AI-Based Decision Support for Personalized Drug Recommendations from a Medical and Clinical Pharmacological Perspective

Purpose and goal

The project aims to make accurate information for drug prescriptions more easily accessible and to develop support for efficient and safe drug treatment. The Swedish Medical Products Agency has developed an AI model that will be able to present information from national knowledge support compiled in an easily accessible way. The main objective of the project is to validate the model. Medical accuracy and patient safety need to be confirmed, before the model can be used in clinical practice.

Expected effects and result

Through testing and, if necessary, adaptation and repeated testing, a validated AI-based model that can be used for clinical drug questions is expected to be obtained. It can facilitate access to compiled, quality-assured information from national knowledge support sources and save time and reduce the risk of errors. More patients can be treated according to updated and evidence-based guidelines, resulting in more effective and safer drug treatment.

Planned approach and implementation

The project is carried out by Clinical Pharmacology at the Karolinska University Hospital and the Swedish Medical Products Agency. After development of validation protocols, the AI-model will be tested on simulated scenarios and adapted. When requirements are met, tests based on synthetic data/anonymized medical record data will be performed. The model´s performance will be reviewed by experts from clinical pharmacology. Results will be presented as a manuscript for a scientific journal.