Graduate Thesis Or Dissertation | Associations Between the PlayerLoad™ Metric and Common IMU-based Physical Activity-Related Measures of Human Movement | ID: 1r66j749n | translation missing: zh.hyrax.product_name
Over the last 60 years, supplemental strength and conditioning for improving sport performance has gone from being nearly non-existent to a cornerstone feature of elite collegiate, amateur, and professional athletics. Concurrent with this profession-related evolution, technologies for monitoring and tracking physical performance have become more prevalent and useful. As an example, the Optimeye S5 from Catapult Sports is currently the most widely used sport performance-related inertial measurement unit (IMU) available for monitoring sport-related movement and is used by more than 2,500 athletic teams globally. The primary metric of interest quantified by this device is known as PlayerLoad™. PlayerLoad™ represents a cumulative summarization of changes in acceleration across three axes of measurement. Although somewhat similar in its formulation to commonly used physical activity-related acceleration metrics, such as band pass filtered Euclidean norm (BFEN) and Euclidean norm minus one (ENMO), no published reports have evaluated the associations between PlayerLoad™ and any of the aforementioned measures. We obtained two publicly available data sets (REAListic DISPlacement [REALDISP] and MHEALTH) containing high resolution IMU data. The REALDISP data set contained data from 17 participants (10 males, ages 22 to 37 years-old) completing 33 unique activities while wearing 9 IMUs across the body (upper-back; right and left side: upper-arm, lower-arm, thigh, and calf). The MHEALTH data set contained data from 10 participants completing 11 unique activities while wearing 3 IMUs (right wrist, left pectoral, left ankle). Raw acceleration data collected from the IMUs was used to compute per second summaries of PlayerLoad™, BFEN, and ENMO. Association analyses revealed that PlayerLoad™ was highly correlated with BFEN and ENMO at all evaluated IMU locations (r = 0.742 to 0.983); however, the strength of these associations did significantly vary across many locations. Notably, lower-body IMU measurement locations (r = 0.951 to 0.983) generally produced higher correlations for PlayerLoad™ with BFEN and ENMO than did upper-body locations (r = 0.742 to 0.960). Models were also developed for predicting epoch- and summary-level PlayerLoad™ from BFEN and ENMO. Unbiased pseudo R2 values exceeded 0.879 in all instances indicating high predictive ability for all models. More research is needed to better understand PlayerLoad™ and its potential value for quantifying time-based dosages of human movement.