By Walker Evans
Early autism screening could be getting a high-tech upgrade, thanks to a new app that uses artificial intelligence and a tablet camera to observe and analyze toddlers’ behavior.
Developed by researchers at Duke University’s Center for Autism and Brain Development, the app—called SenseToKnow—was tested with nearly 500 toddlers aged 17 to 36 months. The tool records how children react to short video clips, capturing aspects such as where they look, how they move their heads, facial expressions, blinking, and whether they respond when called by name.
What makes this tool exciting is its objectivity. Traditional autism screening often relies on parent questionnaires, which can sometimes vary in accuracy. However, SenseToKnow utilizes machine learning to analyze real-time behavioral data, providing a more standardized approach to identifying early signs of autism.
In the study, the app correctly identified children with autism, even catching nine cases that a commonly used screening tool missed. It also maintained strong accuracy across children of different ages, races, and ethnicities, suggesting it may help close the gap in early autism detection for underserved communities.
Even better? When combined with traditional screening questionnaires, the app’s accuracy improved even further.
Researchers emphasize that SenseToKnow isn’t meant to replace traditional methods, but rather to enhance the screening process. It’s part of a broader push toward using digital tools to improve access to early diagnosis and support.
Ongoing studies are now testing whether the app can be used at home by parents, and whether it can pick up signs of autism in infants as young as 6 months.
As the researchers put it: “Objective, scalable tools like this could help reduce disparities in autism diagnosis and connect children with services sooner.”
Source:
Perochon, S., et al. (2023). Early detection of autism using digital behavioral phenotyping.Nature Medicine, 29, 2489–2497. https://doi.org/10.1038/s41591-023-02574-3
