This work presents the design and validation of a cost-effective, compact, and rechargeable non-invasive glucose monitoring device that combines microwave sensing with machine learning techniques for continuous blood glucose estimation. The handheld system employs a patch antenna operating between 1.15–1.36 GHz to detect variations in signal reflection (S11 parameter) caused by glucose-induced dielectric changes in tissue. A multilayer perceptron (MLP) model, trained on over 60,000 frequency-response samples, was used to predict blood glucose levels (BGL). Experimental validation on diabetic and non-diabetic volunteers demonstrated mean predicted values of 141.11 mg/dL and 98.5 mg/dL, respectively, with strong correlation to invasive methods (R² = 0.7781) and a mean absolute relative difference (MARD) of 11.8%. In contrast to many available non-invasive systems that are often bulky, partially invasive, or expensive, this device integrates microwave sensing with deep learning in a compact, rechargeable design. It supports real-time monitoring and mobile app connectivity, offering pain-free glucose tracking and data management.
Keywords: Microwave Sensors, non-invasive glucose monitoring, rechargeable device, blood glucose levels, continuous monitoring.
How to cite this article: Salvekar SS, Ghongade R, Rechargeable Non-Invasive Glucose Monitoring Device. Int J Drug Deliv Technol. 2026;16(4s): 918-926; DOI: 10.25258/ijddt.16.4s.107
Source of support: Nil
Conflict of interest: None