Proposing a Unified Classification System for AI-Based Medical Devices

Abstract

Background: The rapid integration of AI and Machine Learning (ML) into healthcare has created an urgent need for harmonized regulatory frameworks. Significant fragmentation persists across the United States (FDA), European Union (MDR), Japan (PMDA), Canada (Health Canada), United Kingdom (MHRA), Australia (TGA) and India (CDSCO), each maintaining distinct classification frameworks with varying risk stratification criteria. This fragmentation increases development costs, delays patient access, and creates competitive disadvantages for global manufacturers. Objective: To conduct a comparative analysis of AI medical device regulatory frameworks across seven major regulatory bodies, identify convergence opportunities, and propose a unified risk classification system that harmonizes disparate approaches while accommodating jurisdiction-specific requirements. Methods: A comparative legal and regulatory analysis was conducted through systematic review of primary regulatory documents and official publications from seven regulatory authorities. Four dimensions were examined: classification architecture, pre-market requirements, AI/ML-specific provisions, and post-market surveillance. The IMDRF SaMD risk categorization framework (N12) served as the foundational structure, integrated with practical requirements from each jurisdiction. Results: While all regulatory bodies have adopted risk-based classification, significant variations exist in the number of risk classes (3–4), classification criteria, and pathway requirements. We proposed a unified four-tier system- Class I (Low Risk) through Class IV (Critical Risk) based on information significance (inform, drive, diagnose/treat) and healthcare situation severity. Key convergence areas include Predetermined Change Control Plans (PCCPs) and Good Machine Learning Practice (GMLP) standards. Conclusion: The proposed framework provides a common language and decision matrix adaptable to local regulatory contexts while promoting international harmonization, reducing time-to-market, and supporting manufacturers, regulators, and healthcare stakeholders navigating the evolving AI medical device landscape.

Ähnliche Arbeiten