Infant Developmental Assessment Using Video and Eye Gaze Measurement

Objectively assessing neurodevelopment in infancy is important for identifying developmental characteristics such as autism spectrum disorder (ASD) at an early stage and providing opportunities for early intervention. However, the assessment of spontaneous movements (general movements) in neonates relies on observation by trained specialists, and objective, quantitative indicators are needed. In this research, we are developing methods that measure body movements markerlessly from videos captured with RGB cameras and quantify the spatiotemporal patterns of spontaneous movements. Using techniques such as non-negative tensor factorization, we decompose spontaneous movements into interpretable components and analyze how these features relate to later developmental measures using birth cohort data. By combining expert General Movements assessment with automatically extracted motion features, we aim to complement conventional observation-based assessment with quantitative measures and to capture developmental risk more reliably. Beyond movement, we also work on eye gaze measurement, analyzing characteristics of gaze patterns in infancy to capture autistic tendencies. We additionally participate in analyses of the association between prenatal chemical exposure and developmental risk. Technical approaches: - Markerless motion measurement using RGB cameras - Spatiotemporal decomposition of spontaneous movements via non-negative tensor factorization - Automation and quantification of General Movements assessment - Discrimination of autistic tendencies based on gaze characteristics