Related Experiment Video
Updated: Aug 15, 2026
Evaluation of Commercial-Off-The-Shelf Wrist Wearables to Estimate Stress on Students
Published on: June 16, 2018
Occupational Heat Stress in Industrial Workers: Modifiable Predictors and Machine Learning Risk Classification in
Krishnan Srinivasan1, Dibyajyoti Saikia, Ashikh Seethy
1From the Department of Physiology, All India Institute of Medical Sciences (AIIMS), Guwahati, India (K.S., M.B., A. Sinha, S.L., S.M.A., B.H., J.K.); Department of Pharmacology, All India Institute of Medical Sciences (AIIMS), Guwahati, India (D.S.); and Department of Biochemistry, All India Institute of Medical Sciences (AIIMS), Guwahati, India (A. Seethy).
Background:
Occupational heat stress poses a growing threat to worker health and productivity in low- and middle-income countries. This study investigated modifiable predictors of heat stress among industrial workers and evaluated machine-learning models for risk classification.
Methods:
A cross-sectional field-based study was conducted among 1300 workers from the chemical, construction, and packaging industries. Environmental, physiological, occupational, and behavioral data were collected using standardized questionnaires and field measurements. Multivariable regression and machine-learning models were applied to identify predictors and classify heat stress risk.
Results:
Manual labor, long working hours, alcohol use, and inadequate hydration increased risk, while water availability, showers, and uniforms were protective. Random Forest performed best (accuracy: 72.6%; area under the curve: 0.777).
Conclusions:
Heat stress risk is strongly influenced by modifiable workplace and behavioral factors. Random Forest offers a reliable tool for surveillance and early prevention.