Medical Assistance Systems

29 September was World Heart Day. Cardiovascular diseases remain the leading cause of death in Germany. Many conditions could be treated more effectively if they were detected earlier and more reliably. This is exactly where some of our research comes in. In the Medical Assistance Systems research group at Bielefeld University, we are developing AI-supported assistance systems for medicine. We are particularly focused on two heart-related topics: atrial fibrillation and myocarditis in children.

Detecting atrial fibrillation earlier

Atrial fibrillation is the most common cardiac arrhythmia. Because it often occurs only temporarily, it frequently goes undetected for a long time. However, undiagnosed atrial fibrillation significantly increases the risk of a stroke. Together with the Department of Neurology at the Evangelisches Klinikum Bethel (EvKB) and as part of the FIND-AF2 study, we are developing deep learning methods that can detect signs of atrial fibrillation in ECGs that appear normal. The aim is to enable a faster diagnosis based on a routine examination rather than lengthy monitoring. The results are to be presented in such a way that doctors can understand and critically evaluate them.

For more information on the AF project.

Recognising myocarditis in children

Myocarditis is an inflammation of the heart muscle. Diagnosing it in children is complex and has so far relied heavily on criteria developed for adults. In an interdisciplinary collaborative project with the Heart and Diabetes Centre NRW, we are investigating whether artificial intelligence can be used to reliably diagnose myocarditis in children using MRI images. A key focus is on incorporating cardiac functional parameters. We are also using explainable AI (XAI). This makes the system’s decisions transparent and is intended to provide new insights into which features are truly relevant for diagnosis.

For more information on the myocarditis project.

AI that remains transparent

Both projects share a common goal: AI should support, not replace, healthcare professionals. To achieve this, it must be transparent and understandable. Explainable AI (XAI) is therefore a key focus of our research group, both in the field of cardiology and in our research into assistance systems and human-robot interaction. In the coming months, we want to further develop both areas. We very much look forward to exchanging ideas with hospitals, researchers and anyone else with an interest in this field.