Scholarlink

<p>Heart disease continues to be one of the primary causes of mortality worldwide, underscoring the importance of developing accurate and real-time prediction systems. This study proposes a novel IoT-enabled deep learning framework that combines multimodal health data and intelligent optimization techniques to enhance heart disease prediction. The model integrates both physiological signals and structured clinical features using a dual-path learning approach. Advanced preprocessing ensures feature alignment and noise reduction, while hierarchical feature extraction captures critical cardiac patterns. A lightweight optimization mechanism fine-tunes hyperparameters, boosting learning efficiency and overall predictive accuracy. Experimental evaluations demonstrate strong performance, achieving 99.9% accuracy, and 99.8% <i>F</i>1-score, and clearly outperforming conventional methods. The architecture is designed for compatibility with real-time, resource-constrained healthcare systems, making it ideal for mobile and remote monitoring environments. The results validate the framework’s potential to support early diagnosis and personalized treatment in smart healthcare 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

IoT-Driven Heart Disease Prediction Using Triple Pseudo-Siamese Network and Pyramid Attention with Termite Alate Optimization

  • 期刊论文

  • 发表日期：

    2025年9月15日

查看全文
  • R. Anitha,
  • Praveen Talari,
  • A. Babisha,
  • B. R. Tapas Bapu

摘要

Heart disease continues to be one of the primary causes of mortality worldwide, underscoring the importance of developing accurate and real-time prediction systems. This study proposes a novel IoT-enabled deep learning framework that combines multimodal health data and intelligent optimization techniques to enhance heart disease prediction. The model integrates both physiological signals and structured clinical features using a dual-path learning approach. Advanced preprocessing ensures feature alignment and noise reduction, while hierarchical feature extraction captures critical cardiac patterns. A lightweight optimization mechanism fine-tunes hyperparameters, boosting learning efficiency and overall predictive accuracy. Experimental evaluations demonstrate strong performance, achieving 99.9% accuracy, and 99.8% F1-score, and clearly outperforming conventional methods. The architecture is designed for compatibility with real-time, resource-constrained healthcare systems, making it ideal for mobile and remote monitoring environments. The results validate the framework’s potential to support early diagnosis and personalized treatment in smart healthcare applications.