Scholarlink

<p>Classifying Cardiovascular Diseases (CVDs) describes a crucial task in modern healthcare, as it remains a significant reason of death worldwide and often progresses without clear initial signs. Hence, this article performs the Cardiovascular Disease Classification (CVDC) model using intelligent deep learning framework. The dataset is first collected from standard benchmark sources like Kaggle named to be Automated Cardiac Diagnosis Challenge (ACDC) dataset. The pre-processing is next followed using the Synthetic Minority Oversampling Technique (SMOTE) approach. Next, the segmentation takes place with the help of 3D U-Net technique. The features are further extracted by the Wavelet Scattering Transform (WST) method. Finally, the classification of the proposed CVDC model is accomplished by the novel Enhanced Bidirectional Long Short Term Memory (EBiLSTM) model. The parameter tuning in BiLSTM is performed by nature inspired optimization algorithm known as Artificial Eagle Optimization Algorithm (AEOA). Accuracy maximization is considered to be the fitness function behind the entire procedure. The proposed EBiLSTM-AEOA is 10.34% and 8.65% better than the remaining considered methods for the suggested CVDC model in terms of accuracy and sensitivity, respectively.</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

Enhanced Bidirectional Long Short-Term Memory-based Optimization for Cardiovascular Disease Classification Model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月10日

查看全文
  • G Sekar,
  • T. Manojpraphakar,
  • P. Jayaselvi,
  • V. Murugan

摘要

Classifying Cardiovascular Diseases (CVDs) describes a crucial task in modern healthcare, as it remains a significant reason of death worldwide and often progresses without clear initial signs. Hence, this article performs the Cardiovascular Disease Classification (CVDC) model using intelligent deep learning framework. The dataset is first collected from standard benchmark sources like Kaggle named to be Automated Cardiac Diagnosis Challenge (ACDC) dataset. The pre-processing is next followed using the Synthetic Minority Oversampling Technique (SMOTE) approach. Next, the segmentation takes place with the help of 3D U-Net technique. The features are further extracted by the Wavelet Scattering Transform (WST) method. Finally, the classification of the proposed CVDC model is accomplished by the novel Enhanced Bidirectional Long Short Term Memory (EBiLSTM) model. The parameter tuning in BiLSTM is performed by nature inspired optimization algorithm known as Artificial Eagle Optimization Algorithm (AEOA). Accuracy maximization is considered to be the fitness function behind the entire procedure. The proposed EBiLSTM-AEOA is 10.34% and 8.65% better than the remaining considered methods for the suggested CVDC model in terms of accuracy and sensitivity, respectively.