Fimea
New AI application to support reporting of adverse reactions
The Finnish Medicines Agency Fimea maintains a register of adverse reactions, among other things. In Finland, anyone can report an adverse reaction. Solita has developed an AI solution for Fimea to streamline the processing of adverse drug reaction reports. The new AI implementation helps authorities in their decision-making processes by significantly reducing the amount of time spent on information searches. Adding this feature promotes high-quality and up-to-date reporting that benefits international drug safety.
It was great to see how AI can be applied effectively and responsibly in a regulatory environment.
Kari Salmela Researcher and Product Owner of the Register, Fimea
-
A significant amount of time is saved when searching for information, making comparisons and decisions
-
The AI application improves quality and leads to more consistent reporting
-
Drug safety is improved as the solution provides information to an international network of authorities and the pharmaceutical industry
Solita’s strong and practical AI expertise was evident throughout the development process and helped bring the solution into production quickly.
Joonas Tuominen Head of Technology and Information Security, Fimea
How can AI make daily lives safer?
The application leverages a public, comprehensive, and up-to-date database of known medicine properties and utilises information extracted from reports in a way that keeps personal data and other sensitive information secure.
The solution doesn’t make decisions but supports the processor, whose task is to assess whether the adverse reaction described in the report is known or not. The application supports decision-making by performing intelligent searches in an extensive database of drug descriptions, significantly reducing the need for manual data retrieval. In addition to the search results, the application displays the document references on which the results are based.
The application is based on an architecture developed by Solita, which performs intelligent information searches across multiple sources. Retrieval Augmented Generation (RAG) technology is used for information retrieval and enrichment. This technology enables the use of public and Fimea’s own data in information retrieval, utilising the abilities of large language models (LLMs) to produce comprehensible text summaries from the source material searched. The solution was implemented as a modern cloud service.
There are usually two key challenges when using large language models. They excel at general language but lack the mastery of specialised professional terminology. In addition, the results of language models are often not very transparent.
Solita’s approach solves these challenges. The AI solution securely combines adverse reaction reporting data and publicly available information on drug properties in a way that meets regulatory requirements.
Sami Koskivaara Account Director, Solita
High-quality data behind the AI application
When advanced technology is combined with reliable information and expertise, it can create truly impactful solutions, supporting critical functions in society – such as this one, which enhances patient safety and aids in the assessment of drug safety throughout its use and development.
Personal, confidential material and publicly available information can be combined securely and flexibly, saving experts a significant amount of time.
“We have extensive experience as a provider of practical and reliable solutions for processing and utilising data,” says Koskivaara from Solita.
“AI helps experts find new, unknown side effects by comparing them to the known side effects listed in package inserts. The AI solution was built using Amazon Bedrock, Textract, and the Claude model”, tells Tuominen from Fimea and continues:
The experiment was highly successful, and the AI implementation is now integrated as a permanent part of the adverse reaction system’s production environment.
Joonas Tuominen Head of Technology and Information Security, Fimea
See more of our work
-
Fepod Next generation blood analytics boosted by AIFepod is developing a point-of-care solution for measuring the blood concentration of pain medicines. We helped Fepod to strengthen their artificial intelligence and machine learning capabilities needed for analytics. -
Istekki Three wellbeing services counties implemented a common data pool solution – for the benefit of the taxpayerIstekki, Solita and Productivity Leap implemented a Social and Health Services Data Pool solution, which can be replicated in other wellbeing counties, saving ten times as much time, money and work. -
Fimea Updating Finnish pharmaceutical dataTogether with Fimea, we modernised the register of medicinal products and marketing authorisations. As a result, Fimea now has an up-to-date and easy-to-use register with correct information on all medicines in Finland.
-
Sami Koskivaara Account Director, Solita