A non-invasive approach for the diagnosis of Type 2 diabetes using HRV parameters

Abstract

As per the records of World Health Organisation, diabetes is currently one of the major diseases faced by the world community. This necessitates the introduction of screening tools for diabetes. In this paper, a non-invasive approach is proposed to diagnose the presence of Type 2 diabetes by analysing the relationship between the Heart Rate Variability (HRV) parameters and the arterial blood glucose changes. The HRV analysis is performed using non-linear methods such as Detrended Fluctuation Analysis (DFA) and Poincare plot. A few parameters derived from these non-linear methods are used to introduce two metrics named as Standard Deviation Ratio (SDR) and alpha-ratio. These two metrics are given as input to a machine learning classifier to categorise the subjects as diabetic or non-diabetic. The accuracy analysis of the classification results shows that 94.7% of the subjects are categorised correctly. Therefore, the proposed metrics can be considered as non-invasive screening tools in predicting the presence of Type 2 diabetes.