Scholarlink

<p>Today's leading cause of death is cardiovascular disease, which makes early and precise risk assessment essential for prompt intervention. This study integrates a hybrid deep learning architecture, improved feature selection, and enhanced data preparation to present a novel and intelligent approach for CVD risk prediction. Initially, entropy and Kendall-based preprocessing techniques are used to enhance data quality by eliminating noise, redundancy, and inconsistencies. For effective dimensionality reduction, a hybrid Artificial Hummingbird Algorithm with Energy Valley Optimizer (AHA-EVO) selects the most relevant features. The prediction task is performed using the Stereoscopic Scalable Quantum Multi-Relational Attention Network (SSQMRANet), a fusion of quantum convolutional neural networks and graph attention mechanisms that enables robust, scalable, and interpretable learning. The model's hyperparameters are also adjusted using the Osprey Optimization Algorithm (OOA). Experiments on a Kaggle-based CVD dataset with 70,000 records show excellent performance with 99.65% accuracy, 99.64% precision, 99.63% recall, and 99.62% F1 score. These results highlight the framework’s potential for high-accuracy, real-time CVD risk assessment, making it a valuable tool for clinical decision support systems.</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

Advanced Cardiovascular Disease Risk Prediction Via Stereoscopic Scalable Quantum Multi-relational Attention Network

  • 期刊论文

  • 发表日期：

    2025年8月7日

查看全文
  • Architha Keshavaraju,
  • Lalitha Tammabattula,
  • Jyothika Uppalapati,
  • Shaik Khasim Saheb

摘要

Today's leading cause of death is cardiovascular disease, which makes early and precise risk assessment essential for prompt intervention. This study integrates a hybrid deep learning architecture, improved feature selection, and enhanced data preparation to present a novel and intelligent approach for CVD risk prediction. Initially, entropy and Kendall-based preprocessing techniques are used to enhance data quality by eliminating noise, redundancy, and inconsistencies. For effective dimensionality reduction, a hybrid Artificial Hummingbird Algorithm with Energy Valley Optimizer (AHA-EVO) selects the most relevant features. The prediction task is performed using the Stereoscopic Scalable Quantum Multi-Relational Attention Network (SSQMRANet), a fusion of quantum convolutional neural networks and graph attention mechanisms that enables robust, scalable, and interpretable learning. The model's hyperparameters are also adjusted using the Osprey Optimization Algorithm (OOA). Experiments on a Kaggle-based CVD dataset with 70,000 records show excellent performance with 99.65% accuracy, 99.64% precision, 99.63% recall, and 99.62% F1 score. These results highlight the framework’s potential for high-accuracy, real-time CVD risk assessment, making it a valuable tool for clinical decision support systems.