Scholarlink

Heart disease prediction is a significant challenge in medical science, necessitating robust and interpretable machine learning models. This study utilized tree-based boosting techniques, specifically XGBoost, CatBoost, and LightGBM (LGBM), to develop predictive models for heart disease detection using a cardiovascular dataset. Preprocessing steps included missing value imputation using K-nearest neighbors (kNN) and normalization for optimal model performance. Recursive feature elimination (RFE) was employed for feature selection, enhancing model efficiency. Model performance was assessed through training and testing splits with 5-fold cross-validation. The results indicated that LGBM outperformed the other model, achieving a testing accuracy of 96.55%, precision of 96.73%, recall of 96.82%, and F1-score of 96.68%. These findings suggest that tree-based boosting models are effective for heart disease prediction, enhancing clinical decision-making.

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

Predicting Cardiovascular Disease Risk Using Tree-Based Gradient Boosting Machine Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nikhat Raza Khan,
  • Shekhar Verma,
  • Hemant Kumar,
  • Mamata Mayee Panda,
  • Abhishek Dwivedi,
  • Abhishek Kumar Mishra

摘要

Heart disease prediction is a significant challenge in medical science, necessitating robust and interpretable machine learning models. This study utilized tree-based boosting techniques, specifically XGBoost, CatBoost, and LightGBM (LGBM), to develop predictive models for heart disease detection using a cardiovascular dataset. Preprocessing steps included missing value imputation using K-nearest neighbors (kNN) and normalization for optimal model performance. Recursive feature elimination (RFE) was employed for feature selection, enhancing model efficiency. Model performance was assessed through training and testing splits with 5-fold cross-validation. The results indicated that LGBM outperformed the other model, achieving a testing accuracy of 96.55%, precision of 96.73%, recall of 96.82%, and F1-score of 96.68%. These findings suggest that tree-based boosting models are effective for heart disease prediction, enhancing clinical decision-making.