Integrating Artificial Intelligence to Enhance Scoring Reliability and Efficiency for SHAPE-UP

Theme: Advancing Pediatric Assessment Through AI-Driven Innovation.

The SHAPE-UP measure is a clinically validated, AMC-specific tool designed to evaluate upper-limb functional abilities in children with Arthrogryposis Multiplex Congenita. While SHAPE-UP is highly relevant and meaningful for clinicians, scoring requires careful observation, time, and consistency across raters and centers. To address these challenges, this project integrates artificial intelligence and computer vision to automate and enhance the scoring process.

Using video recordings of children performing SHAPE-UP tasks, the AI model identifies key body landmarks, tracks movements, and analyzes functional performance. The system then generates automated scoring predictions aligned with the SHAPE-UP categories. Early work demonstrates that AI can reliably detect joint angles and movement patterns, even in complex scenarios such as atypical limb positions, overlapping joints, or non-standard compensatory strategies.

This project aims to:
• Improve scoring reliability across clinicians, sites, and time points
• Reduce scoring time and clinician burden
• Increase access to high-quality assessment in remote or busy clinical settings
• Support large-scale research, where manual scoring of hundreds of videos is not feasible

The AI-enhanced SHAPE-UP system represents a major step toward standardized, efficient, and scalable functional assessment in pediatric rehabilitation. By combining clinical expertise with advanced machine learning, this project seeks to transform how upper-limb function is evaluated in children with AMC, enabling informed treatment planning and better monitoring of outcomes over time.

Funding:

Quebec Network for Intersectoral Research in Sustainable Oral and Bone Health awarded to Yongni Zhang, PhD PT (2026)