Scopus Indexed Publications

Paper Details


Title
Early Diabetes Detection Using Health Indicators and Machine Learning Techniques

Author
, Most. Thahmina Akhter,

Email

Abstract

Early detection of diabetes is very important because it can help avoid serious health problems and lower the cost of medical care over time. Most current methods for predicting diabetes depend on lab tests, which take a lot of time and money, making them hard to use for early checks. This study suggests using machine learning (ML) to detect diabetes early by using information about a person's health and lifestyle. A large dataset containing 253,680 records was preprocessed through duplicate removal, outlier handling, feature scaling, and class balancing using the Synthetic Minority Oversampling Technique (SMOTE). Six machine learning models were evaluated using multiple performance metrics. The results show that models that combine several methods perform better than single models. The best model, which uses SMOTE and XGBoost (Extreme Gradient Boosting), reached an accuracy of 90.48%. An analysis of which factors matter most found that age, body weight, and high blood pressure are the top predictors. These findings demonstrate the feasibility of early diabetes prediction without reliance on laboratory-based diagnostic tests.


Keywords

Journal or Conference Name
Proceedings of the 3rd International Conference on Innovations in Cybersecurity and Data Science, ICICDS 2026

Publication Year
2026

Indexing
scopus