The validity, reliability and sensitivity of utilising a wearable GPS based IMU to determine goalkeeper specific training demands

[摘要] Despite the plethora of football focused literature, there is still very little known about the training practices of the goalkeeper (GK). The development of portable Global Positioning System (GPS) and Inertial Measurement Unit (IMU) devices ensured physical activities can be accurately measured within the training environment. The integration of inertial sensor fusion algorithms has allowed the IMU the ability to also detect non-locomotive activities that are specific to a sport. This technology is shown to be a valid method of analysis for the demands of an outfield football player, however, similar research into the GK position is required. Thus, the aim of this study was to investigate the validity, reliability and sensitivity of utilizing a wearable GPS based IMU to determine goalkeeper specific training demands.A total of 123 event variables were recorded via OptimEye G5 GPS units over 14 sessions from 6 professional GKs during the 2017-2018 Scottish Premiership season. GPS data was collected as part of normal daily monitoring and compared against corresponding computerized notational analysis of the same training sessions. Event variables were split into specific IMU events by a GK specific algorithm: Total Dives (TD), Dives Right (DvR), Dives Left (DvL), Dive Returns (DR) and Jumps. The intra-unit variation was derived from reproducibility of trends within the difference between GPS and corresponding Video Analysis (VA) counts. Unit sensitivity was investigated according to the relationship between average DR times and countermovement jump (CMJ) and ballistic press-up (BP) results which corresponded to lower and upper body velocity at peak power (m/s) respectively.There was no significant difference (p<0.05) between TD (91% false positive), DvL (89% false positive), DvR (87% false positive) and DR (78% false positive) but this was not the case for Jumps (18% false positive; p>0.05). Bland Altman 95% Limits of Agreement (LOA) show minimal variation for TD (-3.6 to 5.6), DvL (-1.75 to 4.04) and DvR (-3.38 to 3.13). However, DR (-13 to 12.6) and Jumps (-8.8 to 15.7) showed much wider LOA and variation from VA counts. Intra-unit variability was significantly different across all metrics with GPS units, over-estimating movement event counts compared to VA counts. Inter-unit sensitivity suggested that CMJ and lower body velocity at peak power (m/s) performance had the greatest correlation (r=0.992) with average DR times compared to BP and upper body velocity at peak power (r=0.684) and CMJ + BP combined (r=0.603).Based on these findings, the sensitivity of the OptimEye G5 GPS to count GK specific events was almost perfect (r = 903), however, the specificity of the IMU algorithm to distinguish the different movements was questionable. Jumps were significantly over- estimated, and in the meantime, we would suggest using video footage to compliment GPS data for accurate longitudinal analysis. This study provided novel information regarding the DR action, of which the lower body muscular profile plays the dominant part in. Although there are limitations within this study, these investigations should only act as the first step in understanding if the GPS coupled IMU has a place in accurately determining the training demands of a goalkeeper.

[发布机构] University:University of Glasgow;Department:School of Life Sciences

[关键词] Football, goalkeeper, GPS.