Fimea to use artificial intelligence to process adverse reaction reports

© GettyImages/AleksandarNakic

Fimea has looked into how to process reports on adverse reactions to medicines more efficiently using artificial intelligence. Developments in EU legislation and information systems together with the additional backlog due to the pandemic caused adverse reaction reports to pile up in the period 2020-2023. The estimated processing time for the 18 000 reports that have now been processed by marketing authorisation holders would be four years using conventional methods.  

In 2024 an updated national register of adverse reaction reports was introduced together with a user-friendly e-service, which has significantly improved the processing of reports. Now there is no longer any additional backlog of reports from marketing authorisation holders. 

To deal with the backlog effectively, Fimea explored the use of AI to help with the structured storage of medicines data. The investigation was based on a medicines data extraction process with reference to the information in the reports by marketing authorisation holders and the allocation of these medicines data to the medicines data extracted from Fimea’s basic register, either by means of automation or natural language processing.  

The results obtained in the test environment finally led to the decision to use the rules created with traditional programming rather than an AI language model. The deployment of the automation function is planned for early 2025. 

This development work is an example of the use of AI-based solutions in a way that focuses on everyday tasks in order to increase productivity. The use case selected is one where automation genuinely reduces the amount of work time spent by keyboarders through not having to repeat routine tasks. Fimea’s own information system development makes possible, and even requires, the development of the solution in collaboration with the system supplier itself, which will guarantee the best possible integration with the adverse reaction register.