Analysis of Heart Rate Variability in Normal and Diabetic ECG signals using Fragmentation Approach

In this work, an attempt is made to quantify the dynamics of the heart rate variability timeseries in normal and diabetic population using fragmentation metrics. ECG signals recorded during deep breathing and head tilt up experiments are utilized for this study. The QRS-wave of ECG is extracted using the Pan Tompkins Algorithm. Heart rate variability features such as heart rate, Percentage of Inflection Points (PIP) and Inverse of the Average Length of the acceleration/deceleration Segment (IALS) are extracted to quantify the variation in signal dynamics. The results indicate that the ECG signals and heart rate variability signals obtained in deep breathing and tilt exhibit varied characteristics in both normal and diabetics. Further, in the diabetic condition the fragmentation measures exhibit a higher value in both deep breathing and tilt which indicates increased alternations in the signal. Most of the extracted fragmentation features are statistically significant (p<0.005) in differentiating normal and diabetic population. It appears that this method of analysis has potential towards the development of systems for the noninvasive assessment of diabetes.