Predictive Glucose Monitoring for People with Diabetes Using Wearable Sensors

Diabetes is a chronic non-communicable disease resulting from pancreatic inability to produce the hormone insulin, or a physiological cellular inability to use this hormone effectively. Insulin is responsible for maintaining biological homeostasis by enabling glucose to enter cells as their primary energy source. In the UK, 4.1 million people live with diabetes, while a further 850,000 are currently undiagnosed. Current global estimates identify 1 in 11 people as having diabetes. Unregulated glucose levels cause significant, and often irreversible, damage to blood vessels in the eyes, kidneys, teeth, and skin. Current means of glucose level monitoring range from infrequent capillary blood glucose sampling to continuous interstitial fluid glucose monitoring. While these methods can minimise their accuracy is limited by the cleanliness of skin, adequate hydration, certain medications and appropriate calibration methods.

A potential solution to this shortcoming involves the use of wearable sensors which record various information from an individual’s daily activities which have been shown in the medical literature to influence glucose levels and therein serve as potential predictors for estimating overall glucose level. Five features from the wearable device were applied in this work and include daily metrics such as; calories burned, number of steps taken, distance covered, minutes sedentary and activity calories. These features were in turn fused and post-processed with machine learning algorithms to provide a prediction of an individual’s glucose level and showed potential for being able to provide an Artificial Intelligence driven glucose monitoring platform.

In this paper we conduct a comparison case study involving the use of Quadratic Discriminant Analysis (QDA) and Support Vector Machine (SVM) algorithms for the classification of glucose level with data acquired from the wearable sensors of a Type 1 diabetic individual. Preliminary results demonstrate predicted glucose levels with >70% accuracy, providing that the potential for this approach to be used in in the design of an ergonomic glucose prediction platform utilising wearable sensors.

Further work will involve the exploration of additional datasets from affordable wearables to enhance and improve the prediction power of the machine learning algorithms.