Validity and Accuracy of Impulse-Response Models for Modeling and Predicting Training Effects on Performance of Swimmers.

PURPOSE: The aim of this study was to compare the suitability of models for practical applications in training planning. METHODS: We tested six impulse-response models, including Banister's model (Model Ba), a variable dose-response model (Model Bu), and indirect-response models differing in the way they account or not for the effect of previous training on the ability to respond effectively to a given session. Data from 11 swimmers were collected during 61 wk across two competitive seasons. Daily training load was calculated from the number of pool-kilometers and dry land workout equivalents, weighted according to intensity. Performance was determined from 50-m trials done during training sessions twice a week. Models were ranked on the base of Aikaike's information criterion along with measures of goodness of fit. RESULTS: Models Ba and Bu gave the greatest Akaike weights, 0.339 ± 0.254 and 0.360 ± 0.296, respectively. Their estimates were used to determine the evolution of performance over time after a training session and the optimal characteristics of taper. The data of the first 20 wk were used to train these two models and predict performance for the after 8 wk (validation data set 1) and for the following season (validation data set 2). The mean absolute percentage error between real and predicted performance using Model Ba was 2.02% ± 0.65% and 2.69% ± 1.23% for validation data sets 1 and 2, respectively, and 2.17% ± 0.65% and 2.56% ± 0.79% with Model Bu. CONCLUSIONS: The findings showed that although the two top-ranked models gave relevant approximations of the relationship between training and performance, their ability to predict future performance from past data was not satisfactory for individual training planning.

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

PURPOSE: Our aim was to assess whether we can predict satisfactorily performance in swimming and high frequency power (HF power) of heart rate variability from the responses to previous training. We have tested predictions using the model of Banister and the variable dose-response model. METHODS: Data came from ten swimmers followed during 30 weeks of training with performance and HF power measured each week. The first 15-week training period was used to estimate the parameters of each model for both performance and HF power. Both were then predicted in response to the training done during the second 15-week training period. The bias and precision were estimated from the mean and SD of the difference between prediction and actual value expressed as a percentage of performance or HF power at the first week. RESULTS: With the variable-dose response model, the bias for performance prediction was -0.24 ± 0.06 and the precision 0.69 ± 0.24% (mean ± between-subject SD). For HF power, the bias was 0 ± 21 and the precision 22 ± 8%. When HF power was transformed into performance using a quadratic relation in each swimmer established from the first 15-week period, the bias was 0.18 ± 0.74 and the precision 0.80 ± 0.30%. No clear trend in the error was observed during the second period. CONCLUSIONS: This study showed that the modeling of training effects on performance allowed accurate performance prediction supporting its relevance to control and predict week after week the responses to future training.

A model for the training effects in swimming demonstrates a strong relationship between parasympathetic activity, performance and index of fatigue.

Competitive swimming as a physical activity results in changes to the activity level of the autonomic nervous system (ANS). However, the precise relationship between ANS activity, fatigue and sports performance remains contentious. To address this problem and build a model to support a consistent relationship, data were gathered from national and regional swimmers during two 30 consecutive-week training periods. Nocturnal ANS activity was measured weekly and quantified through wavelet transform analysis of the recorded heart rate variability. Performance was then measured through a subsequent morning 400 meters freestyle time-trial. A model was proposed where indices of fatigue were computed using Banister's two antagonistic component model of fatigue and adaptation applied to both the ANS activity and the performance. This demonstrated that a logarithmic relationship existed between performance and ANS activity for each subject. There was a high degree of model fit between the measured and calculated performance (R(2)=0.84±0.14,p<0.01) and the measured and calculated High Frequency (HF) power of the ANS activity (R(2)=0.79±0.07, p<0.01). During the taper periods, improvements in measured performance and measured HF were strongly related. In the model, variations in performance were related to significant reductions in the level of 'Negative Influences' rather than increases in 'Positive Influences'. Furthermore, the delay needed to return to the initial performance level was highly correlated to the delay required to return to the initial HF power level (p<0.01). The delay required to reach peak performance was highly correlated to the delay required to reach the maximal level of HF power (p=0.02). Building the ANS/performance identity of a subject, including the time to peak HF, may help predict the maximal performance that could be obtained at a given time.