Application of smart watch on blood sugar detection--I

Near-infrared non-invasive blood glucose detection technology

Because glucose has good infrared activity, we use the most promising near-infrared spectroscopy method to measure blood glucose, which has high sensitivity, high accuracy and develop-well.The design uses diffuse reflection for infrared spectroscopy analysis, which can overcome the strong absorption of infrared light on water-containing samples and reduce errors.Because glucose contains the absorption information of the combined and harmonic frequencies of hydrogen-containing groups such as C-H/N-H and O-H, it has good infrared activity.Experiments have shown that infrared light with wavelengths above 1800nm has a good activity on glucose, so we used near-infrared light with wavelengths of 1310nm and 1800nm for signal acquisition. 1310nm infrared light was used as the reference light.

In signal processing, signals within a specific frequency region are extracted to enhance the characteristic components of glucose.In data analysis, feature analysis is performed on the two wavelength data, and according to the blood sugar fluctuation pattern, features related to the blood sugar change pattern are extracted.Principal component analysis (PCA) was used to reduce the dimensionality of these variables to extract eigenvectors. Then, a nonlinear auto-regressive network (NARX)-based blood glucose non-detection model, namely the PCA-NARX model, was established with the eigenvariables as input variables and the measured blood glucose concentration as reference variables.

This enables non-invasive blood sugar testing, with a correlation coefficient of 0.89 between the model and blood sugar values.

With the development of AI technology and AI algorithms, blood glucose monitoring technology has evolved from traditional fingertip blood sampling to the current application of the seventh-generation blood glucose monitoring system on smart watches. This system will gradually mature in the next 5-10 years and achieve painless, non-sensing, and all-weather blood glucose management.At the same time, through the integration of other health data (such as electrocardiogram and blood oxygen, medical-grade products are already mature), smart wearable products will become digital health assistants for diabetic patients.