12</xref></sup> suggests, for example, the development of a free AI created with a common objective and whose feeling is based on the operation according to ethical values.</p>
              </list-item>
              <list-item>
                <p><bold>Time optimization</bold>: the modus operandi of the workforce brings with it jobs that require a large number of hours from human beings in their work, performing repetitive tasks. Many of these tasks can be replaced by machines that perform better or like humans. This creates a window of opportunity in which human beings have the potential to work less, in more specialized jobs. The extra time they have should be invested in further studies, leisure activities, physical activity, family care, etc. This certainly implies a new reformulation of labor and educational policies.</p>
              </list-item>
              <list-item>
                <p><bold>Audit of errors and public surveillance</bold>: AI models can make mistakes and their decisions are not always understandable to humans. Therefore, it is essential that the algorithms be audited periodically and that their performance metrics be informed to patients before making any decisions regarding their health. There is an important challenge here, related to the development of a specific legal apparatus on the subject.</p>
              </list-item>
              <list-item>
                <p><bold>Education</bold>: throughout its life, different sets of human knowledge become useless. In a volatile, unstable, complex and ambiguous world, disruptive technologies can render obsolete previous skills and knowledge over time: it is the half-life of biomedical knowledge. The solution is the process of continuous study, where human beings study forever! Another point refers to what to study; certainly a model centered on memory should be replaced by a model focused on solving real problems in society. To this end, expanding the study of mathematics, computing and basic sciences in graduate and postgraduate courses in health with this objective is mandatory.<sup><xref ref-type="bibr" rid="B1">1</xref></sup></p>
              </list-item>
              <list-item>
                <p><bold>Attention to biases</bold>: Machine Learning models (a subset of AI) produce their responses according to the data that are used as inputs for the algorithms. Thus, it is possible to generate discriminatory behavior in relation to certain groups. For instance, if there is not enough data to be used in algorithm training. An example of this prejudiced bias was Microsoft Tay&#8217;s chatbot, which learned racist and sexist language and needed to be removed on the day of its launch.<sup><xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref></sup></p>
              </list-item>
            </list>
          </sec>
        </sec>
      </sec>
      <sec>
        <title>Case Example</title>
        <p>Consider the following hypothetical example. A 70-year-old man has a heart failure condition: dyspnea on exertion, orthopnea, crackles in the lung bases and edema of the lower limbs. Concerned about his situation, he made a quick visit to an online medical diagnosis website, which showed 99% probability of heart failure. In order to save time, he performed an echocardiogram on his own. An AI diagnostic algorithm showed 97% chances of having idiopathic dilated cardiomyopathy, with a prognosis of 12-month survival of only 13% and contraindication to transplantation. And that was written in the automatic report generated by a computer with AI.</p>
        <p>Upset about the situation, he sold all of his belongings and booked a trip across five continents for the following month, but the airline demanded a medical certificate, claiming that there was an 80% chance of on-board complications, in addition to charging an extra fee of 30% on the final amount. The travel insurance firm did not want to offer an insurance policy to the patient based on his risk profile and the health insurance company went to court to break the contract, as his wristwatch had indicated a period of ventricular arrhythmias that the patient denied when he signed the contract.</p>
        <p>However, at the medical appointment, the doctor found that the patient was born in a risk area for Chagas&#8217; disease, which omitted. As a result, Elisa serology was conducted, which allowed the initiation of treatment and delayed the progression of the disease.</p>
        <p>By reading this excerpt, what possible misuses of AI were identified?</p>
        <p>Data privacy, respect for autonomy, data input errors, diagnostic algorithm bias. Do you find this very difficult to happen? Some people still think that the video and voice connection is a scene from the Jetsons!</p>
      </sec>
      <sec sec-type="conclusions">
        <title>Conclusions</title>
        <p>AI certainly brings a potential revolution in healthcare. However, its improper use can be a harmful source for patients. Ethical precepts must, therefore, be the guiding pillar of any implementation of this technology. An important issue that must be always key in clinical practice is empathy; the capacity to understand or feel what another person is experiencing from their point of view. Cardiologists need to use their clinical skills, wisdom, empathy and ethical principles to use artificial intelligence-based assistance tools in the best interest of their patients. In this context, the recognition and identification of vulnerabilities and challenges associated with the theme must be part of the routine of health institutions.</p>
      </sec>
    </body>
    <back>
      <fn-group>
        <fn fn-type="other">
          <p>Study Association</p>
          <p>This article is part of the thesis of Doctoral submitted by Erito Marques de Souza Filho, from Universidade Federal Fluminense.</p>
        </fn>
        <fn fn-type="other">
          <p>Ethics approval and consent to participate</p>
          <p>This article does not contain any studies with human participants or animals performed by any of the authors.</p>
        </fn>
      </fn-group>
      <fn-group>
        <fn fn-type="supported-by">
          <p>Sources of Funding</p>
          <p>There were no external funding sources for this study.</p>
        </fn>
      </fn-group>
    </back>
  </sub-article>
</article>
