An Impulse-Response Model using Pace and Duration Correlates to Performance of a 1500 m Olympic Finalist

Abstract

INTRODUCTION: The Impulse-Response model (Banister 1976) has been used to model performance in running using heart rate based training impulse (TRIMPS), but runners typically track mileage/pace to quantify training load. The purpose of this study was to determine the utility of a performance model that quantifies training load by integrating pace (intensity) and duration in an elite runner. METHODS: Procedures were approved by the EMU-HSRC. Training data from 2000 and 2001 was used to calculate training doses (TSS) similarly to that proposed by Coggan (2006). Pace was exponentially weighted to calculate a normalized pace (NP) and intensity factor (IF=NP/21km pace); then, TSS=(IF2)*hours*100. Performance was modeled using a modified equation (Morton 1990) of the form p(t)= g(t)-h(t), where p(t) represents predicted performance, g(t) represents fitness derived by accumulation of TSS doses over time (constant tau1=42 days), and h(t) represents fatigue as a result of recent TSS doses (constant tau2). Tau2 was varied from 3, 7 and 14 days to test the effect of the fatigue time constant on the validity of the model. Mercier scores (MS; percentage of world record) were calculated from competition times and compared to p(t) using Spearman's rho. MS was compared to mileage using Pearson's correlation (a=0.05; SPSS 14.0) RESULTS: Performance (MS; 882.2±45.5) peaked (945) one week before national championships (814; 1st), and two months before the Olympic finals (918; 5th) in 2000. MS in 2001 (873.5±46.5) peaked (935) 10 days after national championships (762; 2nd). The average daily TSS was significantly greater in 2000 vs 2001 (56.1±1.4 vs 51.5±2.1; p<0.05, respectively). Predicted performance p(t) was best correlated with MS (rho=.73 and .75; p<0.01) when tau2 = 14 days. Tau2= 7 days also elicited significant, but lower correlations of MS and p(t)(rho=.45 and .43; p<0.05, 2000 and 2001 respectively). If Tau2=3 days, p(t) was not correlated to MS. MS was not correlated with mileage, although mileage was negatively correlated with p(t)(r=−.38 and -.33; p<01). CONCLUSION: Over two seasons, in an elite runner, mileage was not correlated to performance. An Impulse-Response model that used TSS as the training dose was correlated with performance. TSS may be a more valuable training metric than mileage for competitive runners.