GlucoPredict: an ensemble machine learning method using lifestyle, health, and socioeconomic factors to predict and prevent diabetes risk.

Kanthasamy, Gowryshankary; Poravi, Guhanathan

Abstract

One of the most common chronic illnesses, diabetes affects millions of individuals annually and has a major negative economic impact. The two main characteristics of diabetes are either insufficient insulin production by the body or inefficient use of the insulin that is produced. People become less able to control blood glucose levels, which can lower their quality of life and shorten their life expectancy. During digestion, various foods are converted into sugars, which are then released into the bloodstream. The pancreas releases insulin in response to this. Insulin makes it possible for the body’s cells to use the blood sugars as fuel. Chronically elevated blood sugar levels in diabetics are linked to complications such as kidney disease, lower limb amputation, heart disease, and vision loss. Diabetes cannot be cured, but many patients can lessen its effects by adopting healthy eating habits, exercising, decreasing weight, and getting medical care. Predictive models for diabetes risk are crucial tools for the general population and public health officials since early diagnosis can result in lifestyle modifications and more successful treatment. Gradient Boosting, AdaBoost, XGBoost, LightGBM, Bagging Classifier, Extra Trees, Neural Network, Stacking Classifier, Random Forest, KNN, Decision Tree, Random Forest, and Logistic Regression were all tested in this study. Because of its highest accuracy, F1-Score, and ROC-AUC, gradient boosting is the best model. It also performs well in precision and recall, and it has good class differentiation.

Citation

KANTHASAMY, G. and PORAVI, G. 2025. GlucoPredict: an ensemble machine learning method using lifestyle, health, and socioeconomic factors to predict and prevent diabetes risk. In Arai, K. (eds.) Intelligent computing: proceedings of the 2025 Computing conference (CompCom 2025), 19-20 June 2025, London, UK. Lecture notes in networks and systems, 1424. Cham: Springer [online], 2, pages 345-360. Available from: https://doi.org/10.1007/978-3-031-92605-1_21

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2025 Computing conference (CompCom 2025)
• Start Date: Jun 19, 2025
• End Date: Jun 20, 2025
• Acceptance Date: Jun 19, 2025
• Online Publication Date: Jun 19, 2025
• Publication Date: Dec 31, 2025
• Deposit Date: Jul 21, 2025
• Publicly Available Date: Jun 20, 2026
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: 2
• Pages: 345-360
• Series Title: Lecture notes in networks and systems
• Series Number: 1424
• Series ISSN: 2367-3379; 2367-3389
• Book Title: Intelligent computing
• ISBN: 9783031926044
• DOI: https://doi.org/10.1007/978-3-031-92605-1_21
• Keywords: Diabetes prediction; Ensemble learning; Machine learning; Health factors; Socioeconomic factors; Lifestyle factors
• Public URL: https://rgu-repository.worktribe.com/output/2934819