In medical treatment development, generative and explainable AI models offer great potential to accelerate research by creating diverse and synthetic data, especially of under-represented patient subpopulations to reduce dependency on patient data and animal testing. Explainable models can enhance transparency, increasing trust of clinicians, device developers, and even regulatory bodies. Furthermore, synthetic data overcomes privacy concerns associated with using real patient data.  

This project will also rigorously assess bias in the generated synthetic data and ensure transparent and reproducible validation while using synthetic data for developing and evaluating novel treatments. Addressing these challenges is especially relevant in the stroke context where new devices are rapidly transforming the treatment landscape. 

This project brings together various end-users to identify technical, cultural, and economic roadblocks in the current regulatory process of enabling the reduction of animal testing and the number of clinical trials. The project’s non-academic partner, Gravity Medical Technology, will provide treatment devices and perform preclinical flow model studies, based on the 3D-printed synthetic data generated from the models developed in the project. 

Project team: 

• Praneeta. R. Konduri (Amsterdam UMC, FdG)

• Henk Marquering (Amsterdam UMC, FdG)

• Erik Bekkers (FNWI) 

• Somendra Narayan (FEB) 

• Shashvat Desai (Gravity Medical Technology)