The prediction of glucose concentration in blood plasma using artificial intelligence techniques

Abstract

During the last few decades diabetes has become a major newlinedisease worldwide and hence an increasing measure of attention has been paid newlineto it because of its social and economic implications Glucose sensors can newlineplay a crucial role for a better treatment of diabetes mellitus In particular newlineContinuous Glucose Monitoring Systems CGMS are of great interest for newlineseveral reasons such as retrospective tuning and optimization of diabetes newlinetherapy along with on newlineThe predictive monitoring is very much essential to have an early newlinewarning of the impending hypo hyper glycemia so that preventive measures newlinecould be applied to avoid diabetic complications However the true scenario newlineis that the accuracy of this prediction process with the existing CGMSs is only newline50 with the remaining that are false or missing predictions Lack of newlineadvanced denoising techniques and non inclusion of glucose variability newlinemeasures in the prediction methodologies could be the reason for the newlinesuboptimal performances of the earlier works The proposed research work newlinehad endeavored to enhance the accuracy of prediction models through an newlineadvanced denoising technique which improves the quality of input CGM newlinesensor data and with adaptive customized prediction models that includes newlinevariability features of Blood Glucose BG newline newline