For families with young children showing early developmental differences, the wait for an autism diagnosis can stretch years. A team at the University of Dubai has built a screening tool designed to shorten that wait, using eye-tracking technology and artificial intelligence to flag early signs of autism spectrum disorder in toddlers.
The gap is stark. Many children show identifiable signs of autism during their first two years of life, yet formal diagnosis often does not arrive until around age four. Those lost years matter. Intervention programmes begun earlier are known to strengthen long-term developmental outcomes, and every month of delay is a month of support a child does not receive.
Additional reference context is available at https://gulfnews.com/uae/health/dubai-researchers-create-ai-tool-for-earlier-autism-screening-in-children-1.500635240.
The new system works differently from the questionnaire-based tools that families typically encounter first. Rather than relying primarily on instruments like the Modified Checklist for Autism in Toddlers (M-CHAT), the technology objectively measures how a child’s eyes move and focus while the child watches short video sequences containing both social and geometric content. Parents complete a developmental milestone questionnaire at the same time, and the AI model draws on both streams of data to produce an early risk assessment.
What sets this approach apart is how it handles raw eye-tracking information. The system analyzes gaze patterns, fixation duration and visual scanning behaviour directly, without first converting that data into images, a step conventional AI models typically require. That streamlined processing means the software runs on tablets and smartphones, removing the need for expensive high-performance computing equipment. For clinics in under-resourced settings, that distinction is not trivial.
The screening itself is brief. The entire process takes two to three minutes and is non-invasive. Researchers envision eventual home-based use that could extend access to early detection well beyond clinic walls. Initial benchmark testing showed accuracy around 96 per cent, which the research team described as encouraging.
The tool is designed as a screening aid and referral prioritization mechanism, not a replacement for formal medical diagnosis. Clinical validation studies are continuing across different healthcare environments to establish its reliability in real-world practice.
By contrast with many research projects that remain confined to a single institution, this one draws on a broad collaboration. The College of Engineering and Information Technology at the University of Dubai leads the effort alongside Emirates Health Services, Al Amal Psychiatric Hospital, and researchers from the University of New South Wales, the University of Wollongong in Dubai and Macquarie University. Funding comes from the Dubai Research, Development and Innovation Grant Initiative, which operates under the Dubai Future Foundation.
The goal, as the team frames it, is to help clinicians identify children who could benefit from specialist evaluation sooner, reducing the years many families currently spend navigating diagnostic pathways while their children wait for access to support services. Additional reporting on the technology’s design and intended applications is available at gulfnews.com/uae/health/dubai-researchers-create-ai-tool-for-earlier-autism-screening-in-children-1.500635240.
Whether the tool can deliver that 96 per cent accuracy consistently across diverse clinical environments and home settings remains the central question the ongoing validation studies will need to answer.