Deep Learning-Coupled Metabolic Heat Integrated Sensing System for Noninvasive Continuous Monitoring of Blood Glucose

The dynamic and continuous monitoring of blood glucose (BG) concentration is crucial for the health management of diabetic patients. Despite its importance, significant challenges remain in the development of effective BG monitoring technologies. Metabolic heat conformation (MHC) offers a promising solution due to its noninvasiveness and reliability. However, progress in MHC technology is hindered by the complexities of multisensor integration and the intricate correlation between MHC and BG. Herein, a wearable BG continuous monitoring device based on metabolic heat integrated sensing (MHIS) is developed, combined with a deep learning network to enable continuous BG detection using MHC principles. The MHIS device integrates a miniaturized circuit and intelligent secondary signal processing, allowing for the simultaneous acquisition and integrated operation of various physiological parameters, including metabolic heat production. By integrating a gate recurrent unit neural network, a model is established to facilitate continuous BG monitoring. The wearable device has a certain accuracy, and when analyzed in comparison with commercial noninvasive glucose meters, the mean absolute relative error meets international standards. The deep learning-enhanced MHIS system proposed in this work enables noninvasive BG monitoring, paving the way for advancements in personalized healthcare management and offering new opportunities in digital healthcare consultation. A portable, wearable device based on metabolic heat integrated sensing and deep learning enables continuous blood glucose (BG) monitoring. The system uses a gate recurrent unit model for real-time BG prediction, achieving accuracy comparable to commercial noninvasive meters. This innovation presents new possibilities in personalized healthcare and digital health solutions.image (c) 2024 WILEY-VCH GmbH