Fortschritte in der EKG-Analyse mit Künstlicher Intelligenz: Nutzung von Transfer-Learning und Hybridmodellen zur Erkennung von Herz-Kreislauf-Erkrankungen
Abstract
This thesis develops innovative algorithms to improve AI performance in ECG analysis, focusing on transfer learning. A "Segment Label" feature is introduced using a CNN pre-trained on an additional database, enabling superior heartbeat classification. A hybrid model combining CNN and transformer enhances ECG classification and reduces computational demands. Additionally, a self-supervised approach significantly improves disease detection, especially atrial fibrillation. These advances highlight the potential of integrating AI into early CVD screening.
Fields of Study & Subjects
ECG Monitoring and Analysis
Artificial Intelligence in Healthcare
Non-Invasive Vital Sign Monitoring
Publication details
- Journal / source
- mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich)
- Publication date
- 19 January 2026
- Language
- EN
- Document type
- Thesis