Association of Blood Glucose Data with Physiological and Nutritional Data from Dietary Surveys Investigated Using Publicly Available Wearable-type Databases (Preprint)

BACKGROUND Wearable devices can simultaneously collect data on multiple items in real-time and are used for disease detection, prediction, diagnosis, and treatment decision-making. Many factors such as diet and exercise influence blood glucose levels; however, the relationship between blood glucose and these factors has not been evaluated in real practice. OBJECTIVE This study aimed to investigate the association of blood glucose data with each physiological index and nutritional values using wearable devices and dietary survey data from PhysioNet, a public database. METHODS Three analytical methods were used: 1) analysis of the correlation between each physiological indicator and blood glucose; 2) multiple regression analysis and 3)one-way analysis of variance on the pre- and post-peak slopes in postprandial glucose over time, and the association between each physiological indicator and nutritional value. RESULTS Three analytical methods were used: 1) analysis of the correlation between each physiological indicator and blood glucose and 2) multiple regression analysis and 3) one-way analysis of variance on the pre- and post-peak slopes in postprandial glucose over time and the association between each physiological indicator and nutritional value. CONCLUSIONS Similar results were obtained from the three analyses, consistent with previous reports, and the relationship between blood glucose, diet, and physiological indices in the real world was examined. Data sharing facilitates accessibility of wearable data and enables statistical analyses from various angles.