Faculty of Sport, Allied Health and Performance Science, St Mary's University, Twickenham, UK.
Ballet Healthcare, The Royal Ballet, London, UK.
J Sports Sci. 2023 Mar;41(5):463-469. doi: 10.1080/02640414.2023.2223048. Epub 2023 Jun 28.
The aim was to determine the validity of an open-source algorithm for measuring jump height and frequency in ballet using a wearable accelerometer. Nine professional ballet dancers completed a routine ballet class whilst wearing an accelerometer positioned at the waist. Two investigators independently conducted time-motion analysis to identify time-points at which jumps occurred. Accelerometer data were cross-referenced with time-motion data to determine classification accuracy. To determine the validity of the measurement of jump height, five participants completed nine , nine and three double from a force plate. The jump height predicted by the accelerometer algorithm was compared to the force plate jump height to determine agreement. Across 1440 jumps observed in time-motion analysis, 1371 true positives, 34 false positives and 69 false negatives were identified by the algorithm, resulting in a sensitivity of 0.98, a precision of 0.95 and a miss rate of 0.05. For all jump types, mean absolute error was 2.6 cm and the repeated measures correlation coefficient was 0.97. Bias was 1.2 cm and 95% limits of agreement were -4.9 to 7.2 cm. The algorithm may be used to manage jump load, implement periodization strategies, or plan return-to-jump pathways for rehabilitating athletes.
目的是确定一种使用可穿戴加速度计测量芭蕾舞跳跃高度和频率的开源算法的有效性。九位专业芭蕾舞演员在腰部佩戴加速度计完成一节常规芭蕾舞课。两位研究人员独立进行时间-动作分析，以确定跳跃发生的时间点。将加速度计数据与时间-动作数据交叉参考，以确定分类准确性。为了确定测量跳跃高度的准确性，五名参与者从力量板上完成了九次、九次和三次双跳。使用加速度计算法预测的跳跃高度与力量板上的跳跃高度进行比较，以确定一致性。在时间-动作分析中观察到的 1440 次跳跃中，算法识别出 1371 个真阳性、34 个假阳性和 69 个假阴性，灵敏度为 0.98，精度为 0.95，漏报率为 0.05。对于所有跳跃类型，平均绝对误差为 2.6 厘米，重复测量相关系数为 0.97。偏差为 1.2 厘米，95%一致性界限为-4.9 至 7.2 厘米。该算法可用于管理跳跃负荷、实施周期性策略或计划康复运动员的回归跳跃路径。
