Modelling Physical Fatigue Through Physiological Monitoring Within High-Risk Professions

Abstract

Physical fatigue is not only a tiredness condition but can also be a symptom of various diseases or a precursor of workplace accidents and incidents. High-risk professions like firefighters and military personnel are more frequently exposed to it and its consequences. To contribute to current research efforts towards physical fatigue quantification approaches for these professionals, the current work aimed to test a previously developed machine learning algorithm based on physiological monitoring on a dataset of soldiers engaged in an incremental running protocol. Personal characteristics and signals from heart rate, breathing rate and core temperature were recorded from participants and used to extract 20 features. These were used as inputs to an XGBoosted algorithm to classify four levels of physical fatigue: low, moderate, heavy and severe. Outcomes showed an overall accuracy of 85% and accuracies of 100 and 92% for detecting the low and heavy levels. Also, the importance of individualised monitoring and the need for more than one physiological indicator to accurately characterise the subject physical fatigue status were demonstrated. Overall, the results confirm the model's contribution to current fatigue detection methods. Future studies will be oriented to test its validity within a bigger sample while assessing more occupational activities. (c) The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.