The cardiovascular diseases are major cause of mortality across the globe contributing to 31% of global burden of mortality. In clinical practice, 2-5 minutes electrocardiography (ECG) recordings are used to detect abnormalities in heart rhythms during the onset of cardiovascular disease when abnormal events have already made significant damage to physiological component. The rhythms of heart start to change long before the onset of disease. Therefore, it is important to detect abnormal heart rhythm before any damage has been made. To detect such abnormality in the cardiac rhythms well in time, long-term ambulatory ECG (AECG) recordings of 24 to 48 hours are required. It is difficult for human to analyze these long-term recordings. To resolve this problem, researchers used both linear and nonlinear computational techniques to perform heart rate variability (HRV) analysis. Due to non-linear nature of heart rate dynamics, linear techniques have limited capability to detect these abnormalities. Among numerous nonlinear approaches, visibility graph-based complexity measures have shown promising result to extract dynamical information encoded in the biological signals. We developed a novel complexity analysis measure named as grouped horizontal visibility graph entropy (GHE) that is able to quantify the dynamical information encoded in the synthetic and biological signals of healthy and pathological subjects. Preliminary results revealed that proposed method is more accurate in detecting dynamical fluctuations occurring due to disease, aging, disease severity and activity level (sleep and wake periods). This work was accepted and published in the IEEE Access journal and find the article from here.
For experiments and analysis purpose we used publicly available data-sets are downloadable from physionet database.
If you are bit curious or beginner then you can read more about the complexity and variability from here.