Wearable-band type visible-near infrared optical biosensor for non-invasive blood glucose monitoring

Abstract

Diabetes is a worldwide-serious problem that can only be delayed or prevented by a regular monitoring of blood glucose (BG) concentration level. Continuous monitoring systems allow subjects to prepare the diabetes management strategy and prevent the long-term complications diseases. Until now, most studies utilize various biofluids such as sweat, tears and saliva that have serious unresolved setback such as expensive material, sensor stability, sensor calibration and long-settling time. Therefore, we developed a novel BG sensor which is cost efficient and highly wearable with a small data acquisition time window that allow a non-invasive, long-term continuous blood glucose monitoring (CGM) system. The novel biosensor exploits a unique information of the pulsatile to continuous components of the arterial blood volume pulsation during the change of blood glucose (BG) concentration at the wrist tissue. The reflected optical signal was measured in the combine visible-near infrared (Vis-NIR) spectroscopy. An in-vivo experiment which enclosed 12 volunteers in a two-hour modified carbohydrate-rich meals reached the average correlation coefficient (Rp) between the estimated and reference BG concentration of 0.86, with the standard prediction error (SPE) of 6.16 mg/dl. Moreover, the full-day experiment was also conducted to test the reliability of the proposed sensor. Results showed that the created model in the previous day, may estimate a full-day BG concentration which was done in next day with an adequate performance.