REG-TRACE

REG-TRACE (Regulatory-Compliant Artificial Intelligence Development and Certification Framework for Healthcare) aims to be a large-scale, multinational R&D project addressing the main regulatory, technical, and operational barriers hindering the adoption of artificial intelligence (AI) and data-driven software in healthcare. The project aims to ensure the efficient, reliable, and regulatory-compliant development of AI-based medical device software (MDSW) within the increasingly complex European regulatory environment, including the MDR, IVDR, GDPR, AI Act, European Health Data Space (EHDS), and Cybersecurity Act.

Although AI has high potential to improve the quality, efficiency, and sustainability of healthcare services, the number of AI-based solutions that successfully reach clinical application and the market is limited. Key barriers include fragmented and conflicting regulations, lengthy and costly certification processes, limited access to high-quality health data, and insufficient transparency, traceability, and explainability of AI models. REG-TRACE directly addresses these challenges by providing an integrated framework that encompasses regulatory compliance throughout the entire AI development lifecycle.

At the heart of the project is the CARE-Framework (Certification of AI in the Regulatory Environment), a modular, extensible ecosystem of methods and tools designed to support all stages of AI-based medical software development. The CARE-Framework covers early requirements analysis, regulatory and ethical assessment, data collection and preparation, AI model development, product integration, certification, and post-marketing surveillance. It combines general modules (e.g., regulatory text analysis, quality and risk management, ethics and data protection, compliant documentation creation) with use-case-specific modules tailored to different AI application areas such as medical imaging, time series analysis, and clinical decision support.

REG-TRACE advances the concept of Secure Processing Environments (SPE) by offering an advanced AI development environment (ASPE) that enables privacy-preserving, traceable, and reproducible machine learning workflows. This environment is designed to comply with EHDS requirements and supports advanced approaches such as federated and swarm learning, enabling AI models to be trained on sensitive health data without direct data sharing. Integrated quality gates, version control mechanisms, and automated pipelines ensure regulatory compliance throughout the development process.

The CARE-Framework and related tools are validated through various real-world healthcare use cases, including health risk prediction, predictive care recommendations, acute patient monitoring, statistical shape modeling for implant design, AI-enhanced musculoskeletal assessment, AI-focused cardiovascular disease management, live image analysis in minimally invasive interventions, ophthalmic diagnosis, and AI-based clinical decision support. These use cases ensure that the framework is both technically robust and practically applicable in numerous clinical and market contexts.

REG-TRACE is expected to significantly reduce the time and cost required for MDR certification, improve documentation quality, increase certification success rates in the initial phase, and lower entry barriers for SMEs developing AI-based medical software. By creating a harmonized, regulation-compliant AI development framework, the project strengthens Europe's digital health ecosystem, accelerates the adoption of innovation, and contributes to safer, more efficient, and more reliable AI solutions in healthcare.