RegTech Insights: Medical device manufacturers have high hopes for artificial intelligence in post-market surveillance

16. April 2021

RegTech Insights: Medical device manufacturers have high hopes for artificial intelligence in post-market surveillance

Artificial intelligence can be used to improve post-market surveillance for medical devices. That, at least, is the expectation expressed by medical technology manufacturers in an industry survey. Reduced effort, faster results and fewer errors – that is what artificial intelligence is supposed to make possible.

Suppliers of medical devices must monitor their safety. This is a clear requirement by the regulatory authorities. And the toolkit for this is what is known as post-market surveillance (PMS): Publicly available data about the manufacturer’s own products are collected, analyzed and evaluated by the manufacturer. The data are relevant for the risk assessment of a product and are used for potentially necessary product improvements.

Getting the right data is anything but trivial: Databases have to be searched, publications sifted through and specialist forums scoured. This is time-consuming and error-prone, as confirmed by the feedback received in an industry survey in medical technology. In as many as 82% of the companies, information is still searched for manually; more than three-quarters of the companies consider PMS to be burdensome; and more than half of the companies surveyed express dissatisfaction with current practices. Small and medium-sized companies in particular feel strongly challenged.

But if the search for information could be automated, if one could proactively receive notifications of new reports on one’s own products, and if trends for specific products or product groups could be determined across manufacturers, that would be a substantial relief for companies. And this is exactly what is possible with artificial intelligence methods and technologies.

Using a mix of technologies and methods including natural language processing, machine learning, deep learning and data analytics, publicly available databases – for example, the Manufacturer and User Facility Device Experience (MAUDE) database of the U.S. Food and Drug Administration (FDA) or the PubMed database for scientific literature in the life sciences and biomedicine – can be searched automatically. Detailed evaluations, notifications based on search profiles and trend determinations are also possible. This is currently being worked on in the “SmartVigilance” research project.

Partners in the “SmartVigilance” project are DHC Business Solutions, the German Research Center for Artificial Intelligence (Deutsche Forschungszentrum für Künstliche Intelligenz, DFKI) in Saarbrücken, and Freiburg-based Averbis GmbH, a specialist in text mining, machine learning and terminology management. SmartVigilance is funded by the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF), FKZ 01|S20028A. The industry survey was conducted as part of the project.


The survey results are summarized in a white paper and can be obtained here.