A new AI system called CAMI at the University of Pennsylvania offers a quicker, more objective approach to autism assessment by analysing how children imitate movements, potentially transforming current diagnostic practices.
Researchers at the University of Pennsylvania are testing an artificial intelligence system they say could make autism assessments quicker and more objective by focusing on how children imitate movement rather than relying solely on lengthy behavioural reviews.
The tool, called CAMI, short for Computerised Assessment of Motor Imitation, asks children to copy a set of simple actions shown in a standard video. The system then analyses the quality of the imitation, which researchers say is a behavioural marker linked to autism. Kaleab Kinfu, a doctoral researcher involved in the work, said the model is designed to filter out distractions such as camera angle and lighting and concentrate on the child’s movement patterns. Rene Vidal said the approach offers a more detailed motor assessment than traditional standardised testing and could eventually become one part of a broader diagnostic process.
In a clinical study involving 183 children aged 6 to 13, the researchers said CAMI achieved diagnostic accuracy of about 80% to 85%. Similar work described by Nottingham Trent University and the Kennedy Krieger Institute in January found that a one-minute version of CAMI using motion-tracking technology correctly identified autism in 183 children aged 7 to 13 with about 80% accuracy. A separate Penn State pilot study, involving 23 children with autism and 17 typically developing children, found that motor imitation could be quantified automatically with a single 2D camera, suggesting the method could scale more easily than some current assessments.
The researchers stressed that the system is not intended to replace doctors. Instead, they say it could help clinics, schools and families by providing faster support while keeping physicians at the centre of diagnosis. The next stage of the project will look at larger and more diverse groups, including adults, while also trying to improve accuracy.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
