Glucose Prediction Data Analytics for Diabetic Patients Monitoring

Diabetes Mellitus (DM) is one of the leading health complication around the world causing national economic burden and low quality of life. This increases the need to focus on prevention and early detection to improve the management and treatment of diabetes. The aim of this paper is to present a comprehensive critical review focusing on recent glucose prediction models and a best fit model is proposed based on the evaluation to perform data analytics in a wireless body area network system. The proposed glucose prediction algorithm is based on autoregressive (ARX) model which consider exogenous inputs such as CGM data, blood pressure (BP), total cholesterol (TC), low-density lipoprotein cholesterol (LDL), high density lipoproteins (HDL). A dataset of 442 diabetic patients is used to evaluate the performance of the algorithm through mean absolute error (MAE), root-mean-square error (RMSE), and coefficient of determination (R ² ). The experimental results demonstrate that the proposed prediction algorithm can improve the prediction accuracy of glucose. Potential research work and challenges are pointed out for further development of glucose prediction models.