A research team from the Human-Tech Institute has developed a new system that leverages virtual reality and artificial intelligence to detect Autism Spectrum Disorder (ASD) in early childhood—with an accuracy rate of over 85%. This marks a notable improvement over traditional detection methods, which typically rely on manual psychological tests and interviews.
Image Credit: Universitat Politècnica de València
In the study, the researchers analyzed children's movements as they completed various tasks in a virtual reality environment to determine which AI techniques were most effective in identifying ASD indicators.
The use of virtual reality allows us to use recognizable environments that generate realistic and authentic responses, imitating how children interact in their daily lives. This is a significant improvement over laboratory tests, in which responses are often artificial. With virtual reality, we can study more genuine reactions and better understand the symptoms of autism.
Mariano Alcañiz, Director, Human-Tech Institute, Universitat Politècnica de València
The virtual system involves displaying a simulated environment onto a large-format screen or the walls of a room. The camera records the child's movements while they are executing various tasks.
“This method standardizes the detection of autism by analyzing biomarkers related to behavior, motor activity and gaze direction. Our system only requires a large screen and a type of camera that is already on the market and is cheaper than the usual test-based evaluation method. Without doubt, it would facilitate access to diagnosis as it could be included in any early intervention space,” emphasized Mariano Alcañiz.
New Artificial Intelligence Model
According to researcher Alberto Altozano, who developed the AI model alongside Professor Javier Marín, the UPV team used insights from motor data analysis to compare standard AI methods with a novel deep learning approach.
The results reveal that the proposed new model can identify ASD with greater precision and in a greater number of tasks within the VR experience.
Alberto Altozano, Researcher, Universitat Politècnica de València
After processing the child’s movements during the virtual session, the system generates a diagnosis that researchers say is both more accurate and more efficient than current techniques.
Eight Years of Collaboration to Improve Early Detection
This system is the result of eight years of research and collaboration between the Human-Tech Institute and the Red Cenit cognitive development center. Together, they’ve been refining and validating the semi-immersive VR system for early ASD detection.
As part of this work, researcher Eleonora Minissi recently completed her doctoral thesis, which not only validated the system through studies involving children with autism but also assessed the effectiveness of various biomarkers captured during the experience. Her research underscores that while there’s growing interest in social-visual attention in ASD diagnosis, motor patterns have been relatively underexplored.
The researcher concludes that the "ease with which this data can be collected and its high effectiveness in detecting autism make the motor activity a promising biomarker".
The latest findings from the Human-Tech Institute also suggest that the new AI model can be adapted to analyze motor behavior in other tasks. “This opens the door to future explorations of the motor symptomatology of autism such as: what are the motor characteristics of autistic children when walking or talking?” added Alcañiz.
Journal Reference:
Altozano, A., et al. (2025). Introducing 3DCNN ResNets for ASD full-body kinematic assessment: A comparison with hand-crafted features. Expert Systems with Applications. doi.org/10.1016/j.eswa.2024.126295