A Hybrid System with Hidden Markov Models and Gaussian Mixture Models for Myocardial Infarction Classification with 12-Lead ECGs

Abstract

This study presented a new diagnosis system with integrating 12-lead ECG data into a density model for increasing accuracy rate and flexibility of diseases detection. A hybrid system with HMMs and GMMs was employed for data classification. For myocardial infarction, data of lead-V1, V2, V3 and V4 were selected and HMMs were used not only to find the ECG segmentations but also to calculate the log-likelihood value which was treated as statistical feature data of each heartbeatpsilas ECG complex. The 4-dimension feature vector was clustered by GMMs and different numbers of distribution (disease and normal data) were examined in experiment. The main idea in this study relied on the multiple ECG channels which could be combined. There were total 1129 samples of heartbeats from clinical data, including 582 data with myocardial infarction and 547 normal data. The sensitivity of this diagnosis system achieved 79% and predictivity achieved 68.70% statistically.