Streamlining Medical Device Classification using Ontology : Automating MDR Annex VIII Compliance

This thesis explores the potential for automating the risk classification of medical devices under the European Medical Device Regulation (MDR) Annex VIII. The classification process, typically requiring expert regulatory knowledge, poses a significant challenge for start-ups and small manufacturers. To address this, the project first examined the MDR and the classification process in detail to identify key decision points and rule structures. Based on this analysis, a rule-based decision tree was implemented in Python, and an ontology was developed in Protégé using OWL and GCAs to structure and interpret classification logic. The decision tree offers a transparent and user-friendly pathway through MDR rules, while the ontology demonstrates the potential for representing semantic relationships within regulatory content. A case study evaluation reveals that both approaches are effective for well-defined scenarios but fall short in edge cases where interpretation beyond the text of Annex VIII is required, in this case, external guidance documents such as MEDDEV. These findings suggest that while partial automation is achievable, full automation is limited by the need for contextual interpretation and expert input. The project concludes that automated tools can support, but not replace, regulatory judgment. They offer value as educational and decision-support resources, particularly for non-expert users.