Linear and non-linear analysis of cardiac health in diabetic subjects

Diabetes is a chronic disease characterized by hyperglycaemia, which leads to specific long-term complications: retinopathy, neuropathy, nephropathy and cardiomyopathy. Analysis of cardiac health using heart rate variation (HRV) has become a popular method to assess the activities of the autonomic nervous system (ANS). It is beneficial in the assessment of cardiac abnormalities, because of its ability to capture fast fluctuations that may be an indication of sympathetic and vagal activity. This paper documents work on the analysis of both normal and diabetic heart rate signals using time domain, frequency domain and nonlinear techniques. The study is based on data from 15 patients with diabetes and 15 healthy volunteers. Our results show that non-linear analysis of HRV is superior compared to time and frequency methods. Non-linear parameters namely,correlation dimension (CD), approximate entropy (ApEn), sample entropy (SampEn) and recurrence plot properties (REC and DET), are clinically significant.