Author(s): WILLIAMS, D.1,2,3, DUNN, M.4, WORSFOLD, P.2,5, NEWTON, D.2,3, FAULKNER, S.H.3,6, D’ANDREA, F.1, WHEAT, J.1, Institution: SHEFFIELD HALLAM UNIVERSITY, Country: UNITED KINGDOM, Abstract-ID: 1707

INTRODUCTION:
Cycling biomechanics and position optimisation research requires that an appropriate number of crank revolutions are collected to test hypotheses. Despite this, there are no evidence-based recommendations for a representative quantity for cycling experiments. The aim of this study was to determine the number of revolutions necessary to achieve stable sagittal plane cycling biomechanics outcomes whilst riders were in an aerodynamic position.
METHODS:
Eighteen elite cyclists (10 females, 8 males, 22 ± 7 years, 1.74 ± 0.10 m, 69.0 ± 9.5 kg) completed a 3-min maximal bout on a cycling ergometer at a fixed cadence of 103 ± 2 rpm in their pursuit position. Lower-limb biomechanical data were captured using two-dimensional motion capture (100 Hz) and force pedals (200 Hz). Raw data were filtered using a 4th order Butterworth low-pass filter (6 Hz) and interpolated to 100 points per crank cycle using cubic splines. 37 discrete and 15 time-series kinematic, kinetic and inverse dynamics outcomes were calculated from the middle 60 crank revolutions. Sequential Averaging (SA)(1) and Iterative ICC (3, 1)(2) were performed to assess outcome stability.
RESULTS:
SA showed that all discrete and time-series variables reached stability by at most 46 and 59 revolutions, respectively. However, the number of revolutions SA indicated were required for stability reduced with fewer reference revolutions (NREF) and greater standard deviation (SD) thresholds. ICC results showed that all 37 discrete variables, excluding peak knee flexion angular velocity (n = 32), reached maximum stability by 7 revolutions. In addition, all discrete variables achieved good stability (ICC ≥ 0.8) by 3 revolutions and 35 variables that reached excellent (ICC ≥ 0.9) did so by 4 revolutions. Despite this, the 95% confidence interval lower-bound did not reach ≥ 0.9 in 14 variables. Joint kinetics variables were found to be the least stable by both SA and ICC.
CONCLUSION:
This study was the first application of SA and ICC to determine stability of cycling biomechanics outcomes. SA was most conservative and, as expected, results were affected by the SD threshold and NREF(2). However, SA allowed an indication of the stability of time-series data(3). In contrast, ICC did not require subjective inputs(2). The revolutions ICC indicated were required for stability were lower, but results did not exhibit a pattern of increased stability as NREF increased(4). It is possible that stability was assumed when the mean deviation (difference between the moving and cumulative mean) was still converging(3). The results of this study can be used to inform the design of future experiments in terms of the number of crank revolutions analysed, which might depend on the dependent variables included. In addition, the limitations of both stability methods should be considered in future applications.
REFERENCES:
1) Bates et al., J Biomech, 1983
2) James et al., J Sport Sci Med, 2007
3) Taylor et al., J Sport Sci, 2015
4) Chua et al., Sport Biomech, 2017