A new empirical approach for predicting heat strain in workers exposed to hot indoor environments

Abstract

There is still a great need for a comprehensive index that could fully describe heat stress and, at the same time, provides a reliable correlation with the physiological responses of the human body. Using artificial neural networks, this study aims to present a new empirical model for predicting heat strain based on body core temperature in workers exposed to hot indoor environments. The study group consisted of 165 male workers working in heat treatment processes of metal industries in the central Iran. A predictive model was developed using eight parameters: age, metabolism rate, body mass index, body surface area, dry-bulb temperature, globe temperature, air velocity and relative humidity. The multilayer feed forward neural networks with different structures were developed using R 3.2.2 statistical software. The results showed that the mean square error of the core temperature predicted by the proposed model was 0.25℃. Based on the Garson algorithm, the dry-bulb and globe temperatures were found to be the most important factors that could affect the human heat strain. The proposed model can be a useful tool for occupational health professionals in analysing heat strain in hot environments.