Non invasive blood glucose monitoring from human facial skin using wavelet based deep learning models

Abstract

newline Diabetes mellitus (DM) is a metabolic disorder based on how the body handles glucose levels in the human blood. DM are classified as (i).Type 1 (ii). Type 2 and (iii) Gestational. Diabetes has a systemic impact on the skin. The largest body surface is the skin, which is comprised of nerves and blood vessels. Elevated blood glucose levels impact the minor capillaries that comprise the blood vessels in the skin region. Diabetes measured through invasive and non-invasive methods. In invasive method blood from human body is pricked and measure the diabetic value at laboratory instruments. The non-invasive methods are continuous glucose monitoring and fingerstick blood glucose testing. However, the casting non-invasive methods have a number of drawbacks such as (i). A CGM device is costly. (ii). Predictions are inaccurate due to environmental factors and human error during measurement. To solve above problems, Type 1 and Type 2 diabetes are measured from human facial skin region. Facial skin regions have dermatological changes due to elevated glucose levels or even prior to the diagnosis of diabetes, and the changes are termed as skin diseases such as Vitiligo, Bell Palsy, Scleroderma, and Rubeosis Faciei.