Wearable Sensor-Based Assessment of 24-Hour Movement Behaviors and Physical Fitness Performance in Fire Academy Cadets

Abstract

Background: This study analyzed the time allocation characteristics of 24-h movement behaviors among firefighter cadets, clarified their independent associations with physical fitness performance, and identified key predictors of physical fitness using compositional data analysis. Methods: A total of 333 fire academy cadets (326 males, 7 females) from five specialty directions were recruited. ActiGraph GT3X-BT accelerometers were used to objectively measure 24-h movement behaviors over seven consecutive days. Isometric log-ratio (ILR) transformation was applied to compositional time-use data, and bidirectional stepwise regression based on AIC was used to identify key predictors. Robustness checks included bootstrap resampling, 10-fold cross-validation, leave-one-out sensitivity analysis, and influence diagnostics. Results: The optimal model (F = 14.46, p < 0.001, adjusted R2 = 0.260) identified weight (β = −0.42, p < 0.001), sex (β = −16.14, p < 0.001), age (β = 2.00, p < 0.01), height (β = 0.18, p < 0.05), and Fire Rescue Technology (Fire Facilities) specialty (β = −4.87, p < 0.001) as significant predictors. The sex association should be interpreted as exploratory given the very small female subsample (n = 7). All ILR-transformed movement components were removed by stepwise regression. The weekend model showed slightly higher explanatory power (adjusted R2 = 0.270) than the weekday model (adjusted R2 = 0.260). Conclusions: Demographic and physical characteristics, rather than 24-h movement composition, are the primary predictors of physical fitness performance in this highly structured training population. The sex-related finding is exploratory and sample-specific due to the small female subsample. These findings provide evidence-based guidance for physical fitness training program design in fire academies and similar structured training environments.