Continuous Glucose Monitoring: A Non-Invasive Approach for Improved Daily Healthcare

Diabetes mellitus is a prevalent health condition affecting approximately one in eight individuals worldwide. Insufficient management of this disease can lead to various medical complications, including vision loss and limb amputation. Continuous glucose monitoring (CGM) is crucial for effective diabetes management. However, current one-time glucose monitoring methods still rely on invasive techniques, causing pain and lacking continuous monitoring capability. This study aims to develop a non-invasive technology for measuring blood glucose levels that can be conveniently monitored through a smartphone using wireless communication. The device utilizes an ESP32 microprocessor and a PPG MAX 30102 sensor capable of recording pulse waves, a biomarker, at the fingertips. In this preliminary study, non-invasive data from 22 participants were used and divided into three classes according to conventional invasive measurements. Artificial intelligence (AI) methods, specifically Convolutional Neural Networks (CNN) and Artificial Neural Networks (ANN) models, were utilized to achieve convergence between the conventional invasive measurements and the proposed approach. The trained model demonstrated the highest accuracy of 0.944, with a minimum model loss of 0.177. In the validation phase, the proposed approach proved its reliability by achieving an average accuracy of 0.750. The non-invasive approach has the potential to provide a convenient and pain-free alternative to traditional monitoring methods, significantly enhancing diabetes management and improving the overall quality of life for those affected by this condition.