This project, carried out in cooperation with the Harvard Medical School, Boston (USA), investigates the use of federated and privacy-friendly machine learning for computer-aided stratification of pre-eclampsia patients in multicenter studies.

Pre-eclampsia is one of the most common and serious pregnancy complications worldwide, affecting over 70,000 mothers and 500,000 infants annually. The disease is associated with a high risk of maternal mortality, premature birth and long-term health consequences for mother and child.

Since pre-eclampsia has different clinical subtypes and its course varies greatly, precise risk stratification is essential. Current predictive models based on ultrasound and biomarker analyzes are expensive, require specialized personnel and often demonstrate low external validity. The project aims to develop a federated AI system that analyzes medical data from different hospitals without the need to centrally store or share sensitive patient data. This enables secure, data protection-compliant, cross-border collaboration.

Using AI to identify high-risk patients could enable early interventions, reduce maternal and infant mortality, and reduce premature births. The project is also intended to serve as a model for the safe use of AI in medicine and pave the way for future personalized therapy approaches.
 

Contact: Mohammad Tajabadi

Funding reference number: HE 6220/13-1