Researchers at the University of Pennsylvania have created CAMI, an AI system that improves autism screening accuracy by analysing children’s motor imitation skills using simple video recordings, potentially transforming early diagnosis practices.
Researchers at the University of Pennsylvania have developed an artificial intelligence tool they say could make autism screening quicker and more consistent for children by focusing on a specific behavioural marker: motor imitation. The system, called CAMI, or Computerised Assessment of Motor Imitation, records a child copying movements shown on video and then uses machine learning to judge how closely the child matches them, according to reporting by 6abc and The Engineer.
The approach matters because imitation difficulties are often seen in children on the autism spectrum, yet standard assessments can depend heavily on human observation, parental questionnaires and specialist equipment. By using ordinary video rather than motion-capture hardware, CAMI is designed to reduce the cost and complexity of early screening and may make objective assessment more accessible in clinics, schools and family settings.
In a study of 183 children aged 6 to 13, the system produced diagnostic accuracy of about 80% to 85%, according to the researchers, whose work was published in IEEE Transactions on Biomedical Engineering. That is enough to make CAMI a potential aid to clinicians, but not a substitute for a full diagnosis, which still requires broader evaluation of behaviour, development and history.
Kaleab Kinfu, a PhD candidate involved in the project, told 6abc that the system is trained to ignore distractions such as lighting and camera placement and instead focus on the quality of imitation itself. René Vidal, senior author of the study and director of the Innovation in Data Engineering and Science Initiative, told The Engineer that the tool could offer a more detailed read of motor behaviour than many existing standard tests. The researchers now plan to test CAMI in larger and more diverse groups, including adults, and to improve its accuracy further.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
