This multicenter diagnostic and prognostic accuracy study tested whether quantifying selective motor control from brief spontaneous infant movement recordings provides diagnostic and prognostic information for cerebral palsy (CP) beyond established movement assessments. The authors aimed to determine whether a 1-minute video scored with a targeted appraisal could discriminate CP diagnosis, predict functional severity by Gross Motor Function Classification System (GMFCS) level, and identify motor distribution (including unilateral CP).
The study included 302 infants drawn from cohorts in Australia and the United States. Of these, 151 infants had a diagnosis of CP. Spontaneous movement recordings were obtained between 10 and 20 weeks corrected age. The authors included all eligible infants with CP and selected a random comparison sample of infants without CP.
Recordings were evaluated using three movement-based tools:
Scores from these instruments were compared for diagnostic discrimination of CP and for prognostic discrimination by later functional outcome.
Primary outcomes were assessed at age 2 years or older and included:
Follow-up at or beyond 2 years provided the reference standard for diagnostic and prognostic accuracy analyses.
BabyOSCAR demonstrated high diagnostic discrimination for CP overall, with an area under the curve (AUC) of 0.98. This discrimination extended to infants who were later classified as GMFCS level I, indicating sensitivity to milder motor impairments.
Among infants with CP, BabyOSCAR differentiated those who later had better versus worse motor function: it discriminated GMFCS levels I–II from III–V with an AUC of 0.89.
BabyOSCAR's measure of absolute asymmetry also provided strong discrimination for motor distribution: it distinguished unilateral CP from all other infants with an AUC of 0.90.
For context, the established movement assessments also showed diagnostic discrimination in this dataset: MOS-R yielded an AUC of 0.94 and GMA an AUC of 0.86. These results indicate that quantifying selective motor control with BabyOSCAR may provide complementary information to MOS-R and GMA rather than replacing them.
The authors interpret these findings to mean that brief recordings of spontaneous infant movement, when quantified for selective motor control, may offer early, complementary diagnostic and prognostic information about cerebral palsy, including likely functional level by GMFCS and motor distribution (e.g., unilateral CP).
Because the study used recordings between 10 and 20 weeks corrected age and outcomes at 2 years or older, the reported discrimination suggests potential for earlier risk stratification than waiting for later clinical milestones. The authors present BabyOSCAR as an adjunctive tool that may enhance early identification and prognostic conversations but do not claim it should replace established assessments.
Clinicians should note that the report is a preprint and has not been peer reviewed; therefore, findings should not be used yet to guide clinical practice without further validation and peer-reviewed publication.
Ethical approval for the work was provided by the Institutional Review Board of Lurie Children’s Hospital and by The University of Queensland ethics committee. The authors confirmed that all necessary participant consent procedures were followed and that relevant reporting guidelines were used.
A competing interest statement reports that authors Peyton, Moulton, and Bos are members of the General Movements Trust Speaker's Bureau.
The source notes that all data produced in the study are available upon reasonable request to the authors.
The authors declared funding from United Cerebral Palsy. The manuscript is posted as a medRxiv preprint and has not undergone peer review. The DOI assigned in the source is https://doi.org/10.64898/2026.08.24.26361217. The preprint explicitly states it should not be used to guide clinical practice pending peer review and further validation.
In this multicenter cohort of 302 infants, a 1-minute spontaneous movement recording scored with BabyOSCAR showed high diagnostic discrimination for cerebral palsy (AUC 0.98) and provided prognostic discrimination for later GMFCS level (AUC 0.89) and motor distribution (absolute asymmetry AUC 0.90). Established assessments (MOS-R AUC 0.94; GMA AUC 0.86) also demonstrated diagnostic value. The authors conclude that quantifying selective motor control from brief videos may provide complementary early information about diagnosis, functional level, and motor distribution, but the findings are reported as a preprint and require peer review and wider validation.