부산대학교 도서관

Performance metrics are extremely important for military instructors and leadership to gauge soldier fitness and readiness and adjust training regimens accordingly. One important performance indicator during military training events is the time taken for a soldier to complete a structured march, hereon referred to as time-to-completion (TTC). During these marches, wearable physiological sensors can be used to infer a soldier’s physiological state and exertion rate, which in turn can be used for predicting TTC. In this work, we present a model that uses signals from a multimodal wearable sensor to predict TTC for soldiers undergoing a 12-mile structured ruck march. Predictions are made at discrete time points (checkpoints) throughout the march using features from skin temperature (SKT), heart rate (HR), estimated core temperature (ECT), and triaxial accelerometry. To utilize the structured nature of these marches, separate models are trained at each checkpoint using features from both the current and past checkpoints. By 120 min (2/3 of the expected 180-min completion time), we achieved a TTC root-mean-square error (RMSE) of 7.12 min and a mean absolute error (MAE) of 5.21 min using this model. Integral to TTC estimation accuracy were gait-related features such as the standard deviation of vertical acceleration (ACC). Features such as HR slope and performance metrics from prior exercises minimally improved accuracy. The deployment of this model will enable continuous monitoring of performance metrics for online TTC estimation.