Predicting thermal comfort of healthcare workers in hot environments based on machine learning algorithms

Song Wenfang, Yu Sijing, Liu Ziyu, Tang Rong, Ding Qiuyue, Zhou Shiqing, Wang Xiaolan,
Predicting thermal comfort of healthcare workers in hot environments based on machine learning algorithms,
Energy and Buildings,
Volume 347, Part B,
2025,
116409,
ISSN 0378-7788,
https://doi.org/10.1016/j.enbuild.2025.116409.
(https://www.sciencedirect.com/science/article/pii/S0378778825011399)
Abstract: During the COVID-19 pandemic, healthcare workers wearing medical protective equipment (MPE) in high-temperature environments have faced significant heat stress, leading to increased health risks and reduced efficiency. Accurate prediction of thermal comfort is essential to mitigate heat stress and ensure their well-being. Traditional models fall short in predicting thermal comfort in dynamic medical settings due to complex interactions between environmental factors, MPE, and individual responses. In contrast, machine learning models offer enhanced accuracy and adaptability, effectively addressing these limitations by capturing complex patterns and interactions. This study integrates environmental and physiological parameters into machine learning models to provide real-time, accurate predictions. In the study, 20 subjects (10 males and 10 females) were recruited to simulate the working scenarios of healthcare workers at different ambient temperatures (26 °C, 28 °C, 30 °C, 32 °C, 34 °C). The subjects performed protocols like those followed by nucleic acid testing personnel, and their physiological and psychological responses were measured. Through correlation analysis and rigorous training and validation of various machine learning models, the Stacking model achieved a cross-validation accuracy of 89 % and an F1 score of 87 %. This model uses only a few easily measurable parameters, i.e., ambient temperature, working duration, gender, chest temperature, and dorsal hand temperature, to deliver high-precision predictions. The study offers a practical solution to manage heat stress in healthcare workers, enhancing their well-being and performance in challenging environments.
Keywords: Healthcare workers, Machine learning algorithms; Thermal comfort prediction; Heat stress management