MDI dose suggestion system based on Bayesian optimization

Abstract

The invention discloses an MDI dose suggestion system based on Bayesian optimization. The MDI dose suggestion system comprises an index calculation module, a Gaussian process learning module and a secure Bayesian optimization module, the index calculation module is used for recording the collected 24-hour blood glucose of a certain patient in the previous day, dividing the blood glucose in sections and converting the blood glucose into an asymmetric penalty index and a hypoglycemia safety index which can describe the blood glucose management level; the Gaussian process learning module takes the index result and the insulin dose as input training to obtain a prediction model, and obtains a prediction result corresponding to the dose by using the prediction model; the safety Bayesian optimization module calculates the maximum improvement expectation by using random distribution of the asymmetric penalty index prediction model obtained by prediction of the Gaussian process learning module, obtains a safety limiting function by using hypoglycemia safety index distribution, adds the safety limiting function as a risk penalty factor into the improvement expectation, and finally obtains a safety expectation improvement value; the blood glucose management level of the diabetic patient is evaluated through multiple indexes, and the optimal blood glucose control effect is gradually achieved.

Description

technical field [0001] The invention belongs to the technical field of insulin dosage adjustment algorithm design, in particular to an MDI dosage suggestion system based on Bayesian optimization. Background technique [0002] Insulin is an important part of the regulation of blood glucose metabolism in the body. Abnormal secretion or inability of insulin to function can lead to diabetes. Diabetic patients need to rely on drugs and exogenous insulin therapy. Currently, most insulin-treated patients take multiple daily insulin injections (Multiple Daily Injection, MDI) to maintain blood glucose levels. Patients using MDI therapy inject multiple injections of insulin every day, one injection of fast-acting insulin before meals to suppress the rise in blood sugar, and one injection of long-acting insulin before going to bed or in the morning to replace basal insulin. These four injections of insulin determine the blood sugar level, but it is difficult to adjust it in practice...

AI Technical Summary

Problems solved by technology

In this algorithm, the amplitude group still needs to be determined according to the patient's physical condition, and the impact of blood sugar fluctuations caused by environmental changes in daily life is not considered in the algorithm design

Images