Physiological state characterization by clustering heart rate, heart rate variability and movement activity information

The objective is to identify whether it is possible to discriminate between normal and abnormal physiological state based on heart rate (HR), heart rate variability (HRV) and movement activity information in subjects with cardiovascular complications. HR, HRV and movement information were obtained from cardiac patients over a period of 6 weeks using an ambulatory activity and single lead ECG monitor. By applying k-means clustering on HR, HRV and movement information obtained from cardiac patients, we obtained 3 clusters in inactive state and one cluster in active state. Two clusters in inactive state characterized by - a) high HR and low HRV b) low HRV and low HR, could be inferred as pathological with abnormal autonomic function. Further, activity information was significant in differentiating between the normal cluster found in active and an abnormal cluster found in inactive states, both with low HRV. This indicates that the activity information must be taken into account while interpreting HR and HRV information.