(39) THE USE OF TRAINING DATA TO CALCULATE CRITICAL VELOCITY AND PREDICT RACE PERFORMANCE TIMES FOR COLLEGIATE DISTANCE RUNNERS DURING INDOOR TRACK SEASON

Introduction: Predictions of running performance can be derived using the relationship between running velocity and the duration in which running velocity can be sustained (T_lim). A linear function is observed when graphing running velocity versus 1/T_lim, with the slope representing critical velocity (CV) and the y-intercept D prime (D’). CV represents the highest sustainable running velocity in which a metabolic steady state may obtained, with D’ being the finite work that can be performed above CV. CV has significant associations with distance running performance and is sensitive to changes in training. Both CV and D’ are typically assessed using a series of intentional exhaustive trials (IETs) which may interfere with the organization of training due to the exhaustive nature. Previous investigations have shown that CV and D’ can be derived from training data of recreational runners and accurately predict marathon race times without IETs.

Purpose: The purpose of this investigation was to determine if race performances for collegiate distance runners could be predicted using CV and D’ derived from the highest sustained running velocities during training preceding competitions without IETs.

Methods: Participants were provided with a global positioning system (GPS) watch (Forerunner 925, Garmin, Olathe, KS, USA) and were instructed to wear it during all training sessions. Training data was uploaded to the online platform Training Peaks (Peaks Software Company, Louisville, CO, USA) where the highest sustained running velocities over a duration of 2-, 5-, 6-, and 10-minute intervals were calculated, with linear regression used to determine CV and D’. Time durations were selected to align with previous investigations of CV and were based on common interval prescription by coaching staff. CV and D’ were calculated prior to every indoor track and field race using training data from the previous four weeks. Modeled race performance times (MRPT) were determined using the equation (Race Distance – D’)/CV. Statistical differences in MRPT and race performance time (RPT) were examined with paired-samples t tests. The agreement between MRPT and PRT were assessed using mean difference, 95% confidence intervals (95%CI) of mean difference, and limits of agreement (LOA). Pearson’s correlation was used to assess the strength of the relationship. Linear regression was used to calculate values for the standard error of the estimate (SEE). Data are reported as mean ± standard deviation.

Results: 16 pairs of RPT and MRPT were examined from members of a collegiate track and field program. Races examined included 1600 meter (m) (n=4), 3000m (n=9), and 5000m (n=3). Paired-samples t tests indicated a significant difference between MPT and RPT (605.1 ± 230.5 seconds (s) vs. 566.1 ± 216.4 S, p=0.00, d=0.17), with a mean difference of 39.0 ± 26.8 s (95%CI: 24.2 to 53.9). Significant association was detected between MRPT and RPT (r=.995, p=0.00), while LOA were -13.5 to 91.5 s with a SEE of 23.7 s.

Conclusion: Though previous investigations have used training data to accurately predict RPTs, the current investigation observed that this method does not accurately predict RPTs in collegiate distance runners. This may be due to a lack of IETs which may better reflect the intensities raced at. Practical applications: Coaches should integrate sessions with intervals that reflect race intensity to potentially be able to use training data to model race times.

Acknowledgements: None