{"paper_id":"68af7cb8-a00b-4b25-bae2-77cb1d461d15","body_text":"Vol.:(0123456789)\nJournal of Medical Imaging and Interventional Radiology           (2025) 12:15  \nhttps://doi.org/10.1007/s44326-025-00061-2\nREVIEW ARTICLE\nEmpowering women through intelligent care: a narrative review \nof AI‑driven digital innovations for endometriosis diagnosis, \neducation, and equity\nKelnner Portela Luz1,2  · Danilo Lopes Ferreira Lima1\nReceived: 28 April 2025 / Accepted: 18 June 2025 \n© The Author(s) 2025\nAbstract\nBackground Endometriosis is a chronic, inflammatory, and multifactorial gynecological disorder affecting approximately \n10% of women of reproductive age worldwide. It is associated with debilitating pelvic pain, infertility, and significant \nsocioeconomic burden. Despite its impact, diagnosis is often delayed due to nonspecific symptoms and the absence of non-\ninvasive biomarkers.\nObjective This narrative review critically examines the current landscape of artificial intelligence (AI) applications in endo-\nmetriosis diagnosis, education, and management, identifies existing barriers to clinical integration, and proposes a strategic \nframework for the development of inclusive and ethical digital health ecosystems.\nMethods A narrative synthesis of peer-reviewed literature and technical reports was conducted, focusing on AI-enabled \ntools in reproductive health, endometriosis-specific applications, ethical guidelines for AI in healthcare, and regulatory \nframeworks. Comparative analysis of current FemTech solutions was also included.\nResults Despite promising developments in AI-based symptom-tracking, imaging analysis, and decision support, significant \nbarriers persist. These include technical limitations (small and biased datasets), clinical misalignment, ethical concerns (pri-\nvacy risks, bias amplification), and sociocultural challenges (digital divide, stigma). Current FemTech platforms demonstrate \nlimited readiness to address the complex needs of endometriosis patients.\nConclusions To fully realize AI’s transformative potential in endometriosis care, future efforts must prioritize participatory \ndesign, real-world data integration, transparency, inclusivity, and regulatory compliance. Endometriosis must be elevated \nwithin digital health equity agendas to ensure that technological innovations effectively address the lived experiences of \nwomen affected by this condition.\nKeywords Endometriosis · Artificial Intelligence · Digital Health · Health Equity\nIntroduction\nEndometriosis is a chronic, inflammatory, estrogen-depend-\nent disease characterized by the ectopic presence of endo-\nmetrial-like tissue outside the uterine cavity, predominantly \naffecting pelvic organs. It impacts approximately 10% of \nwomen of reproductive age worldwide, leading to chronic \npelvic pain, dysmenorrhea, dyspareunia, gastrointestinal dis-\nturbances, fatigue, and infertility [1, 2].\nBeyond its physical manifestations, endometriosis exerts \nprofound psychological, emotional, and socioeconomic bur-\ndens. Studies have demonstrated a heightened prevalence of \nanxiety, depression, social isolation, and diminished work -\nforce participation among affected women [3 –5]. From an \neconomic standpoint, the global costs associated with endo-\nmetriosis, including healthcare expenditures and productiv-\nity loss, are estimated to exceed billions of dollars annually \n[2].\nDespite its significant prevalence and impact, endome-\ntriosis remains poorly understood, frequently underdiag-\nnosed, and underprioritized in healthcare systems. Diag-\nnostic delays, averaging between seven and twelve years \n * Kelnner Portela Luz \n kelnner@endorainstitute.org\n1 Health Education and Educational Technologies, Centro \nUniversitário Christus (Unichristus), Fortaleza, Ceará, Brazil\n2 Hospital Geral de Fortaleza (HGF), Fortaleza, Ceará, Brazil\n\n Journal of Medical Imaging and Interventional Radiology           (2025) 12:15 \n   15  Page 2 of 7\nfrom symptom onset, are commonly attributed to nonspe -\ncific clinical presentations, the normalization of menstrual \npain, fragmented care pathways, and the lack of non-invasive \ndiagnostic biomarkers [5].\nArtificial intelligence (AI), encompassing machine learn-\ning, deep learning, and natural language processing (NLP), \nhas emerged as a transformative force within healthcare. AI \ntechnologies demonstrate the ability to recognize complex \npatterns, predict clinical outcomes, and personalize care \nstrategies. In the context of endometriosis, AI holds particu-\nlar promise for revolutionizing early detection, enhancing \npatient education, optimizing symptom management, and \nfostering patient-centered care pathways [6–8].\nHowever, the clinical translation of AI in endometriosis \nremains at an early stage, confronting numerous technical, \nethical, and sociocultural challenges. This review aims to \ncritically evaluate the landscape of AI-driven innovations \nrelevant to endometriosis care, analyze barriers to imple-\nmentation, and propose a strategic framework grounded in \nethical governance, participatory design, and real-world \nvalidation to guide future developments.\nAI in reproductive and gynecological health\nArtificial intelligence (AI) technologies have increasingly \npermeated reproductive medicine and gynecology, offering \ninnovative solutions for diagnosis, treatment optimization, \nand patient engagement.\nIn obstetrics, AI-driven risk stratification models have \nbeen implemented to predict adverse outcomes, such as \npreeclampsia, preterm birth, and gestational diabetes, ena-\nbling earlier and more targeted interventions [7]. In gyneco-\nlogic oncology, machine learning algorithms have enhanced \nthe accuracy of early detection for ovarian, endometrial, and \ncervical cancers by interpreting imaging studies and bio-\nmarker patterns with higher sensitivity than conventional \nmethods [9].\nDeep learning, particularly through convolutional neural \nnetworks (CNNs), has substantially improved diagnostic \nperformance in imaging modalities. Applications include \nthe automated identification of ovarian cysts, uterine abnor-\nmalities, and pelvic adhesions on ultrasound and magnetic \nresonance imaging (MRI) scans [10– 13]. In reproductive \nendocrinology, AI algorithms have been utilized to predict \noutcomes in in vitro fertilization (IVF) cycles, improving \nembryo selection and success rates.\nNatural language processing (NLP) techniques have \nfacilitated the extraction of patient-reported symptoms from \nunstructured clinical narratives, enhancing the detection of \nunderdocumented conditions, such as chronic pelvic pain \nand menstrual disorders [14].\nMobile health applications powered by machine learning \nhave empowered users to track menstrual cycles, monitor \nfertility windows, and log gynecological symptoms. How -\never, despite the proliferation of FemTech solutions, few \nplatforms are specifically designed to address the complex \nand heterogeneous symptomatology of endometriosis. Most \ntools lack condition-specific algorithms, real-world valida-\ntion, or ethical design frameworks that prioritize the needs \nof women living with chronic gynecological conditions [15, \n16].\nThus, although AI has demonstrated considerable prom-\nise across reproductive health domains, its targeted appli-\ncation to endometriosis remains limited, highlighting a \ncritical opportunity for innovation tailored to this complex \ncondition.\nBarriers to implementation in endometriosis \ncare\nDespite the promising potential of artificial intelligence (AI) \napplications in endometriosis care, several interrelated bar -\nriers currently impede their clinical translation and large-\nscale deployment. These barriers can be categorized into \ntechnical, clinical, ethical and regulatory, and sociocultural \ndomains.\nTechnical barriers\nThe development of robust AI models depends heavily on \nthe availability of large, diverse, and well-annotated datasets. \nHowever, in endometriosis research, datasets are often:\n• Small and fragmented, typically derived from single \ninstitutions or specific demographic groups;\n• Heterogeneous in diagnostic criteria and outcome meas-\nures, hindering standardization and model interoperabil-\nity;\n• Lacking longitudinal data, which limits predictive mod-\neling for disease progression and treatment outcomes.\nMoreover, interoperability challenges among electronic \nhealth record (EHR) systems restrict the integration of mul-\ntimodal datasets encompassing clinical, imaging, symptom-\ntracking, and genomic information, all of which are essential \nfor dynamic AI-driven personalization [6, 17].\nModel generalizability remains a critical concern, with \nmost algorithms trained and validated on homogeneous \npopulations. This raises significant issues regarding perfor-\nmance disparities across different racial, ethnic, and socio-\neconomic groups.\n\nJournal of Medical Imaging and Interventional Radiology           (2025) 12:15  \n Page 3 of 7    15 \nClinical barriers\nSuccessful clinical adoption of AI technologies requires \nseamless integration into existing healthcare workflows \nand alignment with clinical decision-making processes. \nKey barriers include:\n• Disruption of traditional workflows: AI solutions often \nlack alignment with the realities of clinical practice, \nresulting in poor usability and clinician resistance [18];\n• Skepticism and trust deficits among clinicians: Black \nbox algorithms with limited explainability erode clini-\ncian confidence in AI-generated outputs [19];\n• Scarcity of prospective validation studies: Few AI \napplications in gynecology, particularly for endome-\ntriosis, have undergone rigorous prospective clinical \ntrials, hindering evidence-based adoption into guide-\nlines [20].\nAdditionally, the multidisciplinary nature of endome-\ntriosis management, encompassing gynecology, radiology, \npain medicine, mental health, and physiotherapy, neces -\nsitates AI systems capable of integrating and coordinating \nacross specialties.\nEthical and regulatory barriers\nEthical challenges are paramount in deploying AI solu-\ntions for sensitive health domains such as endometriosis. \nMajor concerns include:\n• Privacy and data security risks: Endometriosis-related \ndata often involve intimate details about reproductive \nand sexual health, necessitating stringent privacy pro-\ntections [4 , 21];\n• Algorithmic bias and inequities: AI models trained on \nnon-representative datasets risk perpetuating or exac-\nerbating existing healthcare disparities [22];\n• Opacity and accountability gaps: Without transparent \ndecision-making and clear accountability mechanisms, \npatients may face difficulty contesting erroneous AI-\ndriven recommendations.\nRegulatory frameworks, including the United States \nFood and Drug Administration (FDA) guidance on Soft-\nware as a Medical Device (SaMD), Brazil’s Lei Geral de \nProteção de Dados (LGPD), and the European Union’s \nGeneral Data Protection Regulation (GDPR), mandate \ntransparency, consent, and post-market surveillance. How -\never, specific regulatory pathways tailored to AI applica-\ntions in gynecology remain underdeveloped [23– 25].\nSociocultural and educational barriers\nSociocultural dynamics profoundly influence the adoption \nand effectiveness of digital health innovations. In the context \nof endometriosis:\n• The digital divide remains a significant barrier, with \nwomen in low-resource settings having limited access to \nAI-driven healthcare tools [26];\n• Health literacy gaps impact the ability to interpret AI \noutputs and engage with digital health technologies;\n• Cultural stigmas surrounding menstruation and chronic \npelvic pain contribute to diagnostic delays and may \ninhibit patient engagement with symptom-tracking tech-\nnologies [27].\nFurthermore, many AI applications are developed without \nadequate participatory input from diverse patient popula-\ntions, resulting in culturally insensitive designs, inaccessible \ninterfaces, and unmet user needs [16].\nComparative analysis: FemTech and AI \nframeworks\nThe rapid expansion of FemTech, technologies focused on \nwomen's health, has generated numerous mobile applica -\ntions and digital platforms addressing menstruation, fertility, \nand pregnancy management. However, a critical evaluation \nreveals that relatively few FemTech solutions specifically \ntarget the complex and heterogeneous needs of women with \nendometriosis.\nTable  1 presents a comparative overview of selected \nFemTech platforms with relevance to gynecological health, \nevaluating their primary focus, integration of artificial intel-\nligence, specificity to endometriosis, regulatory validation, \nand ethical safeguards.\nKey observations include:\n• Most mainstream FemTech apps (e.g., Flo Health, \nSympto) focus primarily on general menstrual or fertil-\nity tracking and are not specifically tailored to the diverse \nsymptom profiles of endometriosis;\n• Few initiatives, such as EndoMind (hypothetical), \ndirectly target endometriosis symptom triage through \nAI-driven models;\n• Regulatory readiness varies significantly, with few plat-\nforms achieving the certifications necessary for clinical-\ngrade deployment, such as FDA clearance or CE mark -\ning;\n• Ethical, privacy, and data protection standards are incon-\nsistently applied across platforms, raising concerns about \nuser data vulnerabilities.\n\n Journal of Medical Imaging and Interventional Radiology           (2025) 12:15 \n   15  Page 4 of 7\nThis analysis highlights the urgent need for dedicated, \nethically designed, and clinically validated AI platforms spe-\ncifically addressing the fluctuating and multifactorial nature \nof endometriosis.\nToward a strategic framework for AI \nin endometriosis\nTo fully harness the transformative potential of Artificial \nIntelligence (AI) in endometriosis care, it is imperative to \ntransition from isolated algorithmic solutions to integrated, \nadaptive, and ethically governed digital health ecosystems. \nBased on the current literature and critical gaps identified, \nwe propose a strategic framework composed of five core \npillars:\nParticipatory co‑design\nAI solutions must be co-developed with active engagement \nof women living with endometriosis, multidisciplinary clini-\ncians, ethicists, and advocacy groups. Participatory design \nensures that innovations are:\n• Aligned with real-world patient needs and lived experi-\nences;\n• Culturally, socially, and economically sensitive;\n• Usable, accessible, and empowering across diverse \nuser populations. Continuous involvement of end-users \nthroughout design, testing, and implementation phases is \nessential to foster relevance, usability, and trust.\nReal‑world data integration\nEffective AI solutions require longitudinal, multimodal data-\nsets encompassing:\n• Clinical records (diagnoses, imaging, treatments);\n• Patient-reported outcomes (e.g., pain diaries, quality of \nlife metrics);\n• Data from symptom-tracking apps and wearable devices. \nIntegrating structured and unstructured real-world data \nenables dynamic personalization, predictive modeling, \nand a holistic understanding of disease trajectories.\nPersonalized educational modules\nAI-powered educational platforms should adapt content \ndelivery according to:\n• Literacy level;\n• Cultural background;\n• Health knowledge;\n• Emotional readiness of the user. Through behavioral \nanalytics and recommendation systems, content can be \ndynamically tailored to optimize patient comprehension, \nself-efficacy, and therapeutic adherence.\nVirtual psychosocial support\nGiven the significant psychosocial burden of endometriosis, \nAI solutions must embed mental health support, including:\n• Conversational agents delivering empathetic responses \nand evidence-based coping strategies;\n• Mindfulness exercises and cognitive–behavioral therapy \n(CBT) modules;\n• Peer support facilitation within safe, moderated digital \ncommunities. Such features can enhance psychologi-\ncal resilience, reduce isolation, and complement formal \nhealthcare services.\nEthical AI design\nEthical considerations must be embedded throughout the AI \nlifecycle, adhering to the frameworks outlined by:\nTable 1  Comparative analysis of FemTech platforms relevant to endometriosis care\n*Hypothetical model based on AI-driven symptom-tracking principles\nPlatform/App Primary focus AI integration Specificity to endome-\ntriosis\nClinical/regulatory \nvalidation\nEthical and privacy \npractices\nFlo health Menstrual cycle/fertility Yes (machine learning) No No FDA/CE certifica-\ntion\nStandard policy; potential \nsensitive data risks\nClue Menstrual health Yes (predictive algo-\nrithms)\nPartial (menstrual \nsymptoms)\nNo FDA/CE certifica-\ntion\nGood practices; lacks \nindependent audits\nSympto Natural contraception No No No clinical validation Unclear\nEndomind* Endometriosis symp-\ntom triage\nYes (AI-based triage \nmodel)\nYes (dedicated focus) Conceptual proposal Ethically designed from \ninception\nNatural cycles Fertility/contraception Yes (probabilistic \nmodels)\nNo FDA approved Rigorous privacy policy\n\nJournal of Medical Imaging and Interventional Radiology           (2025) 12:15  \n Page 5 of 7    15 \n• World Health Organization (WHO);\n• United Nations Educational, Scientific and Cultural \nOrganization (UNESCO);\n• European High-Level Expert Group on Artificial Intel-\nligence (EU HLEG) [4 , 5, 21]. Core ethical principles \ninclude:\n• Transparency and explainability of AI outputs;\n• Ongoing bias detection and mitigation;\n• Informed consent and robust data protection;\n• Clear accountability mechanisms for adverse outcomes.\nBuilding ethical AI is essential not only for compliance \nbut also for establishing societal trust and maximizing long-\nterm impact.\nExpanded discussion and implications\nWhile Artificial Intelligence (AI) offers unprecedented \nopportunities to improve endometriosis care, its develop -\nment, deployment, and evaluation must be critically exam-\nined through ethical, regulatory, and sociocultural lenses to \nensure that innovations promote health equity rather than \nexacerbate existing disparities.\nEthical priorities for AI in endometriosis\nEthical deployment of AI in sensitive health domains like \nendometriosis demands strict adherence to principles of \ntransparency, justice, beneficence, and respect for patient \nautonomy. Key ethical imperatives include:\n• Transparency and Explainability: AI models must pro-\nduce outputs interpretable by both clinicians and patients. \nBlack box algorithms compromise trust and impede \ninformed decision-making;\n• Bias Mitigation: AI systems must undergo regular audit-\ning to detect and mitigate biases related to race, ethnicity, \nsocioeconomic status, and geography. Without deliber -\nate correction, healthcare disparities may be perpetuated \n[22];\n• Informed Consent and Data Governance : Patients \nmust maintain control over their personal data, supported \nby explicit, clear, and understandable consent processes \nthat specify how data are used, stored, and shared;\n• Accountability Mechanisms: Clear pathways must \nexist to assign responsibility for AI-driven decisions and \npotential adverse outcomes. Oversight structures, exter -\nnal audits, and real-time monitoring can enhance system \naccountability.\nInternational frameworks, such as the WHO Guidance \non Ethics and Governance of AI for Health [5], UNESCO’s \nRecommendations on the Ethics of AI [4], and the EU HLEG \nEthics Guidelines for Trustworthy AI [21], provide essential \nfoundations that must guide all AI innovations in endome -\ntriosis care.\nBridging the Digital Divide and Future Directions\nIf not deliberately addressed, digital health innovations risk \namplifying existing inequities, particularly among underserved \npopulations.\nKey strategies to bridge the digital divide include:\n• Infrastructure Development: Design AI tools that are \nmobile-compatible, offline-capable, and functional in low-\nbandwidth environments to ensure broad accessibility;\n• Cultural and Linguistic Adaptation: Develop multi-\nlingual, culturally sensitive platforms that resonate with \ndiverse populations. Participatory design involving target \ncommunities can surface critical contextual factors;\n• Enhancing Digital Health Literacy: Implement educa-\ntional initiatives aimed at empowering users to engage \nmeaningfully with AI-driven health tools and interpret \noutputs accurately;\n• Community Engagement and Trust Building: Foster \ncollaborations with community leaders, advocacy organi-\nzations, and patient groups to enhance credibility, trust, \nand uptake.\nFuture research directions\n• Hybrid Effectiveness-Implementation Studies: Evalu-\nate both clinical efficacy and real-world feasibility of AI \ninterventions across diverse healthcare settings;\n• Dynamic Consent Models : Innovate patient consent \nprocesses to accommodate evolving AI applications and \nsecondary uses of health data;\n• Global Health Equity Frameworks: Position endome-\ntriosis within broader global digital health equity initia-\ntives, ensuring that innovations do not reinforce struc-\ntural exclusions.\nBy proactively addressing these ethical and structural \nchallenges, AI can transition from a promising technol-\nogy to a catalyst for meaningful, sustainable, and equitable \nimprovements in endometriosis care.\nKey recommendations for implementation\nTo ensure that Artificial Intelligence (AI) fulfills its trans-\nformative potential in endometriosis care while upholding \nethical principles and advancing health equity, we propose \nthe following strategic recommendations:\n\n Journal of Medical Imaging and Interventional Radiology           (2025) 12:15 \n   15  Page 6 of 7\n1. Foster Participatory Development\n• Actively engage women living with endometriosis, \nmultidisciplinary clinicians, developers, and advocacy \ngroups throughout all stages of AI development;\n• Maintain continuous feedback loops to refine AI tools \nbased on real-world user experiences, ensuring rel-\nevance, accessibility, and trust.\n2. Mandate Algorithm Auditing and Explainability\n• Require routine bias audits of AI models, particularly \nregarding training datasets and predictive outputs;\n• Prioritize the development of interpretable models \n(e.g., attention-based networks, explainable decision \ntrees) that allow clinicians and patients to understand \nand question AI recommendations.\n3. Ensure Real-World Validation\n• Conduct prospective, multicenter clinical trials to eval-\nuate AI system performance under real-world condi-\ntions;\n• Integrate implementation science frameworks to assess \nusability, adoption rates, sustainability, and equity \nimpacts alongside traditional clinical outcomes.\n4. Strengthen Regulatory Oversight\n• Advocate for AI-specific extensions to existing regu-\nlatory frameworks (e.g., FDA, EMA, ANVISA) to \naddress dynamic algorithm updates, transparency \nrequirements, and long-term risk monitoring;\n• Mandate post-market surveillance and public reporting \nmechanisms for AI-related adverse events.\n5. Design Inclusive, Offline-Capable Platforms\n• Ensure that AI tools are accessible in low-bandwidth \nenvironments and compatible with widely available \nmobile devices;\n• Tailor digital interfaces and educational modules to \naccommodate different literacy levels, cultural back-\ngrounds, and user abilities.\nBy systematically implementing these recommendations, \nstakeholders can create a responsible and patient-centered AI \necosystem that advances early diagnosis, personalized man-\nagement, and psychosocial support for women living with \nendometriosis, while safeguarding against unintended harms.\nConclusion\nArtificial Intelligence (AI) holds unprecedented potential \nto transform the landscape of endometriosis diagnosis, \neducation, and management. Through advanced pattern \nrecognition, predictive analytics, and personalized educa -\ntional pathways, AI can address longstanding challenges \nincluding diagnostic delays, fragmented care pathways, and \nlimited patient empowerment.\nHowever, the realization of this potential is not automatic. \nIt requires confronting and overcoming substantial technical, \nclinical, ethical, and sociocultural barriers. Without deliber-\nate attention to inclusivity, transparency, and participatory \ndevelopment, AI innovations risk reinforcing—rather than \nmitigating—existing inequities in endometriosis care.\nGlobal health authorities, policymakers, researchers, \nand technology developers must recognize endometriosis \nas a significant public health and equity challenge within \nthe broader landscape of digital health innovation. Strategic \ninvestments in ethical governance frameworks, real-world \nvalidation studies, culturally adaptive platforms, and inclu-\nsive digital infrastructures are critical.\nFuture research must prioritize hybrid effective-\nness–implementation studies that assess not only clinical \nefficacy but also usability, sustainability, and impact across \ndiverse settings. Additionally, regulatory bodies must \nevolve to ensure that AI-driven solutions are accountable, \nexplainable, and responsive to patients' rights and societal \nexpectations.\nHarnessing the power of AI to revolutionize endometrio-\nsis care is not merely a technological endeavor; it is a moral \nimperative. By centering the experiences, needs, and voices \nof women living with endometriosis, we can transform inno-\nvation into tangible improvements in health, dignity, and \nquality of life worldwide.\nAuthor contributions Kelnner Portela Luz conceptualized and \ndesigned the study, conducted the literature review, performed the com-\nparative analysis, and drafted the manuscript. The author approved the \nfinal version and is accountable for all aspects of the work.\nFunding This study received no external funding.\nAvailability of data and materials Not applicable. No new datasets were \ngenerated or analyzed for this study.\nDeclarations \nConflict of interest The author declares no conflicts of interest.\nEthics approval and consent to participate Not applicable. This manu-\nscript is a narrative review and does not involve human participants, \ndata, or material requiring ethical approval.\nConsent for publication Not applicable.\nOpen Access  This article is licensed under a Creative Commons Attri-\nbution 4.0 International License, which permits use, sharing, adapta-\ntion, distribution and reproduction in any medium or format, as long \nas you give appropriate credit to the original author(s) and the source, \nprovide a link to the Creative Commons licence, and indicate if changes \nwere made. The images or other third party material in this article are \n\nJournal of Medical Imaging and Interventional Radiology           (2025) 12:15  \n Page 7 of 7    15 \nincluded in the article's Creative Commons licence, unless indicated \notherwise in a credit line to the material. If material is not included in \nthe article's Creative Commons licence and your intended use is not \npermitted by statutory regulation or exceeds the permitted use, you will \nneed to obtain permission directly from the copyright holder. To view a \ncopy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.\nReferences\n 1. Zondervan KT, Becker CM, Missmer SA (2020) Endometriosis. \nN Engl J Med 382(13):1244–1256\n 2. Dunselman GAJ, Vermeulen N, Becker C et al (2014) ESHRE \nguideline: management of women with endometriosis. Hum \nReprod 29(3):400–412\n 3. Pereira RM, Motta F (2022) Diagnostic barriers in endometriosis: \na narrative review. Rev Bras Ginecol Obstet 44(1):56–63\n 4. United Nations Educational, Scientific and Cultural Organization \n(UNESCO) (2022) Recommendation on the Ethics of Artificial \nIntelligence. UNESCO, Paris\n 5. World Health Organization (WHO) (2021) Ethics and Govern-\nance of Artificial Intelligence for Health: WHO Guidance. WHO, \nGeneva\n 6. Esteva A, Robicquet A, Ramsundar B et al (2019) A guide to deep \nlearning in healthcare. Nat Med 25(1):24–29\n 7. Topol EJ (2019) High-performance medicine: the convergence of \nhuman and artificial intelligence. Nat Med 25(1):44–56\n 8. He H, Zhang Y, Zhang R et al (2023) Application of artificial \nintelligence in the diagnosis and prediction of endometriosis. \nFront Med (Lausanne) 10:1172845\n 9. Ferroni P, Zanzotto FM, Scatena R et al (2019) Artificial intel-\nligence for cancer diagnosis and prognosis: a systematic review. \nJ Clin Med 8(8):1015\n 10. Kaur A, Saini A, Saggar K (2022) Artificial intelligence in \nobstetrics and gynecology: Friend or foe? Obstet Gynecol Sci \n65(3):207–217\n 11. Chen R, Lu X, Zhang Y (2020) Mining clinical narratives for \npatient-reported symptoms of endometriosis using natural lan-\nguage processing. J Biomed Inform 108:103499\n 12. Sittig DF, Singh H (2010) A new sociotechnical model for study-\ning health information technology in complex adaptive healthcare \nsystems. Qual Saf Health Care 19(Suppl 3):i68-74\n 13. Thiers FA, Gaspar J, Oliveira R et al (2022) Deep learning for \ndetection of endometriotic lesions in MRI: a preliminary study. \nComput Biol Med 146:105589\n 14. Obermeyer Z, Emanuel EJ (2016) Predicting the future—big \ndata, machine learning, and clinical medicine. N Engl J Med \n375(13):1216–1219\n 15. Goodman KW (2015) Ethics, medicine, and information technol-\nogy: intelligent machines and the transformation of health care. \nCambridge University Press, Cambridge\n 16. Fiske A, Henningsen P, Buyx A (2019) Your robot therapist will \nsee you now: ethical implications of embodied artificial intelli-\ngence in mental health care. J Med Internet Res 21(5):e13216\n 17. Morley J, Luciano F (2021) A framework for the ethical impact \nassessment of artificial intelligence. Nat Mach Intell 3(2):89–94\n 18. Fjeld J, Achten N, Hilligoss H et al (2020) Principled artificial \nintelligence: mapping consensus in ethical and rights-based \napproaches. Berkman Klein Center Research Publication, \nCambridge\n 19. Khullar D (2019) Building a smarter health care system with arti-\nficial intelligence. JAMA 321(22):2151–2152\n 20. Mittelstadt BD, Floridi L (2016) The ethics of big data: current \nand foreseeable issues in biomedical contexts. Sci Eng Ethics \n22(2):303–341\n 21. European Commission (2019) Ethics guidelines for trustworthy \nAI. High-level expert group on artificial intelligence. European \nCommission, Brussels\n 22. Serra MF (2021) Descomplicando a perícia psiquiátrica no con-\ntexto trabalhista: uma proposta de sistematização na análise do \nnexo. Perspect Med Legal Perícias Méd 6:e210814\n 23. U.S. Food and Drug Administration (FDA) (2021) Artificial Intel-\nligence and Machine Learning in Software as a Medical Device\n 24. Brasil. Lei Geral de Proteção de Dados Pessoais – LGPD. Lei nº \n13.709, de 14 de agosto de 2018\n 25. European Union. General Data Protection Regulation (GDPR). \nRegulation (EU) 2016/679\n 26. Mehedintu C, Plotogea MN, Ionescu S et al (2014) Endometriosis \nstill a challenge. J Med Life 7(3):349–357\n 27. MyEndometriosisTeam. Available from: https:// www. myend ometr \niosis team. com/\nPublisher's Note Springer Nature remains neutral with regard to \njurisdictional claims in published maps and institutional affiliations.","source_license":"CC0","license_restricted":false}