Abstract: Objective To summarize the recent research advances in post-market safety data sources and signal detection for medical devices so as to provide evidence for post-market regulation of medical devices. Methods Literature on post-market safety signal detection for medical devices was searched to keep track of the latest progress in post-market safety data sources and signal detection methods. Results and Conclusion With the rapid development of information technology, such real-world data as medical insurance claims, electronic health records, and registry data have become the main sources of data on post-market safety signal detection for medical devices. With the expansion of sources of data, multiple signal detection approaches have been developed and applied in post-market safety surveillance for medical devices. Social media data and data automatically collected by medical devices are also emerging sources of data. In the course of signal detection for medical devices, a variety of methods are used, including disproportionality methods, pharmacoepidemiology approaches, and machine learning methods. To reduce selection bias in observational studies, propensity scores are often applied in pharmacoepidemiology studies in order to evaluate the safety of medical devices more effectively.

Key words: medical device, active surveillance, post-market surveillance, signal detection, regulatory science, safety