51³Ô¹Ï

Topic

Objective quantification of physical athlete load in women’s rugby sevens

School of Exercise Science, Physical and Health Education

Date & location

• Tuesday, July 8, 2025
• 9:00 A.M.
• Virtual Defence

Examining Committee

Supervisory Committee

• Dr. Marc Klimstra, School of Exercise Science, Physical & Health Education, 51³Ô¹Ï (Supervisor)
• Dr. Ming-Chang Tsai, School of Exercise Science, Physical & Health Education, UVic (Member)
• Dr. David Clarke, Biomedical Physiology & Kinesiology, Simon Fraser University (Outside Member)
• Dr. Nick Clarke, Director, Varsity Performance Sport, UVic Athletics (Outside Member)

External Examiner

• Dr. Jonathon Fowles, School of Kinesiology, Acadia University

Chair of Oral Examination

• Dr. John Burke, Department of Biochemistry and Microbiology, UVic

Abstract

This dissertation investigates objective factors that objectively quantify physical athlete load in women's rugby sevens.

I examined the use of objective metrics to impute missing rating of perceived exertion (sRPE) data from elite international matches. Despite employing various machine learning models, the best-performing random forest classifier achieved 26.5% accuracy, indicating that sRPE, and potentially session rating of perceived exertion (sRPE-CL) may be unreliable measures of athlete physicalworkload, suggesting the need for alternative metrics.

I explored the relationship between contact events and sRPE, as contact has not been previously considered in workload quantification. A linear regression incorporating playing time and contact explained 30.6% of sRPE variance, indicating that contact is associated with perceived exertion.

Building on this knowledge, I developed a speed-deceleration-contact (SDC) model that included athlete mass, number of contacts, speed, and acceleration data to assess athlete workload (sRPE-CL. This model accounted for 48.7% of sRPE-CL variance.

I analyzed the influence of match-specific factors on sRPE-CL the SDC model, and mechanical work. Results showed that sRPE-CL is highly variable, influenced by contextual factors such as score differential, match outcome, match category (e.g., medal final vs. pool match), opposition, and player experience. Conversely, mechanical work and the SDC model provided more objective workload assessments, influenced by fewer external variables.

Finally, I assessed the feasibility of imputing sRPE-CL using machine learning. Results demonstrate that sRPE-CL can be estimated using objective metrics, including mechanical work, and the SDC model.

Overall, this research provides alternative, objective strategies for the monitoring of physical athlete loads.