Three-Heartbeat Multilead ECG Recognition Method for Arrhythmia Classification

Electrocardiogram (ECG) is the primary basis for diagnosis of cardiovascular diseases. However, amount ECG data patients makes manual interpretation time-consuming and onerous. Therefore, intelligent recognition technology an important means to decrease shortage medical resources. This study proposes a novel classification method arrhythmia that uses very first time three-heartbeat multi-lead (THML) in which each fragment contains three complete heartbeat processes multiple leads. The THML pre-processing formulated use MIT-BIH database as training samples. Four models are constructed based on one-dimensional convolutional neural network (1D-CNN) combined with priority model integrated voting optimize effect. experiments followed recommended inter-patient scheme Association Advancement Medical Instrumentation (AAMI) recommendations, practicability effectiveness proved ablation experiments. Results show average accuracy N, V, S, F, Q classes 94.82%, 98.10%, 97.28%, 98.70%, 99.97%, respectively, positive predictive value F being 97.0%, 90.5%, 71.9%, 80.4%, respectively. Compared current studies, can effectively improve morphological integrity continuity information 1D-CNN sequence has higher classification. proposed alleviates problem insufficient samples, meets needs contributes dynamic research cardiac disease.