Defense of Mr. Hassan SERHAL thesis09h30 | Polytech Angers | Amphi E | 62 Notre Dame du Lac - 49000 Angers

The July 11, 2023

Subject : Contribution to the prediction, detection and classification of atrial fibrillation: processing of ECG signals using temporal and frequency methods and contribution of artificial intelligence.

Director of thesis : Ms. Anne HUMEAU-HEURTIER

Abstract

Artificial intelligence (AI) has become increasingly present in biomedical research, particularly in the prediction of atrial fibrillation (AF). In this thesis, we use AI models to classify signals with and without AF, as well as to predict AF early. New approaches are proposed to extract important features, and attention layers are used to normalize the data. Several publicly available electrocardiogram (ECG) signal databases are exploited, and a harmonization process is implemented to take into account different acquisition sources. ECG signal characteristics are analyzed from a variety of perspectives, including morphological, statistical, temporal, frequency and nonlinear analyses. The proposed AI models outperform existing techniques and show promising results for AF classification and early prediction.