Digital Behavioral Phenotyping Detects Atypical Pattern of Facial Expression in Toddlers with Autism.

The study had only 104 children, and the ASD group had 22 children. There was no separate training set and testing set because the sample was small. The computer vision model was trained on adult faces, not toddler faces. The Other expression category includes all non-positive or negative expressions, so it may combine several facial movements. The comparison group included 74 typically developing children and 8 children with a non-ASD delay, but the delay group was too small to test separately. Only 13 children with ASD received Mullen cognitive testing, so cognitive differences in the comparison group may not be fully known. The ASD group was, on average, 4 months younger than the comparison group. The system did not track eye movements, only whether the face was oriented toward the stimulus. Future validation in larger datasets is underway.

Guillermo Sapiro has received basic research gifts from Amazon, Google, Cisco, and Microsoft and is a consultant for Apple and Volvo. Geraldine Dawson is on the Scientific Advisory Boards of Janssen Research and Development, Akili, Inc., LabCorp, Inc., Tris Pharma, and Roche Pharmaceutical Company, a consultant for Apple, Inc, Gerson Lehrman Group, Guidepoint, Inc., Teva Pharmaceuticals, and Axial Ventures, has received grant funding from Janssen Research and Development, and is CEO of DASIO, LLC. Dawson receives royalties from Guilford Press, Springer, and Oxford University Press. Dawson, Sapiro, Carpenter, Hashemi, Campbell, Espinosa, Baker, and Egger helped develop aspects of the technology used in the study. The technology has been licensed, and Dawson, Sapiro, Carpenter, Hashemi, Espinosa, Baker, Egger, and Duke University have benefited financially.

Do not conclude this tablet facial-expression tool can diagnose autism or replace clinical screening. The study is preliminary, used a small sample, and had no separate validation set.