IEEE EMBC 2024

Together with Nicolai Spicher and Tabea Steinbrinker (scientific research group Biosignal Processing), I represented the department of Medical Informatics of the University Medical Center at the EMBC 2024 in Orlando. With more than 2000 participants, the EMBC is one of the largest congresses in the field of biomedical engineering and medical informatics worldwide.

In the presentation on my publication ‘Enhancing Explainability in ECG Analysis through Evidence-Based AI Interpretability’, I presented an extended open-source framework for the diagnosis of cardiac arrhythmias such as first-degree AV block and left bundle branch block.

We were able to show that transparent AI systems can work with evidence-based parameters such as PR interval and QRS duration. For example, 96% of predictions for 1dAVb show a PR duration of more than 200 ms, while for LBBB 98.6% of predictions show a QRS duration of more than 120 ms. These results illustrate that the decisions of the neural network are largely consistent with clinical guidelines and can therefore provide valuable support for cardiologists.

These findings make a current scientific contribution, particularly with regard to the EU AI Act, in that we are helping to develop the basis for applying AI in patient care on a human scale.

In another presentation, my junior research group leader Nicolai Spicher was able to show that our open source frameworks in the field of ‘Explainable and Transparent AI Systems’ can be applied not only to cardiovascular diseases, but also to sleep research. Here, too, we were able to show that human and AI interpretation of brain signals can be harmonised in order to measure sleep quality in accordance with medical guidelines.

At her first international conference, my colleague Tabea Steinbrinker presented the results of a collaboration with Kana Eguchi from Kyoto University entitled ‘Benchmarking Open-Source Algorithms for QRS Detection and RRI Editing in Textile Electrocardiography’. There they were able to show that we are not only researching in the field of AI, but also study other topics of classical signal processing and wearables such as ‘ECG T-shirts’.

Participation in the EMBC also offered numerous opportunities for networking and exchange with international colleagues, which will promote future collaborations and further advance our research.