Machine learning-based detection of workplace stress using wearable and multimodal data: a systematic literature...

Area of Science:

• Occupational Health
• Computer Science
• Biomedical Engineering

Background:

• Workplace stress significantly impacts employee wellbeing, productivity, and contributes to burnout and turnover.
• Detecting stress in real-world settings is challenging, but machine learning (ML) and deep learning (DL) offer potential solutions.
• Advancements in wearable sensors and multimodal data collection facilitate individual stress tracking.

Purpose of the Study:

• To systematically review ML and DL approaches for detecting workplace stress using wearable and multimodal data.
• To analyze dataset characteristics, sensor modalities, detection performance, and workplace contexts.
• To identify research gaps and future directions in workplace stress detection.

Main Methods:

• Systematic literature review of 20 selected studies.
• Focus on ML and DL techniques applied to wearable and multimodal data.
• Examination of dataset features, sensor types, performance metrics, and occupational domains.

Main Results:

• ML and DL models effectively detect workplace stress using physiological, behavioral, and multimodal data.
• High-risk occupations show particular promise for accurate stress detection.
• Studies indicate the potential of these technologies for monitoring and managing workplace stress.

Conclusions:

• ML and DL approaches are effective for workplace stress detection, especially with multimodal and wearable sensor data.
• Further research is needed to address limitations like small sample sizes, data diversity, and integration of central nervous system signals.
• The findings highlight the potential for improved employee wellbeing and organizational productivity through advanced stress detection methods.