Scholarlink

<p>Cognitive load assessment is essential in understanding the human performance. Despite the recent advancements, there still exist challenges in real-time cognitive load (CL) assessment. In this work, we propose a methodology for CL assessment using smartphone camera-based imaging photoplethysmography (iPPG). In this study, we collected smartphone RGB camera videos and contact-based PPG signals (1000 Hz) from N= 10 healthy subjects in three CL states (baseline, Stroop test, and cognitive) in a semi-controlled experiment. Facial iPPG signals are extracted using Local Group Invariance method. Pulse rate variability (PRV) features are extracted from both iPPG and the PPG signals. The PRV features were applied to Kruskal–Wallis test, t-SNE visualization, and higher-order statistical (HOS) analysis. The PRV features are applied to recursive feature eliminator (RFE) and given to the XGBoost classifier. Experiments are performed, and performance is evaluated using various cross-validation techniques. Results show that our proposed methodology could differentiate between the CL states. iPPG-derived PRV features effectively differentiate CL states. HOS, namely third-order cumulants, enhances the distinction by capturing nonlinear interactions. t-SNE visualization shows significant cognitive state clustering, and XGBoost with RFE achieves the highest classification (F-measure = 78%). Thus, the proposed work could be extended to real-time, lead-free CL assessment, enabling scalable applications in workload monitoring, stress detection, and human performance evaluation.</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

Assessment of smartphone-based contactless imaging photoplethysmography signals for cognitive load monitoring

  • 期刊论文

  • 发表日期：

    2025年11月12日

查看全文
  • P. J. Swarubini,
  • Ryunosuke Kirita,
  • Tomohiko Igasaki,
  • Nagarajan Ganapathy

摘要

Cognitive load assessment is essential in understanding the human performance. Despite the recent advancements, there still exist challenges in real-time cognitive load (CL) assessment. In this work, we propose a methodology for CL assessment using smartphone camera-based imaging photoplethysmography (iPPG). In this study, we collected smartphone RGB camera videos and contact-based PPG signals (1000 Hz) from N= 10 healthy subjects in three CL states (baseline, Stroop test, and cognitive) in a semi-controlled experiment. Facial iPPG signals are extracted using Local Group Invariance method. Pulse rate variability (PRV) features are extracted from both iPPG and the PPG signals. The PRV features were applied to Kruskal–Wallis test, t-SNE visualization, and higher-order statistical (HOS) analysis. The PRV features are applied to recursive feature eliminator (RFE) and given to the XGBoost classifier. Experiments are performed, and performance is evaluated using various cross-validation techniques. Results show that our proposed methodology could differentiate between the CL states. iPPG-derived PRV features effectively differentiate CL states. HOS, namely third-order cumulants, enhances the distinction by capturing nonlinear interactions. t-SNE visualization shows significant cognitive state clustering, and XGBoost with RFE achieves the highest classification (F-measure = 78%). Thus, the proposed work could be extended to real-time, lead-free CL assessment, enabling scalable applications in workload monitoring, stress detection, and human performance evaluation.