| Popis: |
This paper presents the modeling approach artificial electrocardiograms (ECG) with T -wave alternans based on extracted parameters from the T -wave alternas (TWA) database. The developed optimal TWA classification system was evaluated by signals from a mixed database, which consisted of the TWA database and artificial ECG modeled records. F1-score about 95,9 % was received for the Random Forest Classifier (RFC) and Sequential Forward Floating Feature Selection method. Using a wrapper method for feature selection, 14 significant features were selected that associate with TWA. |
