Scholarlink

<p>One of the most prevalent diseases causing deaths worldwide is Heart Disease (HD). In the medical domain, the limited availability of patient data has created a growing demand for high-quality datasets. Modern machine learning techniques rely on large datasets to produce stable and reliable results. To address the issue of limited data, synthetic data generation has become increasingly important in healthcare applications. Generative Adversarial Networks (GANs) are commonly used to generate synthetic data that closely resembles the original data. In this study, we employed Generative AI-based synthetic data generation using GANs to increase the size of the dataset and enhance the performance of the HD prediction model. Additionally, we utilized meta-heuristic algorithms, including Harris Hawk Optimization for feature selection chosen for its effectiveness in high-dimensional datasets, faster convergence, and reduced risk of overfitting and Particle Swarm Optimization to optimize the weights of the Artificial Neural Network classifier due to its computational efficiency and robust global optimization capabilities. To ensure statistical robustness, tenfold stratified cross-validation was performed, with performance evaluated using standard metrics (accuracy, precision, recall, F1-score, and AUC) and reported as Mean ± Standard Deviation along with 95% confidence intervals. The model achieved an accuracy of 94.41% ± 0.56 (CI 93.97–94.85%) on the Cleveland dataset and 93.94% ± 0.41 (CI 93.61–94.27%) on the Statlog dataset. Paired t-tests further confirmed that the improvements in classification performance after GAN-based augmentation were statistically significant (<i>p</i> &lt; 0.05), demonstrating that GAN-based augmentation significantly improves both accuracy and stability in data-scarce medical applications.</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

Optimized Heart Disease Prediction Leveraging Generative AI and Metaheuristic Algorithms

  • 期刊论文

  • 发表日期：

    2025年11月20日

查看全文
  • Farzana Begum,
  • J. Arul Valan

摘要

One of the most prevalent diseases causing deaths worldwide is Heart Disease (HD). In the medical domain, the limited availability of patient data has created a growing demand for high-quality datasets. Modern machine learning techniques rely on large datasets to produce stable and reliable results. To address the issue of limited data, synthetic data generation has become increasingly important in healthcare applications. Generative Adversarial Networks (GANs) are commonly used to generate synthetic data that closely resembles the original data. In this study, we employed Generative AI-based synthetic data generation using GANs to increase the size of the dataset and enhance the performance of the HD prediction model. Additionally, we utilized meta-heuristic algorithms, including Harris Hawk Optimization for feature selection chosen for its effectiveness in high-dimensional datasets, faster convergence, and reduced risk of overfitting and Particle Swarm Optimization to optimize the weights of the Artificial Neural Network classifier due to its computational efficiency and robust global optimization capabilities. To ensure statistical robustness, tenfold stratified cross-validation was performed, with performance evaluated using standard metrics (accuracy, precision, recall, F1-score, and AUC) and reported as Mean ± Standard Deviation along with 95% confidence intervals. The model achieved an accuracy of 94.41% ± 0.56 (CI 93.97–94.85%) on the Cleveland dataset and 93.94% ± 0.41 (CI 93.61–94.27%) on the Statlog dataset. Paired t-tests further confirmed that the improvements in classification performance after GAN-based augmentation were statistically significant (p < 0.05), demonstrating that GAN-based augmentation significantly improves both accuracy and stability in data-scarce medical applications.