Prediction of autism symptoms at 36 months from integration of multi-domain clinical data at 8 and 14 months in high-risk siblings and controls

Although symptoms of autism emerge in the first years of life, a robust clinical diagnosis is typically achieved not earlier than 3 years of age. To develop effective strategies of early intervention, it is necessary to ... [ view full abstract ]

Although symptoms of autism emerge in the first years of life, a robust clinical diagnosis is typically achieved not earlier than 3 years of age. To develop effective strategies of early intervention, it is necessary to improve our understanding of the underlying mechanisms. Longitudinal studies on infants at high familial risk for autism helped to find early markers for the disease, yet limited to just one particular domain. The integration of complementary information from different domains can help to clarify previous findings and shed light on the interactive mechanisms underlying autism.
In this study we integrated clinical data from behavioural and cognitive tests at early age to predict autism symptoms. Early developmental trajectories were characterized in a cohort of 150 high-risk siblings (HR) and 66 low-risk controls (LR). We integrated scores from Mullen Scales of Early Learning (MSEL), Vineland Adaptive Behavior Scales (VABS) and Autism Observation Scale for Infants (AOSI) at approximately 8 and 14 months of age separately in a support vector machine (SVM) model trained to classify subjects basing on the Autism Diagnostic Observation Schedule (ADOS) score at 36 months.
The combination of clinical measures provided a good prognostic classifier, more promising than single measures alone. We show that integrated data perform better than single measures in predicting autism symptoms at 36 months, and that for some measures, scores at 8 months perform better than scores at 14 months. This suggests the relevance of studying trajectories of scores over time rather than predicting from a single time point.

The research leading to these results has received funding from the European Commission within the Horizon 2020 Framework, under the Marie Sklodowska-Curie ETN grant “Integrated view on disruptions of early brain... [ view full abstract ]

The research leading to these results has received funding from the European Commission within the Horizon 2020 Framework, under the Marie Sklodowska-Curie ETN grant “Integrated view on disruptions of early brain development” (BRAINVIEW).