Autism spectrum disorder is difficult to diagnose in children under the age of 5, so early care is rarely applied. To remedy this problem, an interdisciplinary team from the University of Geneva (UNIGE), Switzerland, has developed an artificial intelligence (AI) algorithm based on automated video analysis, which makes it possible to study children’s non-verbal communication anonymously and standardized .
Autism spectrum disorder affects one in 54 children and is characterized by difficulties in social interactions, impaired communication skills, and the presence of repetitive behaviors and restricted interests.
This technology correctly classified 80% of cases from short videos showing a child with or without autism under the age of 5 playing with an adult. These results, published in the journal Scientific Reports , pave the way for a tool for the early detection of autistic disorder.
To do this, they first used a technology called OpenPose, developed at Carnegie Mellon University. This computer vision technology extracts the skeletons of people in motion captured in a video and allows the analysis of gestures eliminating all the characteristics that could be discriminatory (age, sex, setting, etc.), keeping only the skeleton relationships in the space and time. The UNIGE research team then developed their adapted artificial intelligence algorithm to detect autism and tested it in 68 typically developing children and 68 children with autism, all under the age of five.
If the diagnosis is made before age 3, it is often possible to compensate for these developmental delays. Specific behavioral interventions can completely change their skill acquisition trajectories and allow them to integrate into a mainstream school