Automating General Movements Assessment with quantitative deep learning to facilitate early screening of cerebral palsy

The Prechtl General Movements Assessment (GMA) is increasingly recognized for evaluating the developing nervous system and predicting motor dysfunctions, particularly cerebral palsy (CP). However, the need for highly trained professionals limits its early screening adoption. This study proposes a deep learning-based motor assessment model (MAM) that combines infant videos and basic characteristics to automate GMA at the fidgety movements (FMs) stage. MAM achieves strong performance (validation AUC 0.967), adheres to GMA principles with robust interpretability, and accurately identifies FMs within videos with substantial agreement to expert assessments. Using the predicted FMs frequency, a quantitative GMA method attains an AUC of 0.956 and improves the diagnostic accuracy of GMA beginners by 11.0%. MAM may streamline early CP screening and advance video-based quantitative medical diagnostics.