Evaluation Frameworks for Clinical AI Incorporating Validation Strategies, Real-World Applicability, and Ethical Principles: Scoping Review

Clinical Snapshot

Key Findings

• P Value: Not reported; descriptive synthesis only

• Effect Size: Not applicable (scoping review; descriptive proportional findings): 88% of frameworks targeted investigational use; 11.4% achieved full methodological rigour; 31.8% reported technical performance metrics; 15.9% reported clinical performance indicators; 5 of 46 frameworks (10.9%) achieved ≥80% UNESCO ethical compliance; 4 frameworks scored <10% on ethical compliance

• Primary Outcome: Mapping and characterisation of 46 clinical AI evaluation frameworks across methodological rigour, validation strategies, and UNESCO ethical alignment, revealing a fragmented and predominantly investigational-use landscape with critical gaps in clinical validation and ethical integration

• Nnt Or Sensitivity: Not applicable to scoping review design. Within included frameworks: area under the curve, sensitivity, and specificity were the most commonly reported technical metrics (31.8% of frameworks); predictive values and calibration metrics were reported in only 15.9% of frameworks. Most frequently addressed UNESCO principles: awareness and education (71.1%) and transparency and explainability (70%). Least represented: human oversight (24.4%) and adaptive governance (33.3%)

• Confidence Interval: Not reported; no inferential statistics presented

Clinical Application

The review's findings are directly actionable for institutions developing internal AI governance frameworks. The UNESCO-based scoring matrix, if validated, could serve as a practical audit tool for ethics committees and digital health governance boards. However, the absence of a validated, standardised framework emerging from this review means clinicians and institutions must continue to navigate a fragmented landscape without a single endorsed standard Highly relevant to the Australian healthcare context. The Therapeutic Goods Administration (TGA) regulates software-as-a-medical-device (SaMD) under the Medical Devices framework, and the Australian Digital Health Agency (ADHA) has published AI in Health guidance. The finding that only 11.4% of reviewed frameworks achieve full methodological rigour, and that human oversight is addressed in only 24.4% of frameworks, directly informs TGA SaMD conformity assessment requirements and RACGP guidance on AI-assisted clinical decision support. The Australian Government's AI Ethics Framework (2019) and the National Health and Medical Research Council (NHMRC) ethical guidelines for AI in health align with the UNESCO principles assessed in this review. PBS and MBS implications are indirect but significant: AI tools influencing prescribing or diagnostic decisions require robust validation frameworks before integration into reimbursed care pathways. Australian health services considering AI procurement should use this review's gap map to interrogate vendor-supplied validation evidence against the dimensions identified as most frequently absent. Health informaticians, clinical AI developers, hospital digital health governance committees, health regulators, and clinical leaders responsible for evaluating, procuring, or implementing AI-based clinical decision support tools across any clinical specialty

Abstract

BACKGROUND: AI shows substantial potential in health care; however, the absence of standardized evaluation frameworks limits its safe and effective clinical implementation because of inconsistent validation requirements and fragmented ethical principles. Existing guidelines vary in structure, methodological rigor, and ethical integration, creating uncertainty. OBJECTIVE: This study aimed to systematically map, characterize, and critically analyze existing evaluation frameworks for clinical AI, focusing on three core dimensions: methodological rigor, validation strategies (internal validation, including reporting of technical and clinical performance; external validation, including real-world applicability), and alignment with the United Nations Educational, Scientific and Cultural Organization (UNESCO) AI ethical considerations. METHODS: A scoping review was conducted following PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. Six databases (PubMed, Embase, BVS, EBSCOhost, ProQuest, and Sage) and the Enhancing the Quality and Transparency of Health Research Network were searched without language or date restrictions up to February 2026. Eligible documents included peer-reviewed papers, gray literature, and organizational guidelines describing evaluation or reporting frameworks for clinical AI. Editorials, commentaries, and conference abstracts lacking a clearly defined evaluative framework or clinical applicability were excluded. Two reviewers independently screened records and extracted data. Data were extracted across three domains: (1) general characteristics, (2) methodological rigor and validation parameters, and (3) ethical integration and were synthesized using a dot plot-based gap map. Ethical adherence was assessed using a 10-domain UNESCO-based scoring matrix. No formal risk-of-bias assessment was conducted, consistent with scoping review methodology. RESULTS: From 3363 records, 46 frameworks met the inclusion criteria. Mapping revealed a rapidly expanding but fragmented landscape. Most frameworks targeted investigational use (88%), with limited focus on clinical applicability. Frameworks varied in structure, methodology, and scope, with a predominance of reporting guidelines and few validated tools. Most (63%) were developed through multi-institutional collaborations, and 32.6% incorporated transdisciplinary participation. Only 31.8% reported technical metrics (commonly area under the curve, sensitivity, and specificity), and 15.9% provided clinical indicators (eg, predictive values or calibration). Only 11.4% achieved methodological rigor, incorporating validation aligned with intended use, while most relied on partial validation strategies, highlighting a gap between model development and clinical evaluation. Ethical integration was heterogeneous: only 5 frameworks achieved high compliance (≥80%), whereas 4 scored <10%. The most frequently addressed UNESCO principles were awareness and education (71.1%) and transparency and explainability (70%), while human oversight (24.4%) and adaptive governance (33.3%) were least represented. Findings indicate a misalignment between framework design, validation requirements, and clinical implementation. CONCLUSIONS: Evaluation frameworks for clinical AI remain heterogeneous and oriented toward investigational contexts. Critical gaps persist in methodological rigor, validation aligned with intended use, and fragmented ethical coverage. These findings highlight the need for standardized, robust, and ethically grounded frameworks to enable safe, reliable, and scalable integration of AI into clinical practice.

References

1. 1.López Medina, D. C., Oliveros-Navarro, A., Moreno Angel, N., Herrera-Arellano, A. C., & Henao-Pérez, M. (2026). Evaluation frameworks for clinical AI incorporating validation strategies, real-world applicability, and ethical principles: Scoping review. Journal of Medical Internet Research. https://doi.org/10.2196/78168