BiLSTM calibrated iGLU with demographic data: Non-invasive Glucose Measurement Device

Diabetes is regarded as a 21
^st
century crisis, with new cases more than doubling in the last decade. Diabetes affects nearly 530 million people worldwide, with three out of every four adults living in low and middle-income countries. In general, one out of every two adults with diabetes is undiagnosed. The paper presents the calibration of noninvasive glucose measurement device (iGLU) using BiLSTM (Bidirectional Long Short Term Memory) model. The acquisition module of iGLU has been designed to collect the data from real subjects. The demographic data has been considered along with NIR sensor values for the calibration purpose for a test population under different blood glucose conditions. The validation of the proposed model is carried out using Clarke Error Grid Analysis (CEGA). The performance of the model has been calculated using the different parameters like: Mean Absolute Relative Difference (mARD), Average Error (AvgE), mean absolute deviation (MAD), and Root Mean Square Error (RMSE). The proposed model has 2.21% mARD, 2.32% AvgE, 1.05% MAD and 10.32% RMSE. It has outperformed all the previous models and it is precise as well as is useful for clinical testing as per the CEGA standard.