Combined Approach to Estimate Blood Glucose Level in Noninvasive Monitoring: Ultra-Wide Band Microwave and Cascaded General Regression Neural Network

As the increasing number of diabetic around the world and the deficiencies of invasive blood glucose detection, such as prickly pain, risk of infection, and no continuous monitoring, a proper noninvasive blood glucose monitoring method is highly desired to deal better with it. In this article, an ultrawide band microwave based technique combined with a cascaded general regression neural network (C-GRNN) is proposed for blood glucose level estimation to achieve noninvasive and intelligent monitoring. Different glucose solutions are tested by using the reported monitoring system in which C-GRNN is used for intelligent detection and frequency choice. The proposed C-GRNN can reduce the prediction interval from 12.3858 to 2.1375 mg/dL compared with the conventional GRNN, and the root-mean-square error (RMSE) reaches 2.20 mg/dL. These results show good estimation performance. We also conduct a continuous-time blood glucose level monitoring on volunteer using the proposed noninvasive method. The final results show well accuracy in which the RMSE and mean absolute percentage error are 13.3208 mg/dL and 9.74%, respectively.