Non-invasive Blood Glucose Estimation Methodology Using Predictive Glucose Homeostasis Models

Over the last few years, a series of glucose bio-sensors have been developed that have reported various techniques for monitoring and estimation of blood glucose levels. However, despite their advancements, there has been little progress in the development of non-invasive blood glucose monitoring or estimation, due to complications associated with different technologies. In this paper, we propose a glucose estimation methodology which attempts to replace the invasive sensing mechanism of a glucose sensor with a custom-designed and non-invasive system having inherent re-configurability and adaptability towards dietary habits. To the best of our knowledge, this is the first work of its kind, eliminating the need of a body worn sensors for estimation of glucose level. Physiology of glucose homeostasis is involved in the development of this methodology, which considers the impact of glycemic index and glycemic load with an aim towards getting accurate glucose predictions. Our extensive validations demonstrate that the proposed methodology is capable of accurately estimating blood glucose levels with various customization features, suitable for integration with bionic pancreas systems.