Real-time Non-invasive Blood Glucose Monitoring using Advanced Machine Learning Techniques

Abstract

When left untreated, diabetes, a chronic ailment that affects a vast number of people overall, might result in major unanticipated problems. The risk of complications can be completely reduced and considerable improvements can be achieved with early detection of diabetes. Recently, the use of wearable technology has emerged as a potential tool for diagnosing and checking illnesses. Smartwatches with bioactive sensors are perfect for diabetes screening because they can provide continuous, painless monitoring of bodily vitals. This paper suggests a methodology for building a hybrid AI model to detect the existence of diabetes using patient data. The system combines body vitals calculated using a smartwatch equipped with a bioactive sensor to provide accurate and continuous information on the wearer's health state. The hybrid model combines both deep learning and traditional AI computations to achieve a high level of accuracy while diagnosing diabetes. The framework collects data on many bodily parameters, including skin conductance, circulatory strain, and pulse — all of which are known to be strongly associated with diabetes. The acquired data is pre-processed before being utilized to create the hybrid model. The standard AI calculation is used to classify the information into diabetes or non-diabetic categories, while the profound learning calculation is used to eliminate important level highlights from the raw data. The hybrid approach combines the advantages of both deep learning and traditional AI to improve the accuracy of diabetes localization.