Delves, Robert ORCID: https://orcid.org/0000-0002-7111-5054 (2023) Development and use of a Consistent Approach for Processing GNSS-derived Acceleration Data in Team Sports. PhD thesis, Victoria University.

Abstract

In team sports, player tracking systems are used to monitor training and competition. However, challenges exist for practitioners, as different tracking systems can be used for competition compared to training. Additionally, updates/changes being made to proprietary software or firmware within the tracking system can impact longitudinal analysis of the activity profile. Both scenarios can result in different athlete outputs for the same activity. Acceleration is an important metric in the team sport activity profile, which can help quantify the rate of change in athlete speed. However, acceleration as a variable can also be impacted by changes to software/firmware as well as differences between tracking systems. This thesis explored how acceleration has been quantified in the literature, examined if a common filter to Global Navigation Satellite System (GNSS) data could reduce or eliminate differences due to different systems or changes within a system, before examining two longitudinal-type studies using this common filtering method on acceleration data. Chapter 3 identified how acceleration had previously been quantified in elite team sport research. The quantification of acceleration via counts was chosen in 72% of all studies, however only ~13% of studies included how acceleration was filtered. To determine how acceleration was filtered by providers, Chapter 4 was designed as an anonymous survey, attempting to outline current filtering practices of acceleration data. Only two responses were received from 20 invitations, indicating that system providers were unwilling to provide filtering information. Consequently, this thesis examined the use of a common filter to process GNSS data to improve the consistency in the processing of acceleration both within research and applied environments. The common filter was intended to improve the longitudinal and potentially, the between-system comparison of acceleration data. Chapter 5 examined the impact of applying a 1 Hz, fourth order Butterworth filter to two different GNSS devices during elite rugby league training sessions. Following the application of the common filter, there was no substantial difference between GNSS models for average acceleration (Diff; CI: −0.04; −0.04 to −0.04), whilst the root mean square deviation (RMSD) between devices improved (1.77 ± 0.37 to 0.27 ± 0.23 m·s−2). The results from Chapter 5 indicated that if the use of a common filter could improve a greater similarity of results between tracking systems, then, pending validation against a criterion measure, this filter could be used on longitudinal datasets or where different systems have been used. The filter was then evaluated against three-dimensional motion capture technology (VICON) in Chapter 6 during a series of small-sided football (association) games and circuits. The RMSD for speed (0.17 ± 0.04 m·s-1) was acceptable and acceleration error increased as speed increased (RMSD: 0.55 ± 0.17 m·s-2). Following validation, the effects of the six-again rule change upon acceleration in National Rugby League (NRL) competition was examined. The acceleration intercepts across all positions were substantially greater following the introduction of the six-again rule in the 2020 (mean ± SD; 1.02 ± 0.10 m·s−2) and 2021 seasons (1.05 ± 0.08 m·s−2) compared to the previous competition format (2019; 0.91 ± 0.07 m·s−2), indicating an increase in acceleration outputs. A longitudinal analysis of NRL training weeks was completed, where the distribution of training volume and intensity was examined using the common filter. Speed intensity (the magnitude of speed) was manipulated to facilitate performance when fewer training sessions were completed in shorter microcycles (5-6 days; effect size range = 0.34 – 1.26), whilst the intensity of impulse (acceleration force over time) remained stable. This thesis identified that there’s a lack of information relating to the filtering of acceleration via athlete tracking systems in research. However, this thesis identified that tracking system manufacturers do process acceleration differently which may impact upon comparisons between research and between technology. A common filter reduced the difference in acceleration metrics between GNSS devices during rugby league training sessions before being validated against VICON. With use of a common filter, this research identified that the acceleration intensity in NRL competition had increased and that speed intensity during NRL training weeks was altered with shorter turnarounds between matches, whilst impulse was maintained.

Search Google Scholar

Repository staff login