Implementation of a Decision Support System to Enhance Movement Proficiency Assessment in Sport

Background/Objectives: This study aimed to determine the relationships between seven descriptors of movement proficiency used by an expert to grade an athlete's single-leg squat and the overall subjective 'grade' and the ability to predict a 'grade' based on the descriptors. A secondary aim was to determine the relationships between biomechanical data, the expert-defined descriptors, and the subjective 'grade' and its ability to predict the descriptors' presence and the overall 'grade'. Methods: Single-leg squats in 55 male athletes were graded using expert evaluation, synchronized video, biomechanical data, and decision tree and logistic regression analysis. Results: The model that most accurately predicted 'grade' (94.7%) was a decision tree with the descriptors as inputs. The model with biomechanical data for the descriptor 'foot' was the most accurate one (96.3%), followed by 'lumbar' and 'depth' (85.2%), 'knee' (81.2%), 'pelvis/hip' (71.7%), and 'trunk' (62.3%). These accuracies followed similar order to the intra-rater agreement: 'foot' (0.789), 'lumbar' (0.776), 'knee' (0.725), 'depth' (0.682), 'pelvis/hip' (0.662), and 'trunk' (0.637), indicating that 'trunk', 'pelvis/hip', and 'depth' are potentially the hardest descriptors to assess by the expert. Conclusions: The models developed in this study demonstrate that subjective perceptions can be somewhat accurately explained through a small number of biomechanical indicators. The results of this study support the notion that human movement evaluations should consider both subjective and objective assessments in a complementary manner to accurately evaluate an athlete's movement proficiency.