Scholarlink

<p>Psychosocial stress while driving leads to driver inattention, a major contributing factor in many traffic accidents. Fast and accurate detection of a driver’s mental state is critical to mitigating health risks and preventing road crashes. Identifying driver inattention using wearables remains a significant challenge for advanced driver monitoring systems. In this study, we propose the use of Third-Order Cumulant (TOC) features for automated classification of driver inattention states using Textile Electrocardiogram (tECG) signals. ECG data was collected from 15 subjects using textile electrodes at 256 Hz during two scenarios: normal driving and driving while engaging in a phone call (inattention state). A 2D third-order cumulant matrix was computed from each ECG segment, and discriminative features were extracted using a 1D slice integration method. The extracted features were used to train classifiers including Support Vector Machine, Random Forest (RF), Decision Tree, and 1D Convolutional Neural Network (1D-CNN), evaluated using Leave-One-Subject-Out cross-validation. Results show that the proposed approach effectively identifies driver inattention states, with RF and TOC features yielding a weighted F1-score of 73.44% and an average accuracy of 78.11%. Among various segment lengths, the 20-second window provided superior performance, with 1D-CNN network and TOC features we observed a weighted F1-score of 72.86% and an average accuracy of 76.29% . These findings highlight the potential of TOC-based features in the development of fast and accurate real-world driver monitoring 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

Automated identification of driver inattention state using textile electrocardiograms and higher-order cumulant features based classification

  • 期刊论文

  • 发表日期：

    2026年4月20日

查看全文
  • Kaveti Pavan,
  • Ankit Singh,
  • Digvijay S. Pawar,
  • Nagarajan Ganapathy

摘要

Psychosocial stress while driving leads to driver inattention, a major contributing factor in many traffic accidents. Fast and accurate detection of a driver’s mental state is critical to mitigating health risks and preventing road crashes. Identifying driver inattention using wearables remains a significant challenge for advanced driver monitoring systems. In this study, we propose the use of Third-Order Cumulant (TOC) features for automated classification of driver inattention states using Textile Electrocardiogram (tECG) signals. ECG data was collected from 15 subjects using textile electrodes at 256 Hz during two scenarios: normal driving and driving while engaging in a phone call (inattention state). A 2D third-order cumulant matrix was computed from each ECG segment, and discriminative features were extracted using a 1D slice integration method. The extracted features were used to train classifiers including Support Vector Machine, Random Forest (RF), Decision Tree, and 1D Convolutional Neural Network (1D-CNN), evaluated using Leave-One-Subject-Out cross-validation. Results show that the proposed approach effectively identifies driver inattention states, with RF and TOC features yielding a weighted F1-score of 73.44% and an average accuracy of 78.11%. Among various segment lengths, the 20-second window provided superior performance, with 1D-CNN network and TOC features we observed a weighted F1-score of 72.86% and an average accuracy of 76.29% . These findings highlight the potential of TOC-based features in the development of fast and accurate real-world driver monitoring systems.