Evaluating Assistive Technology Outcomes in Boccia Athletes with Disabilities Using AI-Based Kinematic Analysis

Area of Science:

• Sports Science
• Biomechanics
• Artificial Intelligence

Background:

• Evaluating assistive technology outcomes is crucial for adaptive sports.
• Elite boccia athletes with disabilities require precise performance metrics.
• Current methods may lack the granularity for detailed biomechanical analysis.

Purpose of the Study:

• To explore AI's role in evaluating assistive technology outcomes for elite boccia athletes.
• To develop and validate an AI-driven motion analysis framework for adaptive sports.
• To assess the precision of AI pose estimation techniques in capturing seated joint kinematics.

Main Methods:

• Integrated OpenPose, ViTPose, and Lifting for 2D/3D seated joint kinematics estimation.
• Analyzed match footage from 12 elite boccia athletes across five biomechanical phases.
• Applied Principal Component Analysis to movement data for strategy and variability assessment.

Main Results:

• ViTPose demonstrated superior joint detection accuracy (85%) compared to OpenPose (79.5%).
• Lifting reduced 3D joint position error by 16%, enhancing estimation precision.
• Overhand throws exhibited greater movement consistency than underhand techniques.

Conclusions:

• The AI pipeline offers an interpretable and scalable method for measuring boccia performance.
• Findings support evidence-based coaching, athlete classification, and inclusive assistive technology design.
• AI-enhanced motion analysis significantly advances the evaluation of adaptive sports outcomes.