An individualized decision-making system for preprandial insulin dose based on Gaussian process

Abstract

The present invention provides an individualized decision-making system for preprandial insulin doses based on Gaussian processes. Using artificial intelligence methods, the information contained in the patient's blood glucose monitoring and insulin infusion data is simultaneously mined to establish a postprandial blood glucose prediction model that conforms to the laws of human metabolism. Risk-sensitive optimization control idea, individualized determination of risk-sensitivity coefficients, and the use of Bayesian optimization methods to solve optimization problems, to achieve safe and effective pre-meal insulin doses and improve post-meal blood sugar management even when there are deviations in model predictions; It can be seen that, in order to improve postprandial blood glucose management, the present invention uses historical data of patients to design an individualized decision-making system for preprandial insulin dosage based on Gaussian process learning risk-sensitive control. The postprandial blood glucose prediction model facilitates subsequent implementation of postprandial blood glucose control or early warning of postprandial high and low blood sugar.

Description

technical field [0001] The invention belongs to the technical field of Gaussian process and risk-sensitive control, and in particular relates to an individualized decision-making system for preprandial insulin dosage based on Gaussian process. Background technique [0002] Insulin and glucagon work together to maintain normal blood sugar levels in the body. Diabetes mellitus results from defective insulin secretion or impaired biological action. Type 1 diabetes occurs when insulin is not secreted at all, and type 2 diabetes occurs when insufficient insulin is secreted. Both type 1 and type 2 diabetes can lead to acute symptoms such as hypoglycemia, as well as serious long-term complications such as cardiovascular disease and chronic kidney disease. According to statistics from the International Diabetes Federation in 2017, the number of patients in my country reached 114.4 million, making it the largest in the world. With the improvement of modern living standards in our ...

AI Technical Summary

Problems solved by technology

Although the patent uses the historical data of insulin infusion to adjust the dose of insulin at meal time, it ignores the data of blood glucose monitoring, and it is difficult to see the change pattern of the patient's blood glucose and the inter-individual differences in the metabolic process, which can be used to improve dosage decision-making

Images