Disentangling heteRogeneity in autism: from EArly behavioral markers to individual treatment responses through a Machine learning approach (DREAM)

The DREAM project aims to explore, through new machine learning and deep learning approaches, the heterogeneity of autism under the lens of early motor and gestural development, trying to identify: (i) new early behavioural markers of autism risk; (ii) possible subtypes of autism that share similar behavioural characteristics; (iii) socio-communicative developmental trajectories following individualised psycho-educational interventions. In particular, the project will analyse the development of early motor precursors, such as actions and gestures, and their link with shared attention and language. These markers will be rigorously coded using a micro-analytic system that will be applied to video recordings of spontaneous caregiver-child play interactions.