Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms

Abstract Background Type 2 diabetes is the type with the largest proportion of people with diabetes.With the progression of the disease, patients with type 2 diabetes mellitus will have different degrees of complications, which will seriously reduce the quality of life of the patients and bring a heavy economic burden to the patient's families. Therefore, establishing a predictive model for glycemic control in patients with type 2 diabetes mellitus is of great help in optimizing the treatment of type 2 diabetes mellitus and delaying disease progression. Design and Methods: A retrospective study was conducted on type 2 diabetes mellitus real-world medical data from 4 cities in Sichuan Province, China from January 2015 to December 2020, including basic patient information, medication status, laboratory results, dietary habits, exercise status, and the actual follow-up of the patient after treatment. After data preprocessing, data inputting, data sampling, and feature screening, 16 kinds of machine learning methods were used to construct fasting blood glucose prediction models and glycated hemoglobin prediction models for type 2 diabetes mellitus patients, and 5 prediction models with the best prediction performance were screened respectively. Results A total of 375,723 cases of type 2 diabetes mellitus patients were collected, 10,000 cases were included to establish the fasting blood glucose model, and 2,169 cases were established to establish the HbA1c model. The best prediction model both of fasting blood glucose and HbA1c finally obtained are realized by ensemble learning and modified random forest inputting, the AUC value are 0.819 and 0.970, respectively. The most important indicators of the fasting blood glucose and glycated hemoglobin prediction model were fasting blood glucose and glycated hemoglobin. Medication compliance, follow-up outcome, dietary habits, BMI, and waist circumference also had a greater impact onfasting blood glucose levels. But on the glycated hemoglobin level, laboratory indicators such as platelets, Serum creatinine, Aspartate Transaminase, Hemoglobin, etc. had more impact. Conclusion The prediction accuracy of the models of the two blood glucose control indicators is high and has certain clinical applicability. Glycated hemoglobin and fasting blood glucose are mutually important predictors, and there is a close relationship between them.