Scholarlink

<p>This study presents a non-invasive decision support tool for doctors to minimize cardiac consequences by presenting a Fuzzy Expert System optimized with hybrid evolutionary algorithms to predict the risk of heart disease in patients with left breast cancer following radiation therapy. The method employs four input variables blood pressure, cholesterol, blood sugar, and heart rate to categorize 380 patients from the University Medical Center Hamburg, Germany (32.1% male, 67.9% female, and 55.26% with heart disease) into low-risk and high-risk groups. Three hybrid models were created: Fuzzy-GA, which optimizes worldwide searches using a genetic algorithm; Fuzzy-DE, which optimizes locally effectively using differential evolution; and Fuzzy-GA-DE, which combines the two for optimal results. The models obtained AUCs of 97.93% Fuzzy-GA, 97.67% Fuzzy-DE, and 97.33% Fuzzy-GA-DE, which are significantly better than the baseline fuzzy system’s 85.52%, according to ROC curve analysis and 10-fold cross-validation. The system’s 92.35% accuracy and specificity suggest its usefulness in clinical settings to enhance patient outcomes and treatment planning. In order to broaden its therapeutic utility, future studies will verify the system on publicly available datasets, such the UCI Heart Disease dataset, and look into its suitability for different cancer types.</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

Optimizing cardiac radiation therapy in left breast cancer patients using an intelligent hybrid fuzzy model and differential genetic evolution

  • 期刊论文

  • 发表日期：

    2025年8月12日

查看全文
  • Pengwei Yan,
  • Yesong Guo,
  • Jing Wen,
  • Jingjing Han,
  • Yanxin Fan,
  • Yanhong Luo,
  • Pudong Qian,
  • Qian Zhang

摘要

This study presents a non-invasive decision support tool for doctors to minimize cardiac consequences by presenting a Fuzzy Expert System optimized with hybrid evolutionary algorithms to predict the risk of heart disease in patients with left breast cancer following radiation therapy. The method employs four input variables blood pressure, cholesterol, blood sugar, and heart rate to categorize 380 patients from the University Medical Center Hamburg, Germany (32.1% male, 67.9% female, and 55.26% with heart disease) into low-risk and high-risk groups. Three hybrid models were created: Fuzzy-GA, which optimizes worldwide searches using a genetic algorithm; Fuzzy-DE, which optimizes locally effectively using differential evolution; and Fuzzy-GA-DE, which combines the two for optimal results. The models obtained AUCs of 97.93% Fuzzy-GA, 97.67% Fuzzy-DE, and 97.33% Fuzzy-GA-DE, which are significantly better than the baseline fuzzy system’s 85.52%, according to ROC curve analysis and 10-fold cross-validation. The system’s 92.35% accuracy and specificity suggest its usefulness in clinical settings to enhance patient outcomes and treatment planning. In order to broaden its therapeutic utility, future studies will verify the system on publicly available datasets, such the UCI Heart Disease dataset, and look into its suitability for different cancer types.