論文ID: 21-00076
Self-measurement of blood glucose is usually done using a drop of blood from the fingertip. Non-invasive methods have been studied vigorously, but have not been put into practical use because of difficulties resolving problems of portability, accuracy, and cost. We focused on the phenomenon that the pulse-wave form observed on the peripheral arteries is affected by food intake. The purpose of the present study was to analyze the relationship between blood glucose levels and the pulse-wave augmentation index (Aindex), index of the reflected wave in the artery, and provide a useful non-invasive method to estimate postprandial blood glucose levels. We developed the pressure-pulse-wave sensor using the gyro sensor for detecting the Aindex. Fifty-five diabetic (HbA1c > 6.5%) and five healthy subjects were participated in this study. The pressure-pulse-wave forms of the radial artery and the blood glucose levels were measured during the fasting time and 1 hour after taking the test meal. Regression analysis of Aindex and the blood glucose levels was performed with machine learning. Cluster analysis revealed that the Aindex after the test meal decreased in 36 subjects, increased in 8 subjects, and did not change in the remaining16 subjects. Regression analysis of pulse shape and blood glucose showed that the coefficient of variation (CV) was 11.5% for all data, and that the 87% data had an estimation error within ± 15%. We provided a new technology of measuring the pressure-pulse-wave based on the gyro sensor on radial artery to estimate blood glucose levels with machine learning. The methods are expected to be practically useful for detection of postprandial hyperglycemia and screening of diabetes mellitus.
