Parabolical Integration as Parameter to Detect Informative Components of Electrical Heart Signals

Study is aimed to develop parameter which will lead to accurate detection of ventricular late potentials (VLPs) in electrical heart signals. AUC40 parameter - area under the curve (filtered QRS-complex) of last 40 ms - is proposed to achieve the aim. Parabolical integration by Simpson's 1/3 rule is tested due to its accuracy compare to other methods of numerical integration. Cohorts of signals are analyzed by standard approach to detect VLPs and with AUC40 parameter. Generalized extreme value distributions are selected to calculate threshold value AUC40t due to less error compare to other distributions. As a result, AUC40t = 745.8 ms*μV. Probability of error at calculated value is 0.025. Originality of this study includes less error of AUC40 parameter by proposed method of integration compare to earlier results.