Scholarlink

Background <p>Hyperglycemia, a persistent elevation of blood glucose levels, is a major indicator and complication of diabetes mellitus. According to the International Diabetes Federation, diabetes affects approximately 382 million people worldwide, with projections suggesting an increase to 592 million by 2035. This growing prevalence emphasizes the need for effective diagnostic and predictive tools to support clinical decision-making.</p> Objective <p>This study aims to develop and evaluate robust data-driven models for the prediction and classification of diabetes using advanced machine learning and deep learning techniques.</p> Methods <p>The research utilizes the Behavioral Risk Factor Surveillance System (BRFSS) dataset, which comprises data from over 400,000 adult respondents collected annually. Several supervised learning algorithms—decision tree, random forest, XGBoost, gradient boosting—and a deep learning model, the convolutional neural network (CNN), were implemented. Model performance was assessed based on key metrics including accuracy, precision, recall, and F1 score to determine the most effective predictive approach.</p> Results <p>Among the models evaluated, the convolutional neural network (CNN) demonstrated superior performance, achieving an accuracy of 93.34%, outperforming all other machine learning algorithms in terms of predictive capability and overall robustness.</p> Conclusion <p>The study highlights the potential of deep learning models, particularly CNNs, in improving diabetes prediction accuracy. The findings establish a strong foundation for developing reliable, data-driven diagnostic systems to assist healthcare professionals in early diabetes detection and patient management.</p>

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

A novel method for early detection and prediction of diabetes using various machine learning and deep learning models

  • 期刊论文

  • 发表日期：

    2025年12月10日

查看全文
  • Gothai Ekambaram,
  • Thamilselvan Rakkiannan,
  • Natesan Palanisamy,
  • Suresh Muthusamy,
  • Anuraj Singh,
  • Kishor Kumar Sadasivuni,
  • Mahendran Krishnamoorthy

摘要

Background

Hyperglycemia, a persistent elevation of blood glucose levels, is a major indicator and complication of diabetes mellitus. According to the International Diabetes Federation, diabetes affects approximately 382 million people worldwide, with projections suggesting an increase to 592 million by 2035. This growing prevalence emphasizes the need for effective diagnostic and predictive tools to support clinical decision-making.

Objective

This study aims to develop and evaluate robust data-driven models for the prediction and classification of diabetes using advanced machine learning and deep learning techniques.

Methods

The research utilizes the Behavioral Risk Factor Surveillance System (BRFSS) dataset, which comprises data from over 400,000 adult respondents collected annually. Several supervised learning algorithms—decision tree, random forest, XGBoost, gradient boosting—and a deep learning model, the convolutional neural network (CNN), were implemented. Model performance was assessed based on key metrics including accuracy, precision, recall, and F1 score to determine the most effective predictive approach.

Results

Among the models evaluated, the convolutional neural network (CNN) demonstrated superior performance, achieving an accuracy of 93.34%, outperforming all other machine learning algorithms in terms of predictive capability and overall robustness.

Conclusion

The study highlights the potential of deep learning models, particularly CNNs, in improving diabetes prediction accuracy. The findings establish a strong foundation for developing reliable, data-driven diagnostic systems to assist healthcare professionals in early diabetes detection and patient management.