Implementation of five machine learning methods to predict the 52-week blood glucose level in patients with type 2 diabetes

Q: What is the main objective of this study?

To use machine learning methods to predict blood glucose in patients with type 2 diabetes and determine the most effective algorithm for forecasting the 52-week trajectory.

Q: How were patients grouped for glycemic control?

Patients with glycosylated hemoglobin less than 6.5% after 52 weeks were considered to have achieved glycemic control; others were classified as not achieving it.

Q: Which machine learning method performed best and why?

XGBoost achieved the highest accuracy and AUC values, outperforming the other four compared algorithms, making it the recommended option for clinical prediction.

Aug 3, 2026

Upload by Jiven

For patients with type 2 diabetes, blood glucose trajectories are shaped by multiple interacting factors, making accurate prediction important for clinical decision-making. Using longitudinal glucose measurement as training data, this study evaluated predictors of glycemic control over 52 weeks and compared five machine learning algorithms for forecasting blood glucose outcomes. Among the methods, XGBoost showed the best performance for prediction accuracy and discriminative ability, supporting its use to assist clinicians and guide patient lifestyle choices.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Objective

Patients and methods

Results

Conclusion

Introduction

Epidemiology and burden of diabetes

Diagnostic criteria and pathophysiology

TYPE Original Research PUBLISHED 20 January 2023 DOI 10.3389/fendo.2022.1061507 OPEN ACCESS EDITED BY Huajin Wang, Carnegie Mellon University, United States REVIEWED BY Tadao Ooka, Massachusetts General Hospital, Harvard Medical School, United States Fei Wang, The Afﬁliated Hospital of Qingdao University, China Guifang Yan, Johns Hopkins Medicine, United States *CORRESPONDENCE Hongzhou Liu [Liuhongzhou@301hospital.com.cn](Liuhongzhou@301hospital.com.cn)[ ](Liuhongzhou@301hospital.com.cn)Shuangtong Yan [yanshuangtong@301hospital.com.cn](yanshuangtong@301hospital.com.cn) †These authors have contributed equally to this work ‡These authors share ﬁrst authorship SPECIALTY SECTION This article was submitted to Systems Endocrinology, a section of the journal Frontiers in Endocrinology RECEIVED 04 October 2022 ACCEPTED 30 December 2022 PUBLISHED 20 January 2023 CITATION Fu X, Wang Y, Cates RS, Li N, Liu J, Ke D, Liu J, Liu H and Yan S (2023) Implementation of ﬁve machine learning methods to predict the 52-week blood glucose level in patients with type 2 diabetes. Front. Endocrinol. 13:1061507 . doi: 10.3389/fendo.2022.1061507 COPYRIGHT © 2023 Fu, Wang, Cates, Li, Liu, Ke, Liu, Liu and Yan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Implementation of ﬁve machine learning methods to predict the 52-week blood glucose level inpatients with type 2 diabetes Xiaomin Fu 1‡, Yuhan Wang1‡, Ryan S. Cates 2, Nan Li 3, Jing Liu 4, Dianshan Ke 5, Jinghua Liu 6, Hongzhou Liu 1*† and Shuangtong Yan 3*† 1 Department of Endocrinology, The First Medical Center, Chinese PLA General Hospital, Beijing, China, 2 Department of Emergency Medicine Stanford Healthcare TriValley, Stanford University School of Medicine, Stanford, Pleasanton, CA, United States, 3 Department of Endocrinology, The Second Medical Center & National Clinical Research Center for Geriatric Diseases, Chinese PLA General Hospital, Beijing, China, 4Clinics of Cadre, Department of Outpatient, The First Medical Center, Chinese PLA General Hospital, Beijing, China, 5 Department of Orthopedics, Fujian Provincial Hospital, Fuzhou, China, 6 Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical University, Beijing, China Objective: For the patients who are suffering from type 2 diabetes, blood glucose level could be affected by multiple factors. An accurate estimation of the trajectory of blood glucose is crucial in clinical decision making. Frequent glucose measurement serves as a good source of data to train machine learning models for prediction purposes. This study aimed at using machine learning methods to predict blood glucose for type 2 diabetic patients. We investigated various parameters inﬂuencing blood glucose, as well as determined the most effective machine learning algorithm in predicting blood glucose. Patients and methods: 273 patients were recruited in this research. Several parameters such as age, diet, family history, BMI, alcohol intake, smoking status et al were analyzed. Patients who had glycosylated hemoglobin less than 6.5%after52 weeks were considered as having achieved glycemic control and the rest as not achieving it. Five machine learning methods (KNN algorithm, logistic regression algorithm, random forest algorithm, support vector machine, and XGBoost algorithm) were compared to evaluate their performances in prediction accuracy. R 3.6.3 and Python 3.12 were used in data analysis. Results: The statistical variables for which p< 0 . 05 was obtained were BMI, pulse, Na, Cl, AKP. Compa

FAQs

What is the main objective of this study?

To use machine learning methods to predict blood glucose in patients with type 2 diabetes and determine the most effective algorithm for forecasting the 52-week trajectory.

How were patients grouped for glycemic control?

Patients with glycosylated hemoglobin less than 6.5% after 52 weeks were considered to have achieved glycemic control; others were classified as not achieving it.

Which machine learning method performed best and why?

XGBoost achieved the highest accuracy and AUC values, outperforming the other four compared algorithms, making it the recommended option for clinical prediction.

Aug 3, 2026

Upload by Jiven