Football Movement Profile-Based Creatine-Kinase Prediction Performs Similarly to Global Positioning System-Derived...

Area of Science:

• Sports Science
• Exercise Physiology
• Biotechnology

Background:

• External load monitoring is crucial for athlete recovery and performance optimization.
• Previous research has explored the link between training load and creatine kinase (CK) levels using Global Positioning Systems (GPS).
• The Football Movement Profile (FMP) offers a more comprehensive measure of player load.

Purpose of the Study:

• To compare the predictive accuracy of GPS-derived and FMP-derived creatine kinase (CK) prediction models in national-team soccer players.
• To evaluate the performance of general versus individualized CK prediction models.

Main Methods:

• Four hundred forty-four national-team soccer players (U15 to senior) were monitored using GPS during training and matches.
• Daily CK levels were measured from whole blood samples.
• Machine learning models (using GPS, FMP, or combined data) were developed to predict next-day CK values, comparing general and individualized approaches.

Main Results:

• FMP-based (R2=.60), GPS-based (R2=.62), and combined models (R2=.62) showed similar performance in predicting CK levels.
• General machine learning models demonstrated superior predictive power (R2=.57) compared to individualized models (R2=.37) for 73% of players.
• Root-mean-square error (RMSE) values were comparable across the different model types.

Conclusions:

• FMP-derived CK prediction models are as effective as GPS-derived models for soccer players.
• General machine learning models provide more accurate CK predictions than individualized models.
• These findings support the use of FMP and general models for monitoring athlete recovery and optimizing training periodization.