A blood glucose prediction system by chaos approach

Abstract

For suppressing the development of diabetes mellitus and the onset of complications, an insulin therapy has been used for suppressing and normalizing the change of a blood glucose. In a blood glucose control by linear method such as conventional ARMA, however, there exists problem that results in the frequency of hypoglycemia. In a blood glucose prediction by a chaos theory, there also exists problem that results in the lower accuracy on behalf of the impossibility in the long-time prediction. For the improvement in the prediction accuracy of the blood glucose that looks like complicated time series, we propose a system combining the deterministic chaos theory using equal time interval, local fuzzy reconstruction method, and minimal linear model. By local fuzzy reconstruction method, we can predict the fasting blood glucose in the short term and then we can estimate the appropriate amount of insulin shot based on the measured bedtime blood glucose. Using the system, the change of blood glucose can be suppressed and normalized and the number of the insulin dosage a day can be reduced to once. Here we report the high effective result of applying the system to type II diabetes mellitus patient.

Hypoglycemia Prediction with Subject-Specific Recursive Time-Series Models

“…Other empirical models proposed in the literature may require additional inputs, such as food intake, physical condition information, or insulin infusion rate. 9,[18][19][20][21][22][23] We reported 3.83 ± 1.63% relative absolute deviation and accurate readings of 90% or more with CG-EGA on 14 ambulatory patients with diabetes when predicting 30 minutes into the future. 3 This work further evaluated the algorithm to predict hypoglycemia and provide early hypoglycemic alarms.…”

Adaptive system identification for estimating future glucose concentrations and hypoglycemia alarms

“…The first two approaches require manually entered inputs such as insulin infusion rate, meal information and insulin resistance. Similarly, other empirical models proposed in the literature require manual inputs (Bellazzi, 2000; Katayama, 2004; Yamaguchi, 2006), or training data for model development (Mougiakakou, 2000; Florian, 2005; Zainuddin, 2009). The algorithm discussed in this paper does not require any training data or manual inputs from patients.…”

Multiple Linear Regression and Artificial Neural Network to Predict Blood Glucose in Overweight Patients

“…It is inconvenient to monitor blood glucose plasma level, because it requires the patient fast for about 8 h prior to testing -this makes a novel, highly accurate fs-GLU prediction model very valuable. Several blood glucose prediction methods have been published including the response surface method [24], jump neural network method [25], and local fuzzy reconstruction method [26], which are used for type I and type II diabetes. We developed and tested a novel prediction model for fs-GLU in this study.…”