Related Experiment Video
Updated: Apr 23, 2026
08:33
A Cross-Disciplinary and Multi-Modal Experimental Design for Studying Near-Real-Time Authentic Examination Experiences
Published on: September 4, 2019
6.1K
Predicting Emotional Exhaustion with Multimodal Sensor Data During Return-to-Work Trajectories: A 6-Month
Lea Berkemeier1,2,3, Wim Kamphuis4, Hilbrand Oldenhuis5
1Department of Psychology, Health and Technology, Centre for eHealth Research and Wellbeing, University of Twente, 7522NB, Enschede, The Netherlands. l.berkemeier@utwente.nl.
Journal of Occupational Rehabilitation
|April 22, 2026
Summary
Multimodal sensor data can modestly predict emotional exhaustion during return-to-work. Individual patterns of emotional exhaustion (EE) vary, requiring personalized return-to-work strategies.
Area of Science:
- Occupational Health
- Digital Health
- Psychology
Background:
- Emotional exhaustion (EE) is central to burnout and affects return-to-work (RTW) after sick leave.
- Previous studies linked EE to sleep, activity, mobility, and smartphone use, but individual variations over time remain unclear.
- Monitoring and predicting EE during RTW using sensor data requires further investigation.
Purpose of the Study:
- To examine how emotional exhaustion (EE) can be monitored and predicted using multimodal data from smartphones and smart rings.
- To investigate within- and between-person associations of EE with sensor-derived data during RTW trajectories.
- To assess the predictive performance of multimodal features for EE.
Main Methods:
- Eighteen employees on sick leave for burnout/stress provided daily Ecological Momentary Assessments (EMAs) of EE.
- Multimodal sensor data on sleep, physical activity (Oura ring), mobility, and phone usage (Avicenna app) were collected for six months.
- Correlational analyses, linear mixed models, and subject-dependent Random Forest (RF) models were used to analyze associations and predictive performance.
Main Results:
- Subject-dependent RF models achieved an average predictive accuracy (Spearman's ρ̄) of 0.38.
- Sleep and physical activity features showed more consistent associations with EE than mobility and smartphone usage.
- Significant inter- and intra-individual variability was observed in EE patterns and predictive features.
Conclusions:
- Multimodal sensor data offer modest predictive capabilities for emotional exhaustion, with notable individual differences.
- The high variability in EE patterns underscores the need for tailored return-to-work counseling.
- Future research should focus on longer monitoring periods and exploring inter- and intra-individual variability in EE.
