An IoT enabled multi sensor platform for noninvasive blood glucose monitoring via breath analysis.

Chronic metabolic diabetes mellitus requires timely and consistent blood glucose assessment to reduce long-term complications, yet invasive finger-prick techniques often discourage regular screening and follow-up. Volatile organic compounds (VOCs) in exhaled breath have therefore been explored as non-invasive metabolic indicators of altered glucose regulation, although existing breath-based approaches frequently show limited accuracy and reduced sensitivity to prediabetic states. To address these limitations, we present GlucoSense, a low-cost, Internet of Things (IoT)-enabled, non-invasive screening and monitoring-support platform that integrates breath VOC sensing with routinely measured physiological parameters, including heart rate, blood pressure, oxygen saturation, age, and gender. The system combines a six-sensor metal-oxide-semiconductor (MOS) array, signal preprocessing and feature engineering, cloud-connected hardware, and a multimodal machine learning pipeline to estimate both categorical glucose severity and continuous blood glucose levels. Importantly, GlucoSense is designed to complement invasive glucose measurements rather than replace standard clinical diagnostics. Evaluated on a clinically acquired dataset of 488 participants, GlucoSense achieved a mean absolute error of 17.12mg/dL for numerical glucose estimation and a balanced accuracy of 89% for classifying glucose severity (non-diabetic, prediabetic, and diabetic). Clarke Error Grid Analysis showed that 100% of predictions fall within clinically acceptable Zones A and B, indicating low potential risk for inappropriate clinical decision-making. Overall, the results demonstrate the feasibility of an affordable, interpretable, and scalable screening-support and monitoring-support system for glucose assessment in resource-constrained clinical settings, while explicitly maintaining invasive blood glucose measurements as the clinical reference standard.