IoMT-Driven Non-Invasive Glucose Measurement Using Artificial Neural Networks

Abstract

Patients with Type 1 diabetes (diabetes mellitus) must frequently monitor their blood glucose levels to control their condition. This process becomes challenging due to the difficulties and discomfort caused by traditional blood glucose testing. To make this process more convenient and less time consuming, this study presents a non-invasive glucose monitoring system based on the Internet of Medical Things (IoMT) that offers a more user-friendly and painless alternative. The proposed system uses a light sensor connected to an ESP32 microcontroller to collect light intensity data from the user's fingertip. This data is transmitted to a remote server using FastAPI, where it is processed by a machine learning model using artificial neural networks (ANN). By analyzing the relationship between light absorption and glucose concentration, the ANN model estimates glucose levels, eliminating the need for invasive blood tests. This approach offers a pioneering alternative to traditional methods. Initial results demonstrate the system's real time glucose monitoring capability, although challenges such as sensitivity to external factors such as finger pressure are observed. These findings demonstrate the potential of integrating IoT technologies and machine learning to improve diabetes care by enabling more continuous, comfortable, and effective glucose monitoring. The proposed system in this study is a step forward in developing accessible and patient centered tools for diabetes management.