Characterization of ventricular tachycardia and ventricular fibrillation using semantic mining

Abstract

Ventricular tachycardia and ventricular fibrillation are ventricular cardiac arrhythmia that could be calamitous and life threatening. The ability to provide accurate predictions of ventricular tachycardia or ventricular fibrillation events can save lives. The purpose of this work was to investigate the possibility of using a semantic mining algorithm to predict the onset of ventricular tachycardia and ventricular fibrillation in electrocardiogram (ECG) signals. A total of eighteen subjects were obtained from Creighton University Ventricular Tachyarrhythmia Database and MIT-BIH Arrhythmia Database. Normal patient ventricular tachycardia, ventricular tachycardia, ventricular fibrillation for the same subject were classified based on annotations supplied by specialists at the Creighton University Cardiac Center. Based on these downloaded data damping ratios, natural frequencies and input parameters were developed using Semantic mining algorithm. The average value of the three developed parameters was determined. These average values were tabulated in sequence with time. Based on true observation, three ratios were taken: first, between the derivative of input parameter and natural frequency; second, between the input parameter and damping ratio; and third, between natural frequency and damping ratio. These ratios are characterized as new parameters and depending on the maximum amplitude for these new parameters, a threshold value is selected to predict the onset of ventricular tachycardia and ventricular fibrillation. In this study, it was found that the new parameters had different amplitude patterns with time according to conditions for the same subject, and the same patterns emerged for the same condition among different subjects. In addition, applying the selected threshold achieved successful result was one to four minute in the forecasting the onset of both Ventricular tachycardia and ventricular fibrillation. In brief this work provides a new method for advanced researches in distinguish and predict of heart disease.