ºìÐÓÊÓƵ

Topic

Invisible Monitoring Using Athlete Worn Sensors in Elite Women's Soccer

School of Exercise Science, Physical and Health Education

Date & location

• Tuesday, April 1, 2025

• 10:00 A.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Marc Klimstra, School of Exercise Science, Physical and Health Education, University of Victoria (Supervisor)

• Dr. Ming-Chang Tsai, School of Exercise Science, Physical and Health Education, UVic (Member)

• Dr. Cesar Meylan, School of Kinesiology, University of British Columbia (Outside Member)

• Dr. Dave Clarke, Department of Biomedical Physiology and Kinesiology, SFU (Outside Member) 

External Examiner

• Dr. Jean-Benoit Morin, Sport Sciences and Physical Education, Universite Jean Monnet 

Chair of Oral Examination

• Dr. Iain McKechnie, Department of Anthropology, UVic

   

Abstract

The primary objective of this dissertation was to develop valid and reliable monitoring of soccer player maximal acceleration and speed sprinting capacity in an inobtrusive manner. To accomplish this goal, we studied methods that used observational data collected with sensors that are typically worn in every training and match in international level women’s soccer (i.e., invisible monitoring).  

First, to achieve this objective, it was necessary to evaluate the validity and reliability of different athlete athlete-worn sensors (global navigation satellite systems [GNSS]) against gold standard devices for measurement of instantaneous velocity (radar) and standard time-distance sensors (timing gates). These sensors can be used to model and calculate various force-velocity (FV) metrics related to horizontal acceleration, velocity, force, power, and efficiency of the oriented force throughout linear sprinting protocols. It was demonstrated that GNSS sensors could be used to model linear sprints with adequate reliability and validity depending on the GNSS sensor used. 

Second, to provide a more comprehensive assessment of the state of GNSS for FV linear sprint profiling, a systematic review with quantitative (meta-analysis) and narrative analysis was carried out on the literature comparing GNSS to radar and laser. It was found through this process that GNSS can be valid and reliable, however, there are several methodological challenges that need to be considered which informed our research designs going forward and allowed for improvement of modeling recommendations for future research and application. 

Third, although having the ability to model linear FV sprint capacity in the field using GNSS sensors is impactful, it still requires a degree of protocol standardization and requires anthropometric and environmental metrics to be collected for effective modelling. Therefore, we evaluated a novel athlete monitoring approach that requires no dedicated sprint testing or additional metrics to be collected. Acceleration-speed (AS) profiles using regular training and game data were compared to FV profiles in an elite women’s soccer cohort to determine whether AS profiles could provide practitioners with similar information to FV profiling without the need for any isolated sprint testing. It was found that within a 4-week national team camp, it was possible to construct valid AS profiles that can inform practitioners on the acceleration and speed of the athletes in aggregated data sets of training and games. Since this data can contain outliers, multiple outlier techniques to construct AS profiles were also evaluated and shown to be an important aspect to consider. 

Fourth, since AS profiling was determined to be valid with 4-weeks of data which may not be ecologically valid due to variations in national team camp lengths (typically a week or longer), it was then necessary to determine the minimal number of events necessary to construct valid AS profiles. Therefore, an optimization approach whereby all possible combinations of 19 training or game events were performed, and it was determined that nine events were necessary for a valid profile to be constructed, and that the inclusion of maximal sprint efforts can improve the reliability and (reduce) the number of events necessary. 

In the fifth and final study, we conducted a larger scale analysis on 3-years of women’s national team training camp and international match data. Since the AS profiles represent the athletes AS capacity during and surrounding the camp, we could use the AS regression as a reference to normalize the maximal speed and acceleration points across velocity bands typically used in soccer (i.e., low, moderate, high, and very-high speed running). This resulted in the introduction of novel invisible monitoring metrics such as normalized acceleration and power, which allowed for the quantification of athlete’s relative effort in matches compared to their physiological and biomechanical maximal capacity. These data were then used to demonstrate the differences and possible benefits of normalization or non-normalization of acceleration and power and how it varies based upon position, goal differential, and match half. 

Altogether, the findings from this thesis suggests that it is possible to reliably generate “invisibly” monitored acceleration-speed profiles in international women’s soccer contexts using GNSS technology. Further, AS profiling provides a benchmark for physical and tactical staff with which can be compared to absolute athlete data and may inform rehabilitation, strength & conditioning programs, training sessions and match tactics. Further research is necessary to determine effective application in other team sports, standardization and optimization of data processing algorithms as well as refinement of technological accuracy and precision to best support AS profiling.