Scholarlink

As heart disease continues to be one of the leading causes of death worldwide, the demand for accurate and rapid prediction tools for early detection and risk assessment is becoming increasingly urgent. This study introduces a novel approach that combines a fusion method of feature selection with an ensemble stacking technique for predicting heart disease. The approach leverages a comprehensive dataset that includes personal information, medical records, lifestyle factors, and clinical data. The hybrid feature selection method used in this study is pivotal as it identifies the most relevant features for machine learning models in predicting heart disease. The approach combines the strengths of two statistical techniques: analysis of variance (ANOVA) and the chi-square test. ANOVA is useful for continuous data analysis, while the chi-square method is particularly effective for handling categorical data. Together, these techniques help identify the most important subsets of attributes. In addition to selecting these key features, a stacking ensemble method is applied to further enhance prediction accuracy. This technique integrates multiple machine learning models, optimizing the use of hybrid features to achieve maximum performance. The result is a remarkable accuracy rate of 93.44%, demonstrating the effectiveness of this combined approach in improving cardiovascular disease prediction compared to previous methods.

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

Optimizing Heart Disease Prediction Using Fusion Feature Selection and Stacking Ensemble Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nureen Afiqah Mohd Zaini,
  • Mohd Khalid Awang,
  • Abd Rasid Mamat

摘要

As heart disease continues to be one of the leading causes of death worldwide, the demand for accurate and rapid prediction tools for early detection and risk assessment is becoming increasingly urgent. This study introduces a novel approach that combines a fusion method of feature selection with an ensemble stacking technique for predicting heart disease. The approach leverages a comprehensive dataset that includes personal information, medical records, lifestyle factors, and clinical data. The hybrid feature selection method used in this study is pivotal as it identifies the most relevant features for machine learning models in predicting heart disease. The approach combines the strengths of two statistical techniques: analysis of variance (ANOVA) and the chi-square test. ANOVA is useful for continuous data analysis, while the chi-square method is particularly effective for handling categorical data. Together, these techniques help identify the most important subsets of attributes. In addition to selecting these key features, a stacking ensemble method is applied to further enhance prediction accuracy. This technique integrates multiple machine learning models, optimizing the use of hybrid features to achieve maximum performance. The result is a remarkable accuracy rate of 93.44%, demonstrating the effectiveness of this combined approach in improving cardiovascular disease prediction compared to previous methods.