Cerebral palsy detection from infant using movements of their salient body parts and a feature fusion model.

Early detection of cerebral palsy is crucial, yet existing methods face limitations due to restricted datasets and the inherent complexity of the condition. This study addresses these challenges by enhancing a pose-based feature fusion framework with body part-based analysis, aiming to improve the accuracy and robustness of cerebral palsy detection in infants. Current literature predominantly focuses on either body-part information or frequency and motion data, neglecting the potential benefits of combining both. To bridge this gap, this research integrates a body part-based ensemble framework with the existing pose-based model. This approach involves a comprehensive analysis of body part-based features in both the time and frequency domains, capturing subtle nuances in movement patterns that may be indicative of cerebral palsy. The proposed integrated framework achieves an impressive accuracy of 98.94%, demonstrating its potential to significantly enhance early diagnosis and intervention efforts for cerebral palsy. By providing clinicians with a more accurate and reliable tool for early detection, this research can ultimately improve the quality of life for children with cerebral palsy. [ABSTRACT FROM AUTHOR]