Non-Invasive SPO2 & Blood Glucose Measurement Using Near Infrared Spectroscopy (NIRS)

Abstract

Diabetes mellitus (DM) is a chronic disease in which a body fails to produce sufficient amount of insulin required by the body. It is increasing across the globe at an unprecedented pace and has become a serious health concern these days. The conventional devices available in the market are invasive, pain causing, and require puncturing of skin every single time a person needs to take a glucose reading. Moreover, they are expensive because every time a new test strip and needle are required to check the glucose level. Diabetes is a cureless disease till the date. The only thing diabetic patients can do is to continuously monitor their blood glucose level and take insulin accordingly. Previously, much research has been done in the field of non-invasive glucose measurement by implementing various techniques. However, all previously invented glucometers suffer from inaccuracies due to the relatively weak absorption bands of the glucose in the near infra-red spectrum. They were not proven to be practical, precise, clinically approved and/or economically viable. All of them were either for investigational purpose or market awareness. That is why there was a need to propose a non-invasive approach to deal with all said issues. The proposed prototype is based on NIRS which undergoes photoplethysmography (PPG) and double regression analysis (DRA). DRA helped to increase the accuracy, overcome the deviations and get more reliable results. Near infra-red spectroscopy (NIRS) is most famous because it allows minimum attenuation while measuring the blood glucose level. NIRS helps to deal with the tissues having low absorption energy and permits glucose measurement up to few millimeters depth under the skin. In order to validate the results, a 3 day clinical trial is conducted, to perform in vivo analysis, in Holy Family Hospital, Pakistan, and a total of 132 diabetic and non-diabetic test specimens are analyzed. Non-invasive (NI) system’s results are compared with the invasive Beckman Coulter AU-480 chemistry analyzer. The Clarke error grid (CEG) analysis of all 3 day results yields that 98.48% of the results fall in the clinically accepted zone A and the mean and median absolute relative differences (ARDs) values are 8.25% and 7.94%, respectively. The coefficient of determination 𝑅2, depicts the goodness of a fit that how close the predicted values are to the reference glucose values. After the implementation of double regression model (DRM), the coefficient of determination 𝑅2 gets improve to 0.9471.