Filtering Continuous Glucose Monitoring Signal Using Savitzky-Golay Filter and Simple Multivariate Thresholding

Diabetes mellitus is a chronic disease that causes critical health complications like renal failure, heart disease, stroke, and blindness. For precluding the adverse effects of diabetes mellitus, humans use continuous glucose monitoring systems (CGMS) that represent a method of measuring blood glucose concentrations under real-life conditions. Glucose monitoring is an essential component of diabetes care but there are some limitations on accuracy. Accuracy possibly is restricted owing to manufacturing variances, storage, and aging. Because of their limitations on the environment, like temperature or elevation or to patient factors such as incorrect coding, wrong hand washing, unchangeable hematocrit, or inherently resulting interfering substances. Furthermore, exogenous interfering substances may also contribute errors to the system evaluation of blood glucose, as delay time, random fluctuations and noise concerned with sensor physics and chemistry. This paper suggests De-noising methods Savitzky-Golay Filter with Simple Multivariate Thresholding methods to remove all types of noise in CGM signal. This work has been commented with simulated data received from Glucosim that is an educational software package that simulates blood glucose and insulin dynamics in healthy individuals and patients with type 1 diabetes and approved with the Peak signal to noise ratio.