Modeling of performance and ANS activity for predicting future responses to training.

Area of Science:

• Sports Science
• Exercise Physiology
• Biostatistics

Background:

• Monitoring training load and physiological responses is crucial for optimizing athletic performance.
• Heart rate variability (HF power) offers insights into autonomic nervous system regulation during training.
• Predictive modeling can aid in tailoring training regimens to individual athletes.

Purpose of the Study:

• To evaluate the predictive accuracy of training responses using established models.
• To assess the predictability of swimming performance and high-frequency power (HF power) from previous training data.
• To compare the variable dose-response model against the Banister model for prediction.

Main Methods:

• Ten swimmers were monitored over 30 weeks, with weekly measurements of performance and HF power.
• Models were parameterized using the first 15 weeks of training data.
• Predictions for performance and HF power were made for the subsequent 15 weeks.
• Prediction accuracy was assessed using bias and precision metrics.

Main Results:

• The variable dose-response model demonstrated good predictive accuracy for swimming performance (bias -0.24 ± 0.06%, precision 0.69 ± 0.24%).
• Predictions for HF power showed higher variability (bias 0 ± 21%, precision 22 ± 8%).
• Transforming HF power to performance improved prediction accuracy (bias 0.18 ± 0.74%, precision 0.80 ± 0.30%).

Conclusions:

• Training effect modeling provides accurate predictions of swimming performance.
• This approach is relevant for week-to-week control and prediction of training responses.
• Further refinement may be needed for precise HF power prediction.