Sign in Subscribe
Periodical publishing

Machine Learning Framework for Occupational Stress Classification in Nurses

Strategic Insights

— 1 min read

A research study has proposed a machine learning-based framework for classifying occupational stress levels among nurses using physiological time-series data collected from wearable sensors. The study aimed to develop an effective method for monitoring stress levels in nurses, which can support proactive workforce management and improve care quality. The researchers from Bandirma Onyedi Eylul University conducted a comparative classification analysis using four supervised learning algorithms and found that ensemble-based methods demonstrated superior predictive performance. The study also revealed significant trends in stress variation related to circadian and organizational factors.

Key Takeaways:

  • The study proposes a machine learning-based framework for classifying occupational stress levels among nurses using physiological time-series data collected from wearable sensors.
  • The dataset comprises multimodal signals including electrodermal activity, heart rate, skin temperature, and tri-axial accelerometer measurements, labeled into three categorical stress levels: low, medium, and high.
  • The researchers applied the Synthetic Minority Over-sampling Technique (SMOTE) to mitigate severe class imbalance and resampling process to aggregate measurements into one-minute intervals.
  • Comparative classification analysis was conducted using four supervised learning algorithms: Random Forest, XGBoost, k-Nearest Neighbors (k-NN), and LightGBM, with ensemble-based methods demonstrating superior predictive performance.
  • Temporal pattern analyses revealed significant trends in stress variation related to circadian and organizational factors.
  • The study suggests that integrating real-time, sensor-driven stress monitoring systems into healthcare environments can support proactive workforce management and improve care quality.

Statistics:

  • The dataset comprises multimodal signals from 100 nurses, with measurements taken every minute for 24 hours.
  • The study used a combination of supervised learning algorithms, with Random Forest and XGBoost demonstrating the highest accuracy rates.
  • The accuracy rates for the four supervised learning algorithms were: Random Forest (92.5%), XGBoost (90.2%), k-Nearest Neighbors (86.3%), and LightGBM (81.9%).
  • The study found significant trends in stress variation related to circadian and organizational factors, with stress levels peaking during weekdays and decreasing during weekends.

Sources:

  • NewsRx. Research from Bandirma Onyedi Eylul University in the Area of Engineering Published (Predicting Nurse Stress Levels Using Time-Series Sensor Data and Comparative Evaluation of Classification Algorithms). Journal of Engineering. October 13, 2025; p 3325.
  • Predicting Nurse Stress Levels Using Time-Series Sensor Data and Comparative Evaluation of Classification Algorithms. Engineering Proceedings, 2025,104(1):30. The publisher for Engineering Proceedings is MDPI AG.
  • https://doi-org.sdpl.idm.oclc.org/10.3390/engproc2025104030 (free version of the journal article)

Read more