Assessing infant risk of cerebral palsy with video-based motion tracking
Abstract: Cerebral Palsy (CP) is a common (about 1 in 500 children) health condition caused by abnormal brain development
that affects the ability to control movement. Early risk assessment happens through the General Movements Assessment (GMA), a test administered by trained clinicians at 3-4 months of age that has high predictive value for CP. With recent improvements in video-based motion tracking, automated risk assessment for CP based on the GMA is being explored. However, studies generally have used small datasets or were limited in terms of methodological rigor. Here we acquired a large dataset (1060 infants) of videos from a clinical population with elevated CP risk. In a preregistered pipeline using a lock-box set that was not used before algorithm submission we find that our machine learning predictions are highly predictive of the clinician-assessed GMA (AUC=0.79). Given its low cost, our video-based approach may be useful for clinical screening applications, particularly in low-resource settings.
Loading