An Automatic Patient-Adapted ECG Heartbeat Classifier Allowing Expert Assistance

In this paper, we present a patient-adaptable algorithm for ECG heartbeat classification, based on a previously developed automatic classifier and a clustering algorithm. Both classifier and clustering algorithms include features from the RR interval series and morphology descriptors calculated from the wavelet transform. Integrating the decisions of both classifiers, the presented algorithm can work either automatically or with several degrees of assistance. The algorithm was comprehensively evaluated in several ECG databases for comparison purposes. Even in the fully automatic mode, the algorithm slightly improved the performance figures of the original automatic classifier; just with less than two manually annotated heartbeats (MAHB) per recording, the algorithm obtained a mean improvement for all databases of 6.9% in accuracy A, of 6.5% in global sensitivity S and of 8.9% in global positive predictive value P(+). An assistance of just 12 MAHB per recording resulted in a mean improvement of 13.1% in A, of 13.9% in S, and of 36.1% in P(+). For the assisted mode, the algorithm outperformed other state-of-the-art classifiers with less expert annotation effort. The results presented in this paper represent an improvement in the field of automatic and patient-adaptable heartbeats classification, concluding that the performance of an automatic classifier can be improved with an efficient handling of the expert assistance.