Scholarlink

Heart attacks are among the leading causes of death worldwide. Early identification of cardiovascular risk factors and accurate prediction of heart attacks are critical for prevention and treatment. Recent research has shown that AI algorithms can analyze large medical datasets to predict the likelihood of heart attacks accurately. The aim of this study was to predict heart attacks using Machine Learning. A methodology consisting of 5 passes was applied: dataset acquisition; data normalization; feature selection (LightGBM, DT, RL); model implementation (RF, SVM, KNN, MLP, Gradient Boosting, AdaBoost, XGBoost and CatBoost) and performance evaluation. The best model resulted from the combination of the DT feature selection model and the KNN algorithm, whose metrics were superior to those of the other models with 95.2% Accuracy, 96.25% Precision, 95.85% Recall, 96.04% F1-score, and 98.12% ROC-AUC. In conclusion, the results demonstrate that combining feature selection methods and machine learning algorithms is highly effective for binary classification of medical data.

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 Robust Heart Attack Prediction Model Using Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Fabricio Gutiérrez,
  • Wilfredo Ticona

摘要

Heart attacks are among the leading causes of death worldwide. Early identification of cardiovascular risk factors and accurate prediction of heart attacks are critical for prevention and treatment. Recent research has shown that AI algorithms can analyze large medical datasets to predict the likelihood of heart attacks accurately. The aim of this study was to predict heart attacks using Machine Learning. A methodology consisting of 5 passes was applied: dataset acquisition; data normalization; feature selection (LightGBM, DT, RL); model implementation (RF, SVM, KNN, MLP, Gradient Boosting, AdaBoost, XGBoost and CatBoost) and performance evaluation. The best model resulted from the combination of the DT feature selection model and the KNN algorithm, whose metrics were superior to those of the other models with 95.2% Accuracy, 96.25% Precision, 95.85% Recall, 96.04% F1-score, and 98.12% ROC-AUC. In conclusion, the results demonstrate that combining feature selection methods and machine learning algorithms is highly effective for binary classification of medical data.