Scholarlink

<p>Rapid and precise myocardial infarction (MI) prediction is essential to avert life-threatening outcomes. Traditional ECG diagnostic methods often falter due to signal noise, artifacts, and difficulties in detecting subtle cardiac abnormalities, exacerbated by subpar preprocessing and feature extraction techniques. Addressing these limitations, proposed MyoNet model emerges as a groundbreaking solution that fuses medical informatics with deep learning algorithms. Proposed model employs advanced preprocessing techniques, including Independent Component Analysis and Singular Value Decomposition, to eliminate noise and artifacts while preserving vital signal characteristics. ECG signals are then transformed into spectrograms, enabling multi-scale feature extraction for enriched pattern recognition. This transformation labels ECG data based on cardiac conditions, distinguishing MI from healthy cases with remarkable precision. Powered by a convolutional neural network, proposed model autonomously identifies intricate patterns with unparalleled accuracy, achieving a performance milestone of 97.24% on the Apnea-ECG database. By streamlining diagnosis and minimizing computational overhead, MyoNet offers a robust, efficient, and reliable diagnostic tool, empowering clinicians to detect MI earlier and enhance patient outcomes.</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

MyoNet: intelligent feature engineering for precise early prediction of myocardial infarction

  • 期刊论文

  • 发表日期：

    2026年2月23日

查看全文
  • Abhishek Shrivastava,
  • Santosh Kumar,
  • Nenavath Srinivas Naik

摘要

Rapid and precise myocardial infarction (MI) prediction is essential to avert life-threatening outcomes. Traditional ECG diagnostic methods often falter due to signal noise, artifacts, and difficulties in detecting subtle cardiac abnormalities, exacerbated by subpar preprocessing and feature extraction techniques. Addressing these limitations, proposed MyoNet model emerges as a groundbreaking solution that fuses medical informatics with deep learning algorithms. Proposed model employs advanced preprocessing techniques, including Independent Component Analysis and Singular Value Decomposition, to eliminate noise and artifacts while preserving vital signal characteristics. ECG signals are then transformed into spectrograms, enabling multi-scale feature extraction for enriched pattern recognition. This transformation labels ECG data based on cardiac conditions, distinguishing MI from healthy cases with remarkable precision. Powered by a convolutional neural network, proposed model autonomously identifies intricate patterns with unparalleled accuracy, achieving a performance milestone of 97.24% on the Apnea-ECG database. By streamlining diagnosis and minimizing computational overhead, MyoNet offers a robust, efficient, and reliable diagnostic tool, empowering clinicians to detect MI earlier and enhance patient outcomes.