Scholarlink

<p>In our research, a deep learning-based method for autonomous evaluation of Sleep Apnea-Hypopnea Syndrome (SAHS) utilizing minimal physiological cues is presented. For better classification performance, a hybrid Convolutional Neural Network-Recurrent Neural Network (CNN-RNN) model is used, in contrast to the previous study, which employed a 3-stage support vector machine (SVM) based technique and concentrated on severe SAHS patients. Temporal modeling and feature extraction are performed using ECG (electrocardiograph) and nasal pressure (NP) data. To increase the robustness of the model, data augmentation methods like time shifting, noise injection and scaling are used. For Apnea classification, the proposed model produces precision (95.2%), recall (95.6%) and an overall accuracy of 96.3%. By resolving the limitations of existing approaches and including both OSA (obstructive sleep apnea) and CSA (central sleep apnea) event detection, this work presents a scalable and efficient alternative for real-time SAHS assessment in home-based and clinical healthcare situations.</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

A deep learning approach with data augmentation for robust SAHS classification from nasal pressure signals

  • 期刊论文

  • 发表日期：

    2025年7月30日

查看全文
  • M. Vijay Kumar,
  • Tejaswini Matlapudi,
  • Prema Srilatha Ulliboina,
  • Sri Sai Susmitha Ponnam,
  • Lella Sai Sravani,
  • Dinesh Reddy Gunnam,
  • V N Siddardha Bhimireddy

摘要

In our research, a deep learning-based method for autonomous evaluation of Sleep Apnea-Hypopnea Syndrome (SAHS) utilizing minimal physiological cues is presented. For better classification performance, a hybrid Convolutional Neural Network-Recurrent Neural Network (CNN-RNN) model is used, in contrast to the previous study, which employed a 3-stage support vector machine (SVM) based technique and concentrated on severe SAHS patients. Temporal modeling and feature extraction are performed using ECG (electrocardiograph) and nasal pressure (NP) data. To increase the robustness of the model, data augmentation methods like time shifting, noise injection and scaling are used. For Apnea classification, the proposed model produces precision (95.2%), recall (95.6%) and an overall accuracy of 96.3%. By resolving the limitations of existing approaches and including both OSA (obstructive sleep apnea) and CSA (central sleep apnea) event detection, this work presents a scalable and efficient alternative for real-time SAHS assessment in home-based and clinical healthcare situations.