Movement-related differences in infants with high and low likelihood of autism: A longitudinal study using automated kinematic tracking

Aim: While motor differences are increasingly recognised as early signs of autism, research on the developmental trajectories of specific kinematic features remains limited. This study aimed to examine the early emergence of specific kinematic features (duration, range, velocity, acceleration, entropy, and jerk) in infants with high and Low-Likelihoods of developing autism.Method: We longitudinally evaluated weekly, home-recorded videos of naturalistic, spontaneous movement from 60 infants (comprising 705 total videos) across the first six months of life. Automated pose estimation (computer vision) was utilised to extract objective kinematic data from the shoulders, elbows, hips, and knees, while controlling for environmental and postural variables.Results: Analysis revealed significant group differences in the variability of movement entropy. Specifically, the variability of kinematic entropy was significantly lower in the high-likelihood cohort compared to the low-likelihood cohort throughout the first six months of infancy.Interpretation: These results indicate a reduced complexity in early spontaneous motor behaviour among high-likelihood infants. Detecting these early kinematic differences via scalable computer vision may provide a quantifiable, objective marker of the neural atypicalities associated with autism prior to the onset of traditional diagnostic behaviours.