IA Clinique & Réglementation des Dispositifs Médicaux

Ce qu'elle couvre

Ce programme de niveau praticien dote les professionnels des affaires réglementaires et les équipes dispositifs médicaux d'une maîtrise approfondie du Règlement européen sur les dispositifs médicaux (MDR 2017/745), des voies FDA pour les logiciels en tant que dispositifs médicaux (SaMD), et des exigences du cycle de vie logiciel IEC 62304 appliquées aux systèmes IA/ML. Les participants travailleront sur des exercices réels de classification de dispositifs, d'adaptation des systèmes de management de la qualité (SMQ) aux modèles ML, et de planification de la surveillance après commercialisation pour les algorithmes en apprentissage continu. La formation associe des sessions en direct avec des experts à des études de cas structurées issues de dispositifs IA homologués CE ou approuvés FDA, et se conclut par la rédaction d'un document de stratégie réglementaire directement exploitable.

À l'issue, vous saurez

• Classify an AI/ML-based software product under EU MDR and determine the correct conformity assessment route, including Notified Body involvement
• Draft a Predetermined Change Control Plan (PCCP) aligned with FDA guidance for a continuously learning SaMD product
• Identify and document the gaps between an existing ISO 13485 QMS and the additional requirements imposed by AI/ML model management
• Design a post-market surveillance and clinical follow-up plan that accounts for model drift and algorithm updates
• Map dual-regulation obligations where both the EU AI Act and MDR apply to the same product, identifying which authority takes precedence

Sujets abordés

• EU MDR 2017/745: classification rules and conformity assessment routes for AI-based devices
• FDA SaMD pathway: 510(k), De Novo, PMA, and the Predetermined Change Control Plan (PCCP)
• IEC 62304 software lifecycle applied to ML model training, validation, and retraining
• AI Act intersection with MDR: dual-regulation obligations for high-risk AI systems
• Clinical evaluation and PMCF planning for adaptive algorithms
• Post-market surveillance and vigilance reporting for ML-based devices
• Quality Management System (ISO 13485) adaptations for data-driven products
• Change management and version control strategies for continuously learning models

Modalité

Delivered as a blended programme over 4–6 weeks: approximately 50% live virtual instructor-led sessions (half-day blocks) and 50% asynchronous case study work and reading. In-person intensives can be arranged for groups of 8 or more. Each module includes annotated regulatory document templates (technical file sections, PMCF plans, PCCP templates). A dedicated Q&A clinic with a regulatory consultant is included in the final week. All materials are updated annually to reflect evolving MDCG guidance and FDA Digital Health Center of Excellence publications.

Ce qui fait que ça marche

• Involve regulatory affairs, clinical, and ML engineering teams jointly in training to ensure shared vocabulary and aligned responsibility for compliance artefacts
• Apply learning immediately by working on a live or recent internal device project during case study sessions rather than hypothetical scenarios
• Establish internal change control procedures for model retraining before the product enters post-market phase, not after the first model update is needed
• Assign a designated regulatory owner for AI/ML product lines who maintains ongoing awareness of MDCG guidance updates and FDA Digital Health publications

Erreurs fréquentes

• Treating an AI/ML system as purely software and neglecting device classification rules, leading to incorrect conformity assessment routes and costly rework late in development
• Failing to define model change thresholds in advance, resulting in every algorithm update triggering a full re-submission rather than being covered by a PCCP or equivalent
• Underestimating post-market surveillance obligations for adaptive algorithms, particularly the need for real-world performance monitoring beyond traditional adverse event reporting
• Assuming CE marking under MDR automatically satisfies FDA SaMD requirements, and vice versa, without conducting a gap analysis between the two frameworks

Fournisseurs à considérer

• TOPRA (The Organisation for Professionals in Regulatory Affairs)www.topra.org →
• RAPS (Regulatory Affairs Professionals Society)www.raps.org →
• TÜV SÜD Academy – Medical Device Regulation Trainingwww.tuvsud.com/en/services/academy →
• BSI Group – Medical Devices Trainingwww.bsigroup.com/en-GB/medical-devices/training →

Sources

• EU MDR 2017/745 – Official Journal of the European Union →
• FDA Guidance: Marketing Submission Recommendations for a Predetermined Change Control Plan for AI/ML-Enabled Device Software Functions →