CWI and LUMC to develop explainable AI to detect hereditary cholesterol disease

Familial hypercholesterolemia (FH) is the most common inherited metabolic disorder worldwide, affecting 2.5 million Europeans, including an estimated 60,000 individuals in the Netherlands. While effective treatments exist, high-risk individuals often go undiagnosed. The FH-EARLY project aims to develop new strategies for early diagnosis and co-management of FH patients.

The project utilizes large biomolecular datasets (multiomics), explainable AI models, and co-creation with FH families and caregivers to:

• Enable faster, more affordable diagnoses;
• Map heart disease risks;
• Identify novel mechanisms underlying severe hypercholesterolemia.

CWI and LUMC researchers play a pivotal role in this initiative. Peter Bosman (CWI) and Tanja Alderliesten (LUMC) use explainable AI techniques to predict heart disease risk in FH patients. These models not only identify risk but also explain the underlying causes.

“Our models provide immediate insights from the data, potentially leading to novel questions or the need for more data,” says Bosman. These techniques were developed within their joint ICAI lab, ‘Explainable AI for Health.’

Read more about the project on the CWI website.