Scholarlink

<p>Stress is a pervasive aspect of modern life, significantly impacting overall well-being. This study focuses on analyzing mental stress through Photoplethysmography (PPG) and Galvanic Skin Response (GSR) signals using machine learning algorithms. A preliminary investigation involving 15 subjects under various stress-inducing conditions like Math test (calculating backward arithmetic subtraction) and Stroop color word test was conducted. The methodology involved data acquisition and preprocessing followed by feature extraction and optimization using AI-driven algorithms. Results indicate that features such as Pulse Rate (PR), Pulse Rate Variability (PRV), and characteristics of GSR phasic components are key indicators of stress. Employing machine learning classifiers, notably Random Forest, yielded promising results in stress detection, achieving Accuracy (80 ± 8.31), Precision (86.43 ± 12.04), Recall (86.67 ± 12.47), Specificity (66.67 ± 29.81), and F1-score (85.17 ± 5.98). This study underscores the potential of utilizing autonomic biosignals for dynamic stress monitoring and management. Furthermore, it highlights the feasibility of integrating such systems into wearable devices for real-time stress assessment and personalized interventions, with implications for improving psychological well-being and developing mobile healthcare applications.</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

Detection of Stress from PPG and GSR Signals using AI Framework

  • 期刊论文

  • 发表日期：

    2025年1月3日

查看全文
  • Swagata Barik,
  • Vinay Kumar Thakur,
  • Mohasin Ali Miah,
  • Saurabh Pal

摘要

Stress is a pervasive aspect of modern life, significantly impacting overall well-being. This study focuses on analyzing mental stress through Photoplethysmography (PPG) and Galvanic Skin Response (GSR) signals using machine learning algorithms. A preliminary investigation involving 15 subjects under various stress-inducing conditions like Math test (calculating backward arithmetic subtraction) and Stroop color word test was conducted. The methodology involved data acquisition and preprocessing followed by feature extraction and optimization using AI-driven algorithms. Results indicate that features such as Pulse Rate (PR), Pulse Rate Variability (PRV), and characteristics of GSR phasic components are key indicators of stress. Employing machine learning classifiers, notably Random Forest, yielded promising results in stress detection, achieving Accuracy (80 ± 8.31), Precision (86.43 ± 12.04), Recall (86.67 ± 12.47), Specificity (66.67 ± 29.81), and F1-score (85.17 ± 5.98). This study underscores the potential of utilizing autonomic biosignals for dynamic stress monitoring and management. Furthermore, it highlights the feasibility of integrating such systems into wearable devices for real-time stress assessment and personalized interventions, with implications for improving psychological well-being and developing mobile healthcare applications.