Scholarlink

Diabetes affects millions worldwide, including over 30 million in India, necessitating reliable predictive models for early diagnosis and timely intervention. This study explores the application of machine learning (ML) algorithms for Type 2 diabetes prediction using classifiers such as K-Nearest Neighbors (KNN), Decision Tree (DT), Gradient Boosting (GB), LightGBM, Random Forest (RF), and an Ensemble Voting Classifier. The dataset underwent comprehensive preprocessing, missing value imputation, and class imbalance correction, to enhance model robustness. Among the classifiers, RF achieved the highest accuracy of 95.99%, while LightGBM demonstrated strong sensitivity at 95.70% during 5-fold cross-validation. The Ensemble model showcased consistent performance with an accuracy of 95.84% and an MCC of 0.92, highlighting its ability to combine the strengths of individual classifiers. These results emphasize the effectiveness of ensemble techniques and advanced ML algorithms in diabetes prediction, offering promising avenues for early detection and personalized healthcare interventions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning Models for Predicting Type 2 Diabetes Using Ensemble Techniques and Performance Evaluation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Prince Jain,
  • Anand Joshi

摘要

Diabetes affects millions worldwide, including over 30 million in India, necessitating reliable predictive models for early diagnosis and timely intervention. This study explores the application of machine learning (ML) algorithms for Type 2 diabetes prediction using classifiers such as K-Nearest Neighbors (KNN), Decision Tree (DT), Gradient Boosting (GB), LightGBM, Random Forest (RF), and an Ensemble Voting Classifier. The dataset underwent comprehensive preprocessing, missing value imputation, and class imbalance correction, to enhance model robustness. Among the classifiers, RF achieved the highest accuracy of 95.99%, while LightGBM demonstrated strong sensitivity at 95.70% during 5-fold cross-validation. The Ensemble model showcased consistent performance with an accuracy of 95.84% and an MCC of 0.92, highlighting its ability to combine the strengths of individual classifiers. These results emphasize the effectiveness of ensemble techniques and advanced ML algorithms in diabetes prediction, offering promising avenues for early detection and personalized healthcare interventions.