Early Diagnosis and Prognosis of Cerebral Palsy From a 1-Minute Infant Video

AIM: To evaluate whether selective motor control quantified from spontaneous infant movement recordings provides diagnostic and prognostic information for cerebral palsy (CP) beyond established movement-based assessments. METHOD: This multicenter diagnostic and prognostic accuracy study included 302 infants (151 with CP) with spontaneous movement recordings obtained between 10 and 20 weeks corrected age from cohorts in Australia and the United States. All eligible infants with CP were included, and a comparison sample without CP was randomly selected. Recordings were scored using the Baby Observational Selective Control Appraisal (BabyOSCAR), Motor Optimality Score Revised (MOS-R), and General Movements Assessment (GMA). Outcomes at 2 years or older included CP diagnosis, Gross Motor Function Classification System (GMFCS) level, and motor distribution. RESULTS: BabyOSCAR discriminated CP diagnosis (area under the curve [AUC] 0.98), including children later classified in GMFCS level I. Among infants with CP, BabyOSCAR discriminated GMFCS levels I - II from III - V (AUC 0.89). BabyOSCAR absolute asymmetry also discriminated unilateral CP from all other infants (AUC 0.90). Diagnostic discrimination was also observed for MOS-R (AUC 0.94) and GMA (AUC 0.86). INTERPRETATION: Quantifying selective motor control from brief infant movement recordings may provide complementary early information about CP diagnosis, functional level, and motor distribution.