| description abstract | The rise in heatwaves due to climate change is becoming a significant concern for outdoor workers, particularly leading to an increasing number of heat-related illnesses. To address the challenge, this study aimed to propose, as a process-based approach, a classification model using personal biometric characteristics to identify individuals who are vulnerable to extremely hot environments (i.e., high-risk groups). To this end, an experimental study was conducted, and experimental conditions were set in an environmental chamber by considering the extremely hot summer weather in Korea. With the data collected from a total of 70 people who voluntarily participated in the experiment, the classification model was developed by adopting multiple methodologies such as time-series clustering, independent samples t-test, and machine-learning algorithms. Consequently, it was found that the classification performance was the best with the multilayer perceptron algorithm, resulting in 0.800 in terms of the area under the receiver operating characteristic (AUROC) and 0.811 in terms of the area under the precision-recall curve (AUPRC). This study creates new ground in identifying individuals vulnerable to extremely hot environments in the domain of management in engineering by employing machine-learning-based classification algorithms with personal biometric characteristics. The proposed approach can be realized by utilizing a simple and low-cost bioelectrical impedance method for estimating human body composition (such as body fat mass and skeletal muscle mass) before they are put into the field. It is expected to aid in providing a more systematic and individualized management system for proactively preventing personal heat-related illnesses. | |
