Developing a Linear Mixed Model to Predict RPE and sRPE in Female Elite Football Players Using External Load Measures

Abstract

Abstract

Introduction: Despite the broad consensus on the importance of training load monitoring and management for footballers’ performance, existing research on female football players remains scarce. The present study aimed to develop and evaluate a linear mixed model (LMM) to predict the rating of perceived exertion (RPE) and session RPE (sRPE) in female elite football players using GPS-derived external load measures.

Methods: A sample of 58 players from two clubs in Norway were monitored across the 2020 and 2021 seasons (n = 4311 training observations). Physical performance data were collected using STATSports GPS APEX, while RPE and sRPE were collected using the PM Reporter Pro smartphone application. Associations between RPE, sRPE, and selected independent variables were investigated using LMM analysis.

Results: The final model demonstrated accurate predictive performance of RPE and sRPE, with coefficient of determination (R2) values of 0.65 and 0.66, root mean squared error (RMSE) of 1.16 and 142.2 and mean absolute error (MAE) 0.93 and 104.9, respectively. Key predictors included session duration, total distance (TD), high-speed running distance (HSRD), sprint distance (SpD), peak speed (Peakspeed), and player status (starter or substitute). The high intraclass correlation coefficients (ICC) indicated that a considerable proportion of the total variability in sRPE and RPE responses could be attributed to individual player differences.

Conclusion: This study highlights the potential of using GPS-derived data to predict RPE and sRPE values in female football players, providing practitioners with valuable information to tailor individual training sessions and balance training intensity and recovery.