Title:Modeling 3D Infant Kinetics Using Adaptive Graph Convolutional Networks

Abstract:Reliable methods for the neurodevelopmental assessment of infants are essential for early detection of medical issues that may need prompt interventions. Spontaneous motor activity, or `kinetics', is shown to provide a powerful surrogate measure of upcoming neurodevelopment. However, its assessment is by and large qualitative and subjective, focusing on visually identified, age-specific gestures. Here, we follow an alternative approach, predicting infants' neurodevelopmental maturation based on data-driven evaluation of individual motor patterns. We utilize 3D video recordings of infants processed with pose-estimation to extract spatio-temporal series of anatomical landmarks, and apply adaptive graph convolutional networks to predict the actual age. We show that our data-driven approach achieves improvement over traditional machine learning baselines based on manually engineered features.