Wavelet based Bayesian models for characterizing chagasic high-resolution ECG signals

The study of high resolution ECG signals (HRECG) has proven useful in assessing the risk of arrhythmic events in patients suffering chagasic myocarditis. Various time and frequency domain criteria have been developed to evaluate abnormal potentials in chagasic patients, leading to the characterizations of the disease in terms of ECG features. Here, the authors present a Bayesian wavelet based modeling approach to analyze HRECG signals from 61 subjects classified into 3 groups: a control group of patients with no evidence of cardiac damage and two groups of chagasic patients with two different levels cardiac damage. The authors use a probability model that decomposes the expected value of each wavelet coefficient for a given signal in two term: a mean component common to all signals, and a term capturing group features. The model is able to find differences among the groups in terms of estimated posterior signals and posterior distributions of the wavelet coefficients