Post-Market Surveillance Strategies for Medical Devices

Authors

  • Elena Rossi Department of Computer Science, Advanced Computing University, Paris, France. Author
  • Isabella Silva Department of Machine Learning, Western Europe Data Science University, Madrid, Spain. Author
  • Eva Hansen Institute of Intelligent Systems, Central European Tech University, Vienna, Austria. Author

Keywords:

patient-reported outcomes, adverse event reporting, device registries, real-world evidence, EU MDR, signal detection, medical device safety, post-market surveillance

Abstract

Post-market surveillance (PMS) -- the systematic monitoring of medical device safety and performance after regulatory
approval -- is the cornerstone of medical device lifecycle safety management. The EU MDR (2017/745) dramatically
strengthened PMS requirements, mandating proactive surveillance plans, periodic safety update reports (PSURs),
post-market clinical follow-up (PMCF) studies, and trend reporting. Yet PMS implementation remains inconsistent:
adverse event reporting rates are estimated at 1-10% of actual events, signal detection from sparse data is statistically
underpowered, and the 3-5 year lag between device launch and signal emergence means that millions of patients may
be exposed before safety issues are identified. We present the Post-Market Surveillance Excellence Framework
(PMSEF), evaluating five PMS strategies -- passive adverse event reporting systems, active registry-based surveillance,
real-world evidence from electronic health records, AI-powered signal detection platforms, and patient-reported outcome
monitoring -- across four device risk categories (Class III implants, Class IIb active devices, Class IIa diagnostics, and
software as medical device). Our PMS Effectiveness Score (PMSES) measures signal detection sensitivity, detection
timeliness, evidence quality for regulatory action, patient coverage, and operational sustainability. AI-powered signal
detection achieves the highest PMSES (0.926) through multi-source data integration and statistical learning that detects
safety signals 8-14 months earlier than passive reporting, while registry-based surveillance achieves the highest
evidence quality (0.960) through prospective, structured data collection with long-term follow-up.

Author Biographies

  • Elena Rossi, Department of Computer Science, Advanced Computing University, Paris, France.

    Research Scientist, Department of Computer Science, Advanced Computing University, Paris, France. Email:elena.rossi963@ai-europe-research.org | ORCID: 4333-4695-6487-1124

  • Isabella Silva, Department of Machine Learning, Western Europe Data Science University, Madrid, Spain.

    Associate Professor, Department of Machine Learning, Western Europe Data Science University, Madrid, Spain. Email:isabella.silva325@ai-europe-research.org | ORCID: 4135-9902-8699-3336

  • Eva Hansen, Institute of Intelligent Systems, Central European Tech University, Vienna, Austria.

    Postdoctoral Researcher, Institute of Intelligent Systems, Central European Tech University, Vienna, Austria. Email:eva.hansen437@ai-europe-research.org | ORCID: 3846-8976-7814-7147

Downloads

Published

2022-08-15

How to Cite

Post-Market Surveillance Strategies for Medical Devices. (2022). The Biosis Bulletin: Bioscience and Information Science Journal , 3(2), 46-54. https://stanfordgroup.org/index.php/TBB-BIS/article/view/180