Scholarlink

Left Ventricular Hypertrophy (LVH) is a significant predictor of Cardiovascular Disease (CVD) onset, with early detection being crucial for implementing targeted treatment to prevent further deterioration. Current methods for detecting LVH often have low accuracy due to the challenge of extracting relevant information from complex Electrocardiographic (ECG) signals. This study proposes a novel LVH detection algorithm based on Polygonal Area Feature Selection (PAFS) and a Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model. The algorithm uses PAFS to identify the most relevant features for LVH detection and employs a CNN-LSTM model to train on these features, enhancing the accuracy of ECG waveform analysis and LVH detection. To evaluate the algorithm’s efficacy, ECG data from LVH patients collected in a clinical setting were analyzed using a 5-fold cross-validation approach. The results indicate that our algorithm outperforms existing methods, achieving a sensitivity of 84.7%, a specificity of 86.1%, and an overall accuracy of 85.8%. Notably, the PAFS method identified that the most discriminative features for LVH detection are located around the QRS complex. These findings highlight the potential of AI in improving the diagnostic accuracy of LVH.

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

Left Ventricular Hypertrophy Detection Algorithm Using Feature Selection and CNN-LSTM

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhenxing Zhang,
  • Jun Ge,
  • Yilei Wang,
  • Zheng Wei

摘要

Left Ventricular Hypertrophy (LVH) is a significant predictor of Cardiovascular Disease (CVD) onset, with early detection being crucial for implementing targeted treatment to prevent further deterioration. Current methods for detecting LVH often have low accuracy due to the challenge of extracting relevant information from complex Electrocardiographic (ECG) signals. This study proposes a novel LVH detection algorithm based on Polygonal Area Feature Selection (PAFS) and a Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model. The algorithm uses PAFS to identify the most relevant features for LVH detection and employs a CNN-LSTM model to train on these features, enhancing the accuracy of ECG waveform analysis and LVH detection. To evaluate the algorithm’s efficacy, ECG data from LVH patients collected in a clinical setting were analyzed using a 5-fold cross-validation approach. The results indicate that our algorithm outperforms existing methods, achieving a sensitivity of 84.7%, a specificity of 86.1%, and an overall accuracy of 85.8%. Notably, the PAFS method identified that the most discriminative features for LVH detection are located around the QRS complex. These findings highlight the potential of AI in improving the diagnostic accuracy of LVH.