Detection of myocardial ischemia using hidden Markov models

The paper deals with detection of myocardial ischemia by analysis of electrophysiological changes within QRS complexes of electrocardiograms (ECG). ECG signals were analysed by continuous wavelet transform (CWT). Time-frequency spectra of QRS complexes were used as an input of a detection system based on hidden Markov models (HMMs). Parameters of the used HMMs were assessed to recommend their optimal values. The presented results show that HMM analysis of ECGs preprocessed by CWT can be used for early detection of myocardial ischemia. Eleven Langendorff perfused rabbit hearts were used to record training and test data to learn Markov models. An average value of resulting sensitivity and specificity of detection system was around 0.9 depending on parameter setting of the models.