A comparison between manual and automated event detection for shuffle, deceleration and run cut tasks using motion capture.

Background: Increased adolescent sports participation lead to a rise in sports-related injuries. These injuries impact athletes' health and performance, necessitating improved injury prevention methods. The shuffle, deceleration, and run cut tasks are commonly used in injury prevention protocols to elicit improper movement mechanics. Recent literature examined the use of an automated event detection algorithm to improve the accuracy of 3-dimensional motion capture data processing techniques. Manual and automated event detection methods were compared during these tasks in two different groups of participants.

Methods: Thirty healthy controls and thirty adolescents following anterior cruciate ligament reconstruction, performed a shuffle, deceleration, and run-cut task in a motion capture lab. Specific timepoints of the tasks were manually identified by two raters and automatically detected by custom MATLAB algorithms. Intra- and inter-rater reliability, differences in event timings, and task performance were compared.

Findings: Significant differences in event timings were found between manual and automated methods, particularly with events identifying the lateral, forward, or vertical position of the participant with the absolute difference ranging from 4.7 to 13.5 frames across all three tasks. The identification of the first and last timepoints the foot is contacting the ground were similar between methods.

Interpretation: The results of this study indicate that automated event detection is a more reliable method of identifying timepoints assessing participant's movement, highlighting its value in clinical and research settings. Automated event detection may improve injury risk assessments by minimizing user variability and offering consistent event identification across diverse movement tasks.