Abstract

Type 1 diabetes (T1D) is a chronic disease that requires patients to know the blood glucose values to ensure their normal levels. The methods for predicting blood glucose are one of the areas of interest for clinical researchers. The literature describes many methods of predicting blood glucose, which require the determination of many activities, such as the time of insulin injection or emotional factors that may be susceptible to errors. To reduce the impact of individual activities, continuous glucose monitoring (CGM) is proposed to predict the level of glucose in the blood regardless of other factors. The paper presents the comparison of machine learning methods such as: linear regression, vector (SVR) and selected evolutionary algorithms (AE) to predict glucose concentration. The proposed methods based on SVR, and AE algorithms have achieved high accuracy and quality of predicted results..

Keywords:Machine learning; Evolutionary algorithms; Glucose concentration