Publications

Abstract : Millions of people around the world are affected with diabetes and it is increasing day by day. Non invasive mode of glucose monitoring based on Near infrared (IR) spectroscopy using photoplethysmograph is a promising idea for frequent glucose monitoring and thereby decreasing the risk of diabetic complications. Present non invasive devices in the market are calibrated devices and are very costly. Also they have not looked at the diabetic correlation between IR and Red photoplethys-mography (PPG) signals. In this paper we have developed a glucose prediction algorithm for a non-calibrated non-invasive device that uses PPG signals for getting diabetic correlation. For this we studied the correlation of IR and Red signals with blood glucose and analysed its trend. We compared average amplitude as well as peak amplitude of IR and R signals using a dataset of 172 patients. Our results demonstrate that the current PPG device precision may not be able to pick up the minute variations in the amplitude caused by the change in blood glucose levels, and hence a more precise approach may be required.