Clinical Cyberbioethics and AI-Mediated Clinical Decision-Making: a mapping and narrative review with thematic synthesis (SWiM)

Clinical Cyberbioethics and AI-Mediated Clinical Decision-Making: a mapping and narrative review with thematic synthesis (SWiM) | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Systematic Review Clinical Cyberbioethics and AI-Mediated Clinical Decision-Making: a mapping and narrative review with thematic synthesis (SWiM) Anderson Díaz Pérez, Wendy Acuña Pérez This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8585342/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Artificial intelligence (AI) is rapidly reshaping clinical decision-making, yet it introduces ethically salient risks including bias, automation bias, opacity, privacy threats, data drift, adversarial vulnerability, and unclear accountability. Objective: To define and operationalize Clinical Cyberbioethics as an integrative framework for AI-mediated clinical decision-making, synthesizing ethical principles, governance requirements, and patient-facing digital rights proposed across the literature. Methods: We conducted a mapping and narrative review with SWiM-guided thematic synthesis covering January 2015 to March 2025, and documented identification, screening, and inclusion decisions using a PRISMA 2020–adapted flow diagram. Records were identified from an author-curated reference set compiled during the stated search period; screening and selection are reported in Fig. 1 and Supplementary Files 1–4. Results: Records identified (n = 54); duplicates removed (n = 11); records screened (n = 43); records excluded at title/abstract (n = 1); full-text sources assessed for eligibility (n = 42); studies included in synthesis (n = 42); full-text exclusions with reasons (n = 0). The synthesis yielded operational domains spanning decision authority, explainability/contestability, bias and equity safeguards, lifecycle governance and accountability, and proportional data governance and privacy safeguards. Conclusions: Clinical Cyberbioethics provides a practical framework to guide trustworthy AI-mediated clinical decision-making by integrating ethical principles with governance and patient rights. Medical Ethics Artificial Intelligence and Machine Learning Clinical Ethics Artificial Intelligence Bioethics Clinical Decision-Making Digital Health Computational Bioethics Patient Rights Figures Figure 1 Figure 2 Background Healthcare is undergoing a profound digital transformation. Artificial intelligence (AI), electronic health records (EHRs), telemedicine, and AI-enabled clinical decision support systems (CDSS) are progressively embedded in diagnostic reasoning, risk stratification, workflow prioritization, and treatment planning. When robustly developed and clinically validated, these technologies can improve efficiency, support earlier detection, enable personalization, and reduce certain types of human error fueling the vision of “high-performance” medicine where human expertise is augmented by data-driven systems [ 1 , 2 , 3 ]. Yet the same integration introduces a new class of ethical tensions because clinical decisions are increasingly shaped by algorithmic mediation, not solely by clinician judgment and patient values [ 2 , 3 ]. Classical biomedical ethics autonomy, beneficence, non-maleficence, and justice remains essential for clinical care [ 4 ]. However, AI-mediated decision-making adds moral problems that are not fully captured by traditional principle-based analysis alone: opacity and limited contestability of recommendations, automation bias and overreliance, post-deployment performance decay (data drift), cyber-risk exposure, and multi-actor accountability in hybrid human–machine decisions [ 6 , 7 , 14 , 18 ]. Cyberethics contributes a critical lens on digital privacy, data protection, online harms, and responsibility in technologically mediated environments [ 5 ]. Computational bioethics, in turn, focuses on the ethical properties of algorithmic systems fairness, explainability, bias mitigation, transparency, and governance across the AI lifecycle [ 6 , 7 ]. While each domain offers indispensable insights, their fragmented application leaves institutions without a unified, operational framework for clinical deliberation when AI outputs influence diagnosis and treatment decisions [ 8 , 9 , 10 ]. Empirical evidence underscores that these risks are not theoretical. Algorithmic bias has been documented in widely deployed health management tools, where seemingly “objective” risk predictions amplified racial disparities due to biased proxies and structural inequities [ 11 ]. Pipeline-wide sources of bias can occur at problem framing, data collection, labeling, model design, evaluation, and deployment ultimately distorting clinical decisions and perpetuating inequities [ 12 , 13 ]. Moreover, clinician AI interaction can generate automation bias : clinicians may accept erroneous system outputs, particularly under time pressure or when algorithmic recommendations are accompanied by persuasive explanations [ 14 , 15 ]. In a randomized clinical vignette survey, systematically biased AI predictions reduced clinician diagnostic accuracy, and commonly used explanation outputs did not reliably mitigate the harm [ 16 ]. These findings highlight a core ethical challenge for clinical AI: the risk is not only a flawed model, but also a flawed socio-technical decision process. Additional risk pathways are increasingly salient in real-world deployment. AI models are vulnerable to adversarial manipulation and security threats, raising concerns about patient safety, reliability, and trust in clinical infrastructure [ 17 ]. Even without malicious attacks, models may deteriorate after deployment due to dataset shift and changing clinical practice patterns—requiring ongoing monitoring, recalibration, and governance to avoid silent performance failures [ 18 ]. These risks coexist with legitimate clinical benefits; therefore, ethically defensible adoption requires a structured approach to balancing expected gains against foreseeable harms, with safeguards that are both technical (validation, monitoring, cybersecurity) and normative (accountability, patient autonomy, fairness, transparency, contestability). Global and regulatory-facing guidance increasingly emphasizes these safeguards but operationalization at the bedside remains inconsistent. International frameworks call for transparency, accountability, safety, and equity in AI for health [ 8 , 9 , 10 ]. Reporting guidelines such as CONSORT-AI and SPIRIT-AI strengthen prospective evaluation of AI interventions, while TRIPOD + AI and STARD-AI improve transparency for prediction and diagnostic accuracy studies, respectively; these standards must be translated into institutional routines for procurement, deployment, oversight, and clinical deliberation [ 22 , 23 , 24 , 25 , 41 ]. Complementary governance tools such as Good Machine Learning Practice principles and lifecycle risk-management frameworks reinforce the need for continuous oversight, traceability, and responsibility assignment beyond initial model development and clinical validation [ 20 , 21 , 26 , 28 ]. Taken together, these developments motivate the need for a dedicated applied framework Clinical Cyberbioethics capable of integrating classical bioethical reasoning with cyberethical safeguards and computational bioethics requirements across the full clinical AI lifecycle . Such a framework must explicitly address (i) how patient autonomy is exercised when algorithmic outputs shape choices, (ii) how to prevent inequity and discrimination in AI-supported care, (iii) how accountability and traceability are assigned across multi-actor systems, and (iv) how clinical deliberation is preserved as a human-centered, dignity-protecting practice under algorithmic mediation. This review therefore situates Clinical Cyberbioethics as a necessary bridge between normative ethics and the concrete socio-technical realities of AI-mediated clinical decision-making. Methods Design and reporting framework This study was conducted as a mapping and narrative review with SWiM-guided thematic synthesis, aiming to map ethically salient risk pathways and governance requirements in AI-mediated clinical decision-making and to integrate these findings into an applied framework termed Clinical Cyberbioethics. Reporting uses PRISMA 2020 as an adapted transparency framework (to document identification, screening, and inclusion decisions), recognizing that the source corpus is an author-curated reference set rather than a single exhaustive database export [ 29 , 30 ]. Protocol and transparency Protocol and transparency A protocol was finalized prior to screening and extraction. The protocol and extraction materials are provided in Supplementary Files 1–2 to support traceability of eligibility rules, search logic, and data items. Eligibility criteria We included sources published between January 2015 and March 2025 (English or Spanish) that provided substantive ethical analysis, governance content, or normatively grounded discussion explicitly addressing AI-mediated clinical decision-making in clinical care or healthcare delivery contexts. Eligible source types included empirical studies, methodological/review articles with substantive ethical content, and governance instruments (e.g., international guidelines or regulatory documents) included as a contextual corpus to support policy-to-bedside mapping. We excluded purely technical performance papers without ethical analysis, non-clinical contexts, and items lacking sufficient scholarly substance. Information sources and search strategy Search activity occurred during the stated search period using an author-curated reference set assembled from iterative scoping across PubMed/MEDLINE, Scopus, Web of Science, SciELO, and Google Scholar, complemented by backward and selective forward citation checking of seminal works. An illustrative PubMed query used during piloting is reported in Supplementary File 1 to document keyword logic; the PRISMA counts reflect screening of the curated reference set rather than a database re-run. Study selection and PRISMA-adapted flow Screening proceeded in two stages (title/abstract, then full text). Duplicates were removed prior to screening. Figure 1 summarizes the selection process for the curated reference set: records identified (n = 54), duplicates removed (n = 11), records screened (n = 43), records excluded at title/abstract (n = 1), full-text sources assessed for eligibility (n = 42), full-text exclusions with reasons (n = 0), and studies included in synthesis (n = 42). The screening log is provided in Supplementary File 4 [ 29 ]. Data extraction (data items) Using a standardized extraction form (Supplementary File 2), we captured bibliographic information, study type (empirical vs normative vs review), clinical domain, AI modality (when reported), ethical dilemmas and normative frameworks invoked, and proposed safeguards (e.g., bias monitoring, explainability/contestability, auditability, accountability allocation, privacy and data governance, lifecycle monitoring including drift and security threats). The list of included sources is provided in Supplementary File 3 (Table S1). Quality/credibility appraisal Given heterogeneity across empirical, methodological, and normative sources, appraisal was design-appropriate. Reviews were appraised using AMSTAR 2; empirical studies were appraised using relevant JBI tools; normative analyses were assessed using an adapted rubric focusing on argument clarity, justificatory depth, engagement with counterarguments, and clinical applicability. Appraisal informed weighting during interpretation rather than serving as an automatic exclusion criterion. Data synthesis (SWiM-guided thematic synthesis) Meta-analysis was not appropriate due to conceptual and methodological heterogeneity. We therefore conducted a SWiM-guided narrative thematic synthesis combining deductive and inductive coding to map recurring ethical tensions, governance requirements, and operational safeguards in AI-mediated clinical decision-making. [ 30 ] Use of AI tools (transparency statement) Large language models were used only to support drafting and language refinement. No AI system was used to make eligibility decisions, perform data extraction, or generate interpretive conclusions; the author takes full responsibility for the integrity and accuracy of the work. Results Study selection The study identification and selection process is summarized in the PRISMA 2020 flow diagram (Fig. 1 ) and in the screening log (Supplementary File 4). Fifty-four records were identified from the curated reference set; 11 duplicates were removed; 43 records were screened at title/abstract level; 1 record was excluded at title/abstract; 42 full-text sources were assessed for eligibility; no full texts were excluded with reasons; and 42 sources were included in the narrative thematic synthesis [ 29 ]. Each dot represents a peer-reviewed source (numbered as in the reference list) contributing to a given ethical or clinical dimension. The map illustrates the distribution and concentration of evidence across domains such as accountability, bias, governance, explainability, data quality, and clinical impact Discussion AI is increasingly embedded in clinical workflows, but ethical risk in practice is rarely driven by a single principle in isolation. Across the included corpus, harms and governance failures emerge at the intersection of technical properties (opacity, distribution shift/data drift, and adversarial vulnerability) [ 17 , 18 ], human factors (automation bias, over‑reliance, and accountability diffusion) [ 14 , 15 ], and institutional conditions (data governance, procurement incentives, audit capacity, and regulatory maturity). The central contribution of this review is to translate this multi‑layered problem space into an operational, clinically anchored framework Clinical Cyberbioethics that links classical bioethics, digital ethics, and computational governance into workflow‑ready requirements for AI‑mediated clinical decision‑making. This emphasis on operationalization also responds to the persistent gap between explainability ideals and clinically actionable explanations, including the need for causability‑oriented interpretability in medical contexts [ 19 , 35 , 36 , 37 , 38 ]. Positioning relative to existing guidance: International instruments such as the WHO guidance on AI for health, UNESCO’s Recommendation on the Ethics of AI, and European trustworthy‑AI frameworks converge on high‑level principles (beneficence, non‑maleficence, autonomy, justice, transparency, and accountability) [ 8 , 9 , 10 ]. However, implementation frequently stalls because clinical teams and institutions require actionable specifications—what must be documented, audited, monitored, and governed at each stage of the model lifecycle and clinical use. Clinical Cyberbioethics is proposed as a bridging layer that (i) specifies operational obligations (e.g., traceability of data provenance, model versioning, and decision rationale), (ii) embeds meaningful human oversight and escalation pathways, and (iii) makes equity and patient protections testable through monitoring and governance checkpoints [ 21 ]. From principles to operational requirements: The synthesis supports a set of recurring implementation requirements that can be treated as minimum conditions for ethically defensible clinical AI. These include: (1) transparency and traceability (model documentation, provenance, intended use, and limitations); (2) meaningful human oversight (clear responsibility allocation, override authority, and clinical deliberation); (3) equity and bias mitigation (subgroup evaluation, deployment-context validation, and monitoring for disparate impact) [ 11 , 12 , 13 ]; (4) privacy and data governance (data minimization, access controls, and safeguards against re-identification) [ 6 ]; and (5) lifecycle safety (drift detection, incident reporting, and post-deployment auditing) [ 18 ]. Beyond subgroup performance, the debate on race correction in clinical algorithms and subsequent movement toward race-neutral equations for estimating kidney function illustrates how normative critique can translate into concrete model and policy changes [ 32 , 33 ]. Importantly, several of these requirements are socio-technical: they depend not only on model design but also on training, workflow integration, procurement, and institutional governance. Existing reporting and evaluation extensions provide a practical foundation for this operational layer, including CONSORT‑AI, SPIRIT‑AI, TRIPOD + AI, STARD‑AI, and international consensus guidance on deployable and trustworthy AI [ 22 , 23 , 24 , 25 , 26 , 41 ]. TRIPOD + AI also builds on the original TRIPOD explanation and elaboration for prediction model reporting [ 24 , 40 ]. Complementary governance tools such as Good Machine Learning Practice and risk‑management frameworks further support lifecycle controls and accountability assignment [ 20 , 21 , 28 ]. Patient‑facing digital rights and accountability: A distinctive practical implication of the proposed framework is the emphasis on patient‑facing digital rights as complements to professional duties and institutional safeguards [ 6 , 8 , 9 ]. The corpus repeatedly motivates rights such as the right to be informed when AI materially contributes to a clinical decision, the right to meaningful explanation appropriate to the clinical context, the right to privacy and data stewardship, and the right to contest or seek human review for high‑impact decisions. These rights should not remain abstract: they can be operationalized through consent language, clinical documentation, audit trails, and complaint/escalation mechanisms, and they align with emerging regulatory attention to high‑risk AI systems and accountability requirements [ 27 ]. Given evidence that re-identification can be feasible even in incomplete datasets, privacy safeguards should include explicit re-identification risk assessment and governance beyond de-identification alone [ 39 ]. Implications for research, practice, and policy: For researchers, the findings underscore the need to report clinical AI interventions using AI‑specific extensions of reporting guidelines and to evaluate not only accuracy but also equity, usability, and unintended effects [ 22 , 23 , 24 , 25 , 26 , 28 , 42 ]. Independent external validation of widely implemented proprietary models has revealed limitations in generalizability and real‑world performance, reinforcing the need for transparency and independent evaluation prior to and after deployment [ 34 ]. For clinical leaders, Clinical Cyberbioethics supports procurement and deployment checklists that make responsibilities explicit (who monitors drift, who audits equity, who responds to incidents) and that require governance artifacts (model cards, data documentation, and audit logs) as conditions of adoption [ 20 , 21 ]. For policymakers and regulators, the framework suggests that compliance should incorporate measurable post‑market obligations—monitoring, transparency artifacts, and accountability pathways—rather than relying solely on pre‑deployment approval [ 21 , 27 ]. Future directions: The next step is implementation validation. This includes developing and testing practical tools—checklists, minimum dataset documentation, audit protocols, and patient information templates—followed by prospective evaluation of impact on decision quality, clinician trust calibration, and patient‑centered outcomes [ 21 , 26 ]. Comparative studies across settings and populations are needed to refine equity‑sensitive governance and to reduce documented bias pathways in real‑world systems [ 11 , 12 , 13 ]. Finally, the framework should be iteratively updated as regulatory regimes, reporting standards, and clinical AI architectures evolve [ 27 , 28 ]. Limitations This review has limitations. Identification relied on an author-curated reference set supplemented by backward reference checking; coverage may therefore be incomplete relative to a multi-database systematic search, and selection bias is possible. The corpus includes heterogeneous sources (empirical studies, reviews, normative analyses, and governance instruments), which supports conceptual and operational mapping but precludes quantitative synthesis. Quality appraisal was used to contextualize credibility rather than to exclude sources, and screening was conducted by a single reviewer; future updates should incorporate independent duplicate screening and full database exports. These limitations align with guidance on transparent reporting for evidence syntheses and synthesis-without-meta-analysis approaches (PRISMA 2020; SWiM) [ 29 , 30 ]. Conclusions This mapping and narrative review with thematic synthesis (SWiM) (2015–2025) suggests that classical biomedical ethics, cyberethics, and computational bioethics remain necessary but, when applied in isolation, are operationally insufficient for recurring dilemmas of AI-mediated clinical decision-making [ 29 , 30 ]. Across the included sources, ethical risk is driven by how algorithmic recommendations are embedded into workflows—opacity at the point of care, automation bias and overreliance, uneven subgroup performance and bias, post-deployment drift, diffuse accountability, and privacy vulnerabilities in linked clinical data [ 11 , 12 , 13 , 14 , 15 , 17 , 18 , 19 , 21 , 22 ]. Although global and regional instruments (e.g., WHO guidance on AI in health, UNESCO’s Recommendation on the Ethics of AI, European trustworthy‑AI guidance, the EU AI Act, and the Council of Europe AI Convention) converge on high-level commitments such as transparency, accountability, safety, and equity, they remain under-specified for bedside implementation [ 8 , 9 , 10 , 27 , 31 ]. In particular, the literature reveals persistent gaps around actionable thresholds (what counts as sufficient transparency), enforceable oversight (how override and contestability function under time pressure), auditable traceability (what must be logged and by whom), and the allocation of responsibilities across clinicians, institutions, and vendors; risk-management frameworks such as the NIST AI RMF and GMLP help translate these commitments into lifecycle controls, but clinical operationalization remains uneven [ 20 , 21 ]. This article contributes by defining Clinical Cyberbioethics as an integrative, practice-oriented field at the intersection of clinical ethics, cyberethics, and computational bioethics, explicitly centered on AI-enabled and digitally mediated clinical authority. It further proposes (i) seven operational principles and (ii) five candidate fifth-generation digital rights for AI-mediated care, intended to translate normative commitments into patient protections and workflow-level duties [ 8 , 9 , 10 ]. Finally, the Deliberative Clinical Cyberbioethics Model (D-CCM) is presented as a pragmatic structure to embed ethical reasoning into routine care through: (a) transparency that is clinically meaningful, (b) meaningful human oversight and contestability, (c) equity monitoring and bias mitigation, (d) accountability engineering and traceability, and (e) proportional data governance and privacy safeguards [ 11 , 12 , 13 , 19 , 20 , 21 , 27 ]. The relevance of Clinical Cyberbioethics is threefold. First, it reframes ethical reasoning in digital healthcare by linking principles to the AI lifecycle (design–validation–deployment–monitoring) and to real clinical constraints [ 8 , 20 , 21 ]. Second, it supports shared deliberation by clinicians and patients when AI influences diagnosis, prognosis, or treatment options, thereby strengthening autonomy and calibrated trust and reducing inappropriate reliance [ 14 , 15 , 16 ]. Third, it offers a coherent basis for institutional governance, training curricula, and policy design anchored in human-rights–compatible safeguards [ 9 , 27 , 31 ]. Future research should move beyond conceptual endorsement and empirically evaluate the D-CCM in real clinical environments using measurable outcomes: patient understanding and autonomy, calibrated reliance and reduced automation bias, safety events and workflow burden, subgroup equity metrics, traceability and accountability indicators, and patient trust [ 11 , 12 , 13 , 14 , 15 , 16 , 21 ]. Establishing feasibility, acceptability, and effectiveness across settings—and aligning evaluations with emerging clinical AI reporting standards—will determine whether Clinical Cyberbioethics can serve as a durable ethical infrastructure for responsible AI implementation in healthcare [ 22 , 23 , 24 , 25 , 26 ]. Declarations Ethics approval and consent to participate Not applicable. This study is a mapping and narrative review of published literature and did not involve human participants, individual-level human data, or animal subjects. Consent for publication Not applicable. This manuscript does not report identifiable individual data. Availability of data and materials All information extracted and synthesized for this mapping and narrative review is reported in the main text and in Supplementary Files 1–4. No new datasets were generated or analyzed beyond the cited sources. Competing interests The author declares that he has no competing interests. Funding This research received no external funding. Authors’ contributions ADP (Anderson Díaz Pérez): Conceptualization; methodology; literature search; screening and selection; data extraction; synthesis and interpretation; drafting and critical revision of the manuscript; visualization (PRISMA flow materials); and approval of the final version. WJAP (Wendy Johana Acuña Pérez): Conceptualization (public health and nursing perspective); interpretation and contextualization of findings; writing—review and editing; and approval of the final version.. Acknowledgements The author thanks academic colleagues and research groups engaged in ongoing discussions at the intersection of bioethics, clinical practice, and artificial intelligence, which informed the development of the concept of Clinical Cyberbioethics. Authors’ information Anderson Díaz Pérez, PhD (Bioethics); MSc (Biomedical Basic Sciences); Specialist (Artificial Intelligence). Affiliation: Department of Social and Human Sciences, Universidad Simón Bolívar, Barranquilla, Colombia. ORCID: 0000-0003-2448-0953. Email: [email protected] . Wendy Johana Acuña Pérez, RN; MSc (Public Health). Affiliation: Faculty of Health Sciences, Nursing Program, Corporación Universitaria Rafael Núñez, Barranquilla, Colombia. ORCID: 0000-0001-7755-8588. Email: [email protected] . Use of artificial intelligence in writing Artificial intelligence tools (ChatGPT, OpenAI) were used to assist with language refinement and the drafting of preliminary text. The author critically reviewed, edited, and validated all content and takes full responsibility for the accuracy, originality, and integrity of the work. No AI tool was used to make decisions about eligibility, data extraction, or the interpretation of included evidence. References Topol EJ (2019) High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25(1):44–56. 10.1038/s41591-018-0300-7 Rajkomar A, Dean J, Kohane IS (2019) Machine learning in medicine. N Engl J Med 380(14):1347–1358. 10.1056/NEJMra1814259 Nagendran M, Chen Y, Lovejoy CA, Gordon AC, Komorowski M et al (2020) Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ 368:m689. 10.1136/bmj.m689 Beauchamp TL, Childress JF (2019) Principles of Biomedical Ethics, 8th edn. Oxford University Press, New York Ess C (2009) Digital Media Ethics, 2nd edn. Polity, Cambridge Mittelstadt BD, Allo P, Taddeo M, Wachter S, Floridi L (2016) The ethics of algorithms: mapping the debate. Big Data Soc 3(2):1–21. 10.1177/2053951716679679 Morley J, Machado CCV, Burr C, Cowls J, Joshi I, Taddeo M, Floridi L (2020) The ethics of AI in health care: a mapping review. Soc Sci Med 260:113172. 10.1016/j.socscimed.2020.113172 World Health Organization (2021) Ethics and governance of artificial intelligence for health: WHO guidance. WHO, Geneva UNESCO (2021) Recommendation on the Ethics of Artificial Intelligence. UNESCO, Paris European Commission High-Level Expert Group on AI (2019) Ethics Guidelines for Trustworthy AI. European Commission, Brussels Obermeyer Z, Powers B, Vogeli C, Mullainathan S (2019) Dissecting racial bias in an algorithm used to manage the health of populations. Science 366(6464):447–453. 10.1126/science.aax2342 Cross JL, Choma MA, Onofrey JA (2024) Bias in medical AI: implications for clinical decision-making. PLOS Digit Health 3(11):e0000651. 10.1371/journal.pdig.0000651 Nazer LH et al (2023) Bias in artificial intelligence algorithms and recommendations for mitigation. PLOS Digit Health 2(7):e0000278. 10.1371/journal.pdig.0000278 Goddard K, Roudsari A, Wyatt JC (2012) Automation bias: a systematic review of frequency, effect mediators, and mitigators. J Am Med Inf Assoc 19(1):121–127. 10.1136/amiajnl-2011-000469 Abdelwanis M, Alarafati HK, Tammam MMS, Simsekler MCE (2024) Exploring the risks of automation bias in healthcare artificial intelligence applications: a Bowtie analysis. J Saf Sci Resil 5:460–469. 10.1016/j.jnlssr.2024.06.001 Jabbour S, Fouhey D, Shepard S et al (2023) Measuring the impact of AI in the diagnosis of hospitalized patients: a randomized clinical vignette survey study. JAMA 330(23):2275–2284. 10.1001/jama.2023.22295 Finlayson SG, Bowers JD, Ito J, Zittrain JL, Beam AL, Kohane IS (2019) Adversarial attacks on medical machine learning. Science 363(6433):1287–1289. 10.1126/science.aaw4399 Sahiner B et al (2023) Data drift in medical machine learning: implications and potential remedies. Br J Radiol 96:20220878. 10.1259/bjr.20220878 Holzinger A, Langs G, Denk H, Zatloukal K, Müller H (2019) Causability and explainability of artificial intelligence in medicine. Wiley Interdiscip Rev Data Min Knowl Discov 9(4):e1312. 10.1002/widm.1312 Food and Drug Administration (FDA) (2021) Health Canada, MHRA. Good Machine Learning Practice for Medical Device Development: Guiding Principles National Institute of Standards and Technology (NIST) (2023) Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100–101 Liu X, Cruz Rivera S, Moher D, Calvert MJ, Denniston AK et al (2020) Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. BMJ 370:m3164. 10.1136/bmj.m3164 Rivera SC, Liu X, Chan A-W, Denniston AK, Calvert MJ et al (2020) Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. BMJ 370:m3210. 10.1136/bmj.m3210 Collins GS, Moons KGM, Dhiman P et al (2024) TRIPOD + AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385:e078378. 10.1136/bmj-2023-078378 Sounderajah V, Guni A, Liu X, Collins GS, Karthikesalingam A et al (2025) The STARD-AI reporting guideline for diagnostic accuracy studies using artificial intelligence. Nat Med 31(10):3283–3289. 10.1038/s41591-025-03953-8 Lekadir K et al (2025) FUTURE-AI: international consensus guideline for trustworthy and deployable artificial intelligence in healthcare. BMJ 388:e081554. 10.1136/bmj-2024-081554 Regulation (EU) (2024) 2024/1689 of the European Parliament and of the Council: Artificial Intelligence Act Kolbinger FR et al (2024) Reporting guidelines in medical artificial intelligence. Commun Med. 10.1038/s43856-024-00492-0 Page MJ, McKenzie JE, Bossuyt PM et al (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71. 10.1136/bmj.n71 Campbell M, McKenzie JE, Sowden A et al (2020) Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. BMJ 368:l6890. 10.1136/bmj.l6890 Council of Europe (2024) Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law. Council of Europe Vyas DA, Eisenstein LG, Jones DS (2020) Hidden in plain sight—reconsidering the use of race correction in clinical algorithms. N Engl J Med 383(9):874–882. 10.1056/NEJMms2004740 Inker LA, Eneanya ND, Coresh J et al (2021) New creatinine- and cystatin C–based equations to estimate GFR without race. N Engl J Med. 10.1056/NEJMoa2102953 Wong A, Otles E, Donnelly JP et al (2021) External validation of a widely implemented proprietary sepsis prediction model in hospitalized patients. JAMA Intern Med Ghassemi M, Oakden-Rayner L, Beam AL (2021) The false hope of current approaches to explainable artificial intelligence in health care. Lancet Digit Health 3:e745–e750. 10.1016/S2589-7500(21)00208-9 Rudin C (2019) Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat Mach Intell 1:206–215. 10.1038/s42256-019-0048-x Ribeiro MT, Singh S, Guestrin C (2016) Why should I trust you? Explaining the predictions of any classifier. In: Proc. KDD. 10.1145/2939672.2939778 Lundberg SM, Lee SI (2017) A unified approach to interpreting model predictions. In: Advances in Neural Information Processing Systems (NeurIPS). 10.5555/3295222.3295230 Rocher L, Hendrickx JM, de Montjoye YA (2019) Estimating the success of re-identifications in incomplete datasets using generative models. Nat Commun 10:3069. 10.1038/s41467-019-10933-3 Moons KGM, Altman DG, Reitsma JB et al (2015) Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 162:W1–W73. 10.7326/M14-0698 Cruz Rivera S, Liu X, Chan AW et al (2020) Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nat Med. 10.1038/s41591-020-1037-7 Vasey B, Nagendran M, Campbell B et al (2022) Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. Nat Med. 10.1038/s41591-022-01772-9 Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8585342","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Systematic Review","associatedPublications":[],"authors":[{"id":573521672,"identity":"ea7bf46a-c739-428c-b3c6-bbb74a61366c","order_by":0,"name":"Anderson Díaz Pérez","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIiWNgGAWjYHACAwaGiv88bOwNILYFsVrOMMvw8RwAsSWI1MLYxmwjJ5EA4hChhX9288bHFWxsPGySz69u+FEgwcDf3p2AV4vEnWPFhmd4eHjYpHPKbvYAHSZx5uwG/NbcyDGTbJCQAGlJu8ED1GIgkYtfi/yNHPOfDQYGQIedSbv5hxgtBkBbGBsSEnjYJNiP3SbKFsMbacWSDQcO8LDx5LDdljGQ4CHoF7kbyRs/Nv47YC/ffvzZzTd/bOT423sJeB8BeAzAJLHKQYD9ASmqR8EoGAWjYAQBANy3QoF0lSOFAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-2448-0953","institution":"Universidad Simón Bolívar","correspondingAuthor":true,"prefix":"","firstName":"Anderson","middleName":"Díaz","lastName":"Pérez","suffix":""},{"id":573521673,"identity":"a00f0c5a-230e-40b9-850c-fe6dc1f67427","order_by":1,"name":"Wendy Acuña Pérez","email":"","orcid":"https://orcid.org/0000-0001-7755-8588","institution":"Corporacion Universitaria Rafael Nuñez","correspondingAuthor":false,"prefix":"","firstName":"Wendy","middleName":"Acuña","lastName":"Pérez","suffix":""}],"badges":[],"createdAt":"2026-01-12 21:20:55","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8585342/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8585342/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100410060,"identity":"9b024c58-0642-4126-8967-cb7b8004aaec","added_by":"auto","created_at":"2026-01-16 13:07:45","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":404480,"visible":true,"origin":"","legend":"","description":"","filename":"ClinicalCyberbioethicsandAIMediatedClinicalDecisionMaking8.docx","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/a41bd3a704b39723268b10d9.docx"},{"id":100410314,"identity":"8e08b18d-d7b0-4777-b28b-e3ded3cf17ec","added_by":"auto","created_at":"2026-01-16 13:08:12","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs8585342.json","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/b03e7566dff6a543050d53cf.json"},{"id":100410548,"identity":"2f27b433-0103-4816-906f-60a0acd65d0f","added_by":"auto","created_at":"2026-01-16 13:08:35","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":91350,"visible":true,"origin":"","legend":"","description":"","filename":"rs85853420enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/898b1691b6fba90a3efe5f02.xml"},{"id":100410320,"identity":"dce09fd2-37de-494c-b6f9-dd5021475c13","added_by":"auto","created_at":"2026-01-16 13:08:12","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51564,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/a24c6f027031f25969efc800.png"},{"id":100410403,"identity":"b4a6e487-9332-4161-b9ef-40e8cc658ea0","added_by":"auto","created_at":"2026-01-16 13:08:20","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":61126,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/3a1a8f17bd7f0572fd4f6c04.png"},{"id":100409155,"identity":"d23d4d7a-d0d2-46f9-8010-3cc9d96e9942","added_by":"auto","created_at":"2026-01-16 13:06:54","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43919,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/c13cf742264ad18214d6cc9f.png"},{"id":100410165,"identity":"e3b18e40-9f3f-4dcf-9aef-b0af280c6bfb","added_by":"auto","created_at":"2026-01-16 13:07:56","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22962,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/b5e3e334e726d22cb29ed5a6.png"},{"id":100410614,"identity":"b6055f22-0ede-4f7d-9e7b-c595701b7d3c","added_by":"auto","created_at":"2026-01-16 13:08:40","extension":"xml","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89603,"visible":true,"origin":"","legend":"","description":"","filename":"rs85853420structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/f217cc7fc1fd07a0b7842f86.xml"},{"id":100409398,"identity":"ba5c3f0e-275e-4529-9b4f-8c989636db23","added_by":"auto","created_at":"2026-01-16 13:07:09","extension":"html","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":100327,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/4e051d87ced345e1fd5e4c0c.html"},{"id":100410063,"identity":"1a09d276-d586-4b8e-8e0b-b1401fa7e3c6","added_by":"auto","created_at":"2026-01-16 13:07:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":133414,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA 2020 flow diagram.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/bde54ca582543dfe21e8efea.png"},{"id":100410441,"identity":"11d2d973-da0f-4df9-9afb-054da1145692","added_by":"auto","created_at":"2026-01-16 13:08:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":246453,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThematic evidence map of Clinical Cyberbioethics in AI-mediated clinical decision-making.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/94c7850ac211dfedcea62704.png"},{"id":100415677,"identity":"8abab38e-fe37-4d7f-ad5e-d23b4cc009f2","added_by":"auto","created_at":"2026-01-16 13:21:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1031048,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8585342/v1/c99721aa-2f0f-4b6e-8aa4-bacee41e4369.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eClinical Cyberbioethics and AI-Mediated Clinical Decision-Making: a mapping and narrative review with thematic synthesis (SWiM)\u003c/p\u003e","fulltext":[{"header":"Background","content":"\u003cp\u003eHealthcare is undergoing a profound digital transformation. Artificial intelligence (AI), electronic health records (EHRs), telemedicine, and AI-enabled clinical decision support systems (CDSS) are progressively embedded in diagnostic reasoning, risk stratification, workflow prioritization, and treatment planning. When robustly developed and clinically validated, these technologies can improve efficiency, support earlier detection, enable personalization, and reduce certain types of human error fueling the vision of \u0026ldquo;high-performance\u0026rdquo; medicine where human expertise is augmented by data-driven systems [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Yet the same integration introduces a new class of ethical tensions because clinical decisions are increasingly shaped by algorithmic mediation, not solely by clinician judgment and patient values [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eClassical biomedical ethics autonomy, beneficence, non-maleficence, and justice remains essential for clinical care [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. However, AI-mediated decision-making adds moral problems that are not fully captured by traditional principle-based analysis alone: opacity and limited contestability of recommendations, automation bias and overreliance, post-deployment performance decay (data drift), cyber-risk exposure, and multi-actor accountability in hybrid human\u0026ndash;machine decisions [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Cyberethics contributes a critical lens on digital privacy, data protection, online harms, and responsibility in technologically mediated environments [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Computational bioethics, in turn, focuses on the ethical properties of algorithmic systems fairness, explainability, bias mitigation, transparency, and governance across the AI lifecycle [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. While each domain offers indispensable insights, their fragmented application leaves institutions without a unified, operational framework for clinical deliberation when AI outputs influence diagnosis and treatment decisions [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eEmpirical evidence underscores that these risks are not theoretical. Algorithmic bias has been documented in widely deployed health management tools, where seemingly \u0026ldquo;objective\u0026rdquo; risk predictions amplified racial disparities due to biased proxies and structural inequities [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Pipeline-wide sources of bias can occur at problem framing, data collection, labeling, model design, evaluation, and deployment ultimately distorting clinical decisions and perpetuating inequities [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Moreover, clinician AI interaction can generate \u003cem\u003eautomation bias\u003c/em\u003e: clinicians may accept erroneous system outputs, particularly under time pressure or when algorithmic recommendations are accompanied by persuasive explanations [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. In a randomized clinical vignette survey, systematically biased AI predictions reduced clinician diagnostic accuracy, and commonly used explanation outputs did not reliably mitigate the harm [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. These findings highlight a core ethical challenge for clinical AI: the risk is not only a flawed model, but also a flawed socio-technical decision process.\u003c/p\u003e \u003cp\u003eAdditional risk pathways are increasingly salient in real-world deployment. AI models are vulnerable to adversarial manipulation and security threats, raising concerns about patient safety, reliability, and trust in clinical infrastructure [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Even without malicious attacks, models may deteriorate after deployment due to dataset shift and changing clinical practice patterns\u0026mdash;requiring ongoing monitoring, recalibration, and governance to avoid silent performance failures [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. These risks coexist with legitimate clinical benefits; therefore, ethically defensible adoption requires a structured approach to balancing expected gains against foreseeable harms, with safeguards that are both technical (validation, monitoring, cybersecurity) and normative (accountability, patient autonomy, fairness, transparency, contestability).\u003c/p\u003e \u003cp\u003eGlobal and regulatory-facing guidance increasingly emphasizes these safeguards but operationalization at the bedside remains inconsistent. International frameworks call for transparency, accountability, safety, and equity in AI for health [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Reporting guidelines such as CONSORT-AI and SPIRIT-AI strengthen prospective evaluation of AI interventions, while TRIPOD\u0026thinsp;+\u0026thinsp;AI and STARD-AI improve transparency for prediction and diagnostic accuracy studies, respectively; these standards must be translated into institutional routines for procurement, deployment, oversight, and clinical deliberation [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Complementary governance tools such as Good Machine Learning Practice principles and lifecycle risk-management frameworks reinforce the need for continuous oversight, traceability, and responsibility assignment beyond initial model development and clinical validation [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTaken together, these developments motivate the need for a dedicated applied framework Clinical Cyberbioethics capable of integrating classical bioethical reasoning with cyberethical safeguards and computational bioethics requirements across the \u003cem\u003efull clinical AI lifecycle\u003c/em\u003e. Such a framework must explicitly address (i) how patient autonomy is exercised when algorithmic outputs shape choices, (ii) how to prevent inequity and discrimination in AI-supported care, (iii) how accountability and traceability are assigned across multi-actor systems, and (iv) how clinical deliberation is preserved as a human-centered, dignity-protecting practice under algorithmic mediation. This review therefore situates Clinical Cyberbioethics as a necessary bridge between normative ethics and the concrete socio-technical realities of AI-mediated clinical decision-making.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDesign and reporting framework\u003c/h2\u003e \u003cp\u003eThis study was conducted as a mapping and narrative review with SWiM-guided thematic synthesis, aiming to map ethically salient risk pathways and governance requirements in AI-mediated clinical decision-making and to integrate these findings into an applied framework termed Clinical Cyberbioethics. Reporting uses PRISMA 2020 as an adapted transparency framework (to document identification, screening, and inclusion decisions), recognizing that the source corpus is an author-curated reference set rather than a single exhaustive database export [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eProtocol and transparency\u003c/h3\u003e\n\u003cdiv class=\"Heading\"\u003eProtocol and transparency\u003c/div\u003e \u003cp\u003eA protocol was finalized prior to screening and extraction. The protocol and extraction materials are provided in Supplementary Files 1\u0026ndash;2 to support traceability of eligibility rules, search logic, and data items.\u003c/p\u003e\n\u003ch3\u003eEligibility criteria\u003c/h3\u003e\n\u003cp\u003eWe included sources published between January 2015 and March 2025 (English or Spanish) that provided substantive ethical analysis, governance content, or normatively grounded discussion explicitly addressing AI-mediated clinical decision-making in clinical care or healthcare delivery contexts. Eligible source types included empirical studies, methodological/review articles with substantive ethical content, and governance instruments (e.g., international guidelines or regulatory documents) included as a contextual corpus to support policy-to-bedside mapping. We excluded purely technical performance papers without ethical analysis, non-clinical contexts, and items lacking sufficient scholarly substance.\u003c/p\u003e\n\u003ch3\u003eInformation sources and search strategy\u003c/h3\u003e\n\u003cp\u003eSearch activity occurred during the stated search period using an author-curated reference set assembled from iterative scoping across PubMed/MEDLINE, Scopus, Web of Science, SciELO, and Google Scholar, complemented by backward and selective forward citation checking of seminal works. An illustrative PubMed query used during piloting is reported in Supplementary File 1 to document keyword logic; the PRISMA counts reflect screening of the curated reference set rather than a database re-run.\u003c/p\u003e\n\u003ch3\u003eStudy selection and PRISMA-adapted flow\u003c/h3\u003e\n\u003cp\u003eScreening proceeded in two stages (title/abstract, then full text). Duplicates were removed prior to screening. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the selection process for the curated reference set: records identified (n\u0026thinsp;=\u0026thinsp;54), duplicates removed (n\u0026thinsp;=\u0026thinsp;11), records screened (n\u0026thinsp;=\u0026thinsp;43), records excluded at title/abstract (n\u0026thinsp;=\u0026thinsp;1), full-text sources assessed for eligibility (n\u0026thinsp;=\u0026thinsp;42), full-text exclusions with reasons (n\u0026thinsp;=\u0026thinsp;0), and studies included in synthesis (n\u0026thinsp;=\u0026thinsp;42). The screening log is provided in Supplementary File 4 [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eData extraction (data items)\u003c/h2\u003e \u003cp\u003eUsing a standardized extraction form (Supplementary File 2), we captured bibliographic information, study type (empirical vs normative vs review), clinical domain, AI modality (when reported), ethical dilemmas and normative frameworks invoked, and proposed safeguards (e.g., bias monitoring, explainability/contestability, auditability, accountability allocation, privacy and data governance, lifecycle monitoring including drift and security threats). The list of included sources is provided in Supplementary File 3 (Table S1).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eQuality/credibility appraisal\u003c/h3\u003e\n\u003cp\u003eGiven heterogeneity across empirical, methodological, and normative sources, appraisal was design-appropriate. Reviews were appraised using AMSTAR 2; empirical studies were appraised using relevant JBI tools; normative analyses were assessed using an adapted rubric focusing on argument clarity, justificatory depth, engagement with counterarguments, and clinical applicability. Appraisal informed weighting during interpretation rather than serving as an automatic exclusion criterion.\u003c/p\u003e\n\u003ch3\u003eData synthesis (SWiM-guided thematic synthesis)\u003c/h3\u003e\n\u003cp\u003eMeta-analysis was not appropriate due to conceptual and methodological heterogeneity. We therefore conducted a SWiM-guided narrative thematic synthesis combining deductive and inductive coding to map recurring ethical tensions, governance requirements, and operational safeguards in AI-mediated clinical decision-making. [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eUse of AI tools (transparency statement)\u003c/h2\u003e \u003cp\u003eLarge language models were used only to support drafting and language refinement. No AI system was used to make eligibility decisions, perform data extraction, or generate interpretive conclusions; the author takes full responsibility for the integrity and accuracy of the work.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eStudy selection\u003c/p\u003e \u003cp\u003eThe study identification and selection process is summarized in the PRISMA 2020 flow diagram (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and in the screening log (Supplementary File 4). Fifty-four records were identified from the curated reference set; 11 duplicates were removed; 43 records were screened at title/abstract level; 1 record was excluded at title/abstract; 42 full-text sources were assessed for eligibility; no full texts were excluded with reasons; and 42 sources were included in the narrative thematic synthesis [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eEach dot represents a peer-reviewed source (numbered as in the reference list) contributing to a given ethical or clinical dimension. The map illustrates the distribution and concentration of evidence across domains such as accountability, bias, governance, explainability, data quality, and clinical impact\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eAI is increasingly embedded in clinical workflows, but ethical risk in practice is rarely driven by a single principle in isolation. Across the included corpus, harms and governance failures emerge at the intersection of technical properties (opacity, distribution shift/data drift, and adversarial vulnerability) [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], human factors (automation bias, over‑reliance, and accountability diffusion) [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], and institutional conditions (data governance, procurement incentives, audit capacity, and regulatory maturity). The central contribution of this review is to translate this multi‑layered problem space into an operational, clinically anchored framework Clinical Cyberbioethics that links classical bioethics, digital ethics, and computational governance into workflow‑ready requirements for AI‑mediated clinical decision‑making. This emphasis on operationalization also responds to the persistent gap between explainability ideals and clinically actionable explanations, including the need for causability‑oriented interpretability in medical contexts [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePositioning relative to existing guidance: International instruments such as the WHO guidance on AI for health, UNESCO\u0026rsquo;s Recommendation on the Ethics of AI, and European trustworthy‑AI frameworks converge on high‑level principles (beneficence, non‑maleficence, autonomy, justice, transparency, and accountability) [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. However, implementation frequently stalls because clinical teams and institutions require actionable specifications\u0026mdash;what must be documented, audited, monitored, and governed at each stage of the model lifecycle and clinical use. Clinical Cyberbioethics is proposed as a bridging layer that (i) specifies operational obligations (e.g., traceability of data provenance, model versioning, and decision rationale), (ii) embeds meaningful human oversight and escalation pathways, and (iii) makes equity and patient protections testable through monitoring and governance checkpoints [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFrom principles to operational requirements: The synthesis supports a set of recurring implementation requirements that can be treated as minimum conditions for ethically defensible clinical AI. These include: (1) transparency and traceability (model documentation, provenance, intended use, and limitations); (2) meaningful human oversight (clear responsibility allocation, override authority, and clinical deliberation); (3) equity and bias mitigation (subgroup evaluation, deployment-context validation, and monitoring for disparate impact) [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]; (4) privacy and data governance (data minimization, access controls, and safeguards against re-identification) [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]; and (5) lifecycle safety (drift detection, incident reporting, and post-deployment auditing) [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Beyond subgroup performance, the debate on race correction in clinical algorithms and subsequent movement toward race-neutral equations for estimating kidney function illustrates how normative critique can translate into concrete model and policy changes [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. Importantly, several of these requirements are socio-technical: they depend not only on model design but also on training, workflow integration, procurement, and institutional governance. Existing reporting and evaluation extensions provide a practical foundation for this operational layer, including CONSORT‑AI, SPIRIT‑AI, TRIPOD\u0026thinsp;+\u0026thinsp;AI, STARD‑AI, and international consensus guidance on deployable and trustworthy AI [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. TRIPOD\u0026thinsp;+\u0026thinsp;AI also builds on the original TRIPOD explanation and elaboration for prediction model reporting [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Complementary governance tools such as Good Machine Learning Practice and risk‑management frameworks further support lifecycle controls and accountability assignment [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePatient‑facing digital rights and accountability: A distinctive practical implication of the proposed framework is the emphasis on patient‑facing digital rights as complements to professional duties and institutional safeguards [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The corpus repeatedly motivates rights such as the right to be informed when AI materially contributes to a clinical decision, the right to meaningful explanation appropriate to the clinical context, the right to privacy and data stewardship, and the right to contest or seek human review for high‑impact decisions. These rights should not remain abstract: they can be operationalized through consent language, clinical documentation, audit trails, and complaint/escalation mechanisms, and they align with emerging regulatory attention to high‑risk AI systems and accountability requirements [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Given evidence that re-identification can be feasible even in incomplete datasets, privacy safeguards should include explicit re-identification risk assessment and governance beyond de-identification alone [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eImplications for research, practice, and policy: For researchers, the findings underscore the need to report clinical AI interventions using AI‑specific extensions of reporting guidelines and to evaluate not only accuracy but also equity, usability, and unintended effects [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Independent external validation of widely implemented proprietary models has revealed limitations in generalizability and real‑world performance, reinforcing the need for transparency and independent evaluation prior to and after deployment [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. For clinical leaders, Clinical Cyberbioethics supports procurement and deployment checklists that make responsibilities explicit (who monitors drift, who audits equity, who responds to incidents) and that require governance artifacts (model cards, data documentation, and audit logs) as conditions of adoption [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. For policymakers and regulators, the framework suggests that compliance should incorporate measurable post‑market obligations\u0026mdash;monitoring, transparency artifacts, and accountability pathways\u0026mdash;rather than relying solely on pre‑deployment approval [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFuture directions: The next step is implementation validation. This includes developing and testing practical tools\u0026mdash;checklists, minimum dataset documentation, audit protocols, and patient information templates\u0026mdash;followed by prospective evaluation of impact on decision quality, clinician trust calibration, and patient‑centered outcomes [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Comparative studies across settings and populations are needed to refine equity‑sensitive governance and to reduce documented bias pathways in real‑world systems [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Finally, the framework should be iteratively updated as regulatory regimes, reporting standards, and clinical AI architectures evolve [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThis review has limitations. Identification relied on an author-curated reference set supplemented by backward reference checking; coverage may therefore be incomplete relative to a multi-database systematic search, and selection bias is possible. The corpus includes heterogeneous sources (empirical studies, reviews, normative analyses, and governance instruments), which supports conceptual and operational mapping but precludes quantitative synthesis. Quality appraisal was used to contextualize credibility rather than to exclude sources, and screening was conducted by a single reviewer; future updates should incorporate independent duplicate screening and full database exports. These limitations align with guidance on transparent reporting for evidence syntheses and synthesis-without-meta-analysis approaches (PRISMA 2020; SWiM) [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis mapping and narrative review with thematic synthesis (SWiM) (2015\u0026ndash;2025) suggests that classical biomedical ethics, cyberethics, and computational bioethics remain necessary but, when applied in isolation, are operationally insufficient for recurring dilemmas of AI-mediated clinical decision-making [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Across the included sources, ethical risk is driven by how algorithmic recommendations are embedded into workflows\u0026mdash;opacity at the point of care, automation bias and overreliance, uneven subgroup performance and bias, post-deployment drift, diffuse accountability, and privacy vulnerabilities in linked clinical data [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAlthough global and regional instruments (e.g., WHO guidance on AI in health, UNESCO\u0026rsquo;s Recommendation on the Ethics of AI, European trustworthy‑AI guidance, the EU AI Act, and the Council of Europe AI Convention) converge on high-level commitments such as transparency, accountability, safety, and equity, they remain under-specified for bedside implementation [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. In particular, the literature reveals persistent gaps around actionable thresholds (what counts as sufficient transparency), enforceable oversight (how override and contestability function under time pressure), auditable traceability (what must be logged and by whom), and the allocation of responsibilities across clinicians, institutions, and vendors; risk-management frameworks such as the NIST AI RMF and GMLP help translate these commitments into lifecycle controls, but clinical operationalization remains uneven [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThis article contributes by defining Clinical Cyberbioethics as an integrative, practice-oriented field at the intersection of clinical ethics, cyberethics, and computational bioethics, explicitly centered on AI-enabled and digitally mediated clinical authority. It further proposes (i) seven operational principles and (ii) five candidate fifth-generation digital rights for AI-mediated care, intended to translate normative commitments into patient protections and workflow-level duties [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Finally, the Deliberative Clinical Cyberbioethics Model (D-CCM) is presented as a pragmatic structure to embed ethical reasoning into routine care through: (a) transparency that is clinically meaningful, (b) meaningful human oversight and contestability, (c) equity monitoring and bias mitigation, (d) accountability engineering and traceability, and (e) proportional data governance and privacy safeguards [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe relevance of Clinical Cyberbioethics is threefold. First, it reframes ethical reasoning in digital healthcare by linking principles to the AI lifecycle (design\u0026ndash;validation\u0026ndash;deployment\u0026ndash;monitoring) and to real clinical constraints [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Second, it supports shared deliberation by clinicians and patients when AI influences diagnosis, prognosis, or treatment options, thereby strengthening autonomy and calibrated trust and reducing inappropriate reliance [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Third, it offers a coherent basis for institutional governance, training curricula, and policy design anchored in human-rights\u0026ndash;compatible safeguards [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFuture research should move beyond conceptual endorsement and empirically evaluate the D-CCM in real clinical environments using measurable outcomes: patient understanding and autonomy, calibrated reliance and reduced automation bias, safety events and workflow burden, subgroup equity metrics, traceability and accountability indicators, and patient trust [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Establishing feasibility, acceptability, and effectiveness across settings\u0026mdash;and aligning evaluations with emerging clinical AI reporting standards\u0026mdash;will determine whether Clinical Cyberbioethics can serve as a durable ethical infrastructure for responsible AI implementation in healthcare [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable. This study is a mapping and narrative review of published literature and did not involve human participants, individual-level human data, or animal subjects.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable. This manuscript does not report identifiable individual data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll information extracted and synthesized for this mapping and narrative review is reported in the main text and in Supplementary Files 1\u0026ndash;4. No new datasets were generated or analyzed beyond the cited sources.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author declares that he has no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research received no external funding.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eADP (Anderson D\u0026iacute;az P\u0026eacute;rez): Conceptualization; methodology; literature search; screening and selection; data extraction; synthesis and interpretation; drafting and critical revision of the manuscript; visualization (PRISMA flow materials); and approval of the final version.\u003c/p\u003e\n\u003cp\u003eWJAP (Wendy Johana Acu\u0026ntilde;a P\u0026eacute;rez): Conceptualization (public health and nursing perspective); interpretation and contextualization of findings; writing\u0026mdash;review and editing; and approval of the final version..\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author thanks academic colleagues and research groups engaged in ongoing discussions at the intersection of bioethics, clinical practice, and artificial intelligence, which informed the development of the concept of Clinical Cyberbioethics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAnderson D\u0026iacute;az P\u0026eacute;rez, PhD (Bioethics); MSc (Biomedical Basic Sciences); Specialist (Artificial Intelligence).\u003cbr\u003e Affiliation: Department of Social and Human Sciences, Universidad Sim\u0026oacute;n Bol\u0026iacute;var, Barranquilla, Colombia.\u003cbr\u003e ORCID: 0000-0003-2448-0953.\u003cbr\u003eEmail: [email protected].\u003c/p\u003e\n\u003cp\u003eWendy Johana Acu\u0026ntilde;a P\u0026eacute;rez, RN; MSc (Public Health).\u003c/p\u003e\n\u003cp\u003eAffiliation: Faculty of Health Sciences, Nursing Program, Corporaci\u0026oacute;n Universitaria Rafael N\u0026uacute;\u0026ntilde;ez, Barranquilla, Colombia.\u003c/p\u003e\n\u003cp\u003eORCID: 0000-0001-7755-8588.\u003c/p\u003e\n\u003cp\u003eEmail: [email protected]. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUse of artificial intelligence in writing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eArtificial intelligence tools (ChatGPT, OpenAI) were used to assist with language refinement and the drafting of preliminary text. The author critically reviewed, edited, and validated all content and takes full responsibility for the accuracy, originality, and integrity of the work. No AI tool was used to make decisions about eligibility, data extraction, or the interpretation of included evidence.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eTopol EJ (2019) High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25(1):44\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-018-0300-7\u003c/span\u003e\u003cspan address=\"10.1038/s41591-018-0300-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRajkomar A, Dean J, Kohane IS (2019) Machine learning in medicine. N Engl J Med 380(14):1347\u0026ndash;1358. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/NEJMra1814259\u003c/span\u003e\u003cspan address=\"10.1056/NEJMra1814259\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNagendran M, Chen Y, Lovejoy CA, Gordon AC, Komorowski M et al (2020) Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ 368:m689. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.m689\u003c/span\u003e\u003cspan address=\"10.1136/bmj.m689\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeauchamp TL, Childress JF (2019) Principles of Biomedical Ethics, 8th edn. Oxford University Press, New York\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEss C (2009) Digital Media Ethics, 2nd edn. Polity, Cambridge\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMittelstadt BD, Allo P, Taddeo M, Wachter S, Floridi L (2016) The ethics of algorithms: mapping the debate. Big Data Soc 3(2):1\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/2053951716679679\u003c/span\u003e\u003cspan address=\"10.1177/2053951716679679\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorley J, Machado CCV, Burr C, Cowls J, Joshi I, Taddeo M, Floridi L (2020) The ethics of AI in health care: a mapping review. Soc Sci Med 260:113172. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.socscimed.2020.113172\u003c/span\u003e\u003cspan address=\"10.1016/j.socscimed.2020.113172\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization (2021) Ethics and governance of artificial intelligence for health: WHO guidance. WHO, Geneva\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO (2021) Recommendation on the Ethics of Artificial Intelligence. UNESCO, Paris\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEuropean Commission High-Level Expert Group on AI (2019) Ethics Guidelines for Trustworthy AI. European Commission, Brussels\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eObermeyer Z, Powers B, Vogeli C, Mullainathan S (2019) Dissecting racial bias in an algorithm used to manage the health of populations. Science 366(6464):447\u0026ndash;453. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.aax2342\u003c/span\u003e\u003cspan address=\"10.1126/science.aax2342\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCross JL, Choma MA, Onofrey JA (2024) Bias in medical AI: implications for clinical decision-making. PLOS Digit Health 3(11):e0000651. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pdig.0000651\u003c/span\u003e\u003cspan address=\"10.1371/journal.pdig.0000651\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNazer LH et al (2023) Bias in artificial intelligence algorithms and recommendations for mitigation. PLOS Digit Health 2(7):e0000278. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pdig.0000278\u003c/span\u003e\u003cspan address=\"10.1371/journal.pdig.0000278\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoddard K, Roudsari A, Wyatt JC (2012) Automation bias: a systematic review of frequency, effect mediators, and mitigators. J Am Med Inf Assoc 19(1):121\u0026ndash;127. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/amiajnl-2011-000469\u003c/span\u003e\u003cspan address=\"10.1136/amiajnl-2011-000469\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdelwanis M, Alarafati HK, Tammam MMS, Simsekler MCE (2024) Exploring the risks of automation bias in healthcare artificial intelligence applications: a Bowtie analysis. J Saf Sci Resil 5:460\u0026ndash;469. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jnlssr.2024.06.001\u003c/span\u003e\u003cspan address=\"10.1016/j.jnlssr.2024.06.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJabbour S, Fouhey D, Shepard S et al (2023) Measuring the impact of AI in the diagnosis of hospitalized patients: a randomized clinical vignette survey study. JAMA 330(23):2275\u0026ndash;2284. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2023.22295\u003c/span\u003e\u003cspan address=\"10.1001/jama.2023.22295\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFinlayson SG, Bowers JD, Ito J, Zittrain JL, Beam AL, Kohane IS (2019) Adversarial attacks on medical machine learning. Science 363(6433):1287\u0026ndash;1289. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.aaw4399\u003c/span\u003e\u003cspan address=\"10.1126/science.aaw4399\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSahiner B et al (2023) Data drift in medical machine learning: implications and potential remedies. Br J Radiol 96:20220878. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1259/bjr.20220878\u003c/span\u003e\u003cspan address=\"10.1259/bjr.20220878\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolzinger A, Langs G, Denk H, Zatloukal K, M\u0026uuml;ller H (2019) Causability and explainability of artificial intelligence in medicine. Wiley Interdiscip Rev Data Min Knowl Discov 9(4):e1312. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/widm.1312\u003c/span\u003e\u003cspan address=\"10.1002/widm.1312\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFood and Drug Administration (FDA) (2021) Health Canada, MHRA. Good Machine Learning Practice for Medical Device Development: Guiding Principles\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNational Institute of Standards and Technology (NIST) (2023) Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100\u0026ndash;101\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu X, Cruz Rivera S, Moher D, Calvert MJ, Denniston AK et al (2020) Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. BMJ 370:m3164. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.m3164\u003c/span\u003e\u003cspan address=\"10.1136/bmj.m3164\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRivera SC, Liu X, Chan A-W, Denniston AK, Calvert MJ et al (2020) Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. BMJ 370:m3210. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.m3210\u003c/span\u003e\u003cspan address=\"10.1136/bmj.m3210\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins GS, Moons KGM, Dhiman P et al (2024) TRIPOD\u0026thinsp;+\u0026thinsp;AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385:e078378. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj-2023-078378\u003c/span\u003e\u003cspan address=\"10.1136/bmj-2023-078378\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSounderajah V, Guni A, Liu X, Collins GS, Karthikesalingam A et al (2025) The STARD-AI reporting guideline for diagnostic accuracy studies using artificial intelligence. Nat Med 31(10):3283\u0026ndash;3289. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-025-03953-8\u003c/span\u003e\u003cspan address=\"10.1038/s41591-025-03953-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLekadir K et al (2025) FUTURE-AI: international consensus guideline for trustworthy and deployable artificial intelligence in healthcare. BMJ 388:e081554. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj-2024-081554\u003c/span\u003e\u003cspan address=\"10.1136/bmj-2024-081554\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRegulation (EU) (2024) 2024/1689 of the European Parliament and of the Council: Artificial Intelligence Act\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKolbinger FR et al (2024) Reporting guidelines in medical artificial intelligence. Commun Med. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s43856-024-00492-0\u003c/span\u003e\u003cspan address=\"10.1038/s43856-024-00492-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePage MJ, McKenzie JE, Bossuyt PM et al (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.n71\u003c/span\u003e\u003cspan address=\"10.1136/bmj.n71\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCampbell M, McKenzie JE, Sowden A et al (2020) Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. BMJ 368:l6890. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.l6890\u003c/span\u003e\u003cspan address=\"10.1136/bmj.l6890\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCouncil of Europe (2024) Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law. Council of Europe\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVyas DA, Eisenstein LG, Jones DS (2020) Hidden in plain sight\u0026mdash;reconsidering the use of race correction in clinical algorithms. N Engl J Med 383(9):874\u0026ndash;882. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/NEJMms2004740\u003c/span\u003e\u003cspan address=\"10.1056/NEJMms2004740\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eInker LA, Eneanya ND, Coresh J et al (2021) New creatinine- and cystatin C\u0026ndash;based equations to estimate GFR without race. N Engl J Med. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/NEJMoa2102953\u003c/span\u003e\u003cspan address=\"10.1056/NEJMoa2102953\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWong A, Otles E, Donnelly JP et al (2021) External validation of a widely implemented proprietary sepsis prediction model in hospitalized patients. JAMA Intern Med\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhassemi M, Oakden-Rayner L, Beam AL (2021) The false hope of current approaches to explainable artificial intelligence in health care. Lancet Digit Health 3:e745\u0026ndash;e750. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2589-7500(21)00208-9\u003c/span\u003e\u003cspan address=\"10.1016/S2589-7500(21)00208-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRudin C (2019) Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat Mach Intell 1:206\u0026ndash;215. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s42256-019-0048-x\u003c/span\u003e\u003cspan address=\"10.1038/s42256-019-0048-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRibeiro MT, Singh S, Guestrin C (2016) Why should I trust you? Explaining the predictions of any classifier. In: Proc. KDD. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1145/2939672.2939778\u003c/span\u003e\u003cspan address=\"10.1145/2939672.2939778\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundberg SM, Lee SI (2017) A unified approach to interpreting model predictions. In: Advances in Neural Information Processing Systems (NeurIPS). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5555/3295222.3295230\u003c/span\u003e\u003cspan address=\"10.5555/3295222.3295230\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRocher L, Hendrickx JM, de Montjoye YA (2019) Estimating the success of re-identifications in incomplete datasets using generative models. Nat Commun 10:3069. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-019-10933-3\u003c/span\u003e\u003cspan address=\"10.1038/s41467-019-10933-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoons KGM, Altman DG, Reitsma JB et al (2015) Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 162:W1\u0026ndash;W73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7326/M14-0698\u003c/span\u003e\u003cspan address=\"10.7326/M14-0698\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCruz Rivera S, Liu X, Chan AW et al (2020) Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nat Med. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-020-1037-7\u003c/span\u003e\u003cspan address=\"10.1038/s41591-020-1037-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVasey B, Nagendran M, Campbell B et al (2022) Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. Nat Med. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-022-01772-9\u003c/span\u003e\u003cspan address=\"10.1038/s41591-022-01772-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Universidad Simón Bolívar","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Clinical Ethics, Artificial Intelligence, Bioethics, Clinical Decision-Making, Digital Health, Computational Bioethics, Patient Rights","lastPublishedDoi":"10.21203/rs.3.rs-8585342/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8585342/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Artificial intelligence (AI) is rapidly reshaping clinical decision-making, yet it introduces ethically salient risks including bias, automation bias, opacity, privacy threats, data drift, adversarial vulnerability, and unclear accountability. Objective: To define and operationalize Clinical Cyberbioethics as an integrative framework for AI-mediated clinical decision-making, synthesizing ethical principles, governance requirements, and patient-facing digital rights proposed across the literature.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMethods: We conducted a mapping and narrative review with SWiM-guided thematic synthesis covering January 2015 to March 2025, and documented identification, screening, and inclusion decisions using a PRISMA 2020–adapted flow diagram. Records were identified from an author-curated reference set compiled during the stated search period; screening and selection are reported in Fig.\u0026nbsp;1 and Supplementary Files 1–4.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResults: Records identified (n = 54); duplicates removed (n = 11); records screened (n = 43); records excluded at title/abstract (n = 1); full-text sources assessed for eligibility (n = 42); studies included in synthesis (n = 42); full-text exclusions with reasons (n = 0). The synthesis yielded operational domains spanning decision authority, explainability/contestability, bias and equity safeguards, lifecycle governance and accountability, and proportional data governance and privacy safeguards.\u003c/p\u003e\n\u003cp\u003eConclusions: Clinical Cyberbioethics provides a practical framework to guide trustworthy AI-mediated clinical decision-making by integrating ethical principles with governance and patient rights.\u003c/p\u003e","manuscriptTitle":"Clinical Cyberbioethics and AI-Mediated Clinical Decision-Making: a mapping and narrative review with thematic synthesis (SWiM)","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-16 11:10:53","doi":"10.21203/rs.3.rs-8585342/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"eb1faa0a-6146-4ad2-9e1e-5801e54fed8f","owner":[],"postedDate":"January 16th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":61024275,"name":"Medical Ethics"},{"id":61024276,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-01-16T11:10:53+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-16 11:10:53","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8585342","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8585342","identity":"rs-8585342","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}