Digital Autism Screening Tool Could Enhance Early Identification
• Research Highlight
A tablet-based screening tool that analyzes children’s behavior in response to specific video clips shows promise for enhancing early autism screening, according to a study supported in part by the National Institute of Mental Health. While early autism screening typically depends on parent questionnaires, data suggest the accuracy of these assessments may vary across settings and populations. Objective measurement tools, including digital technologies, could help improve screening in real-world settings and reduce disparities in early screening and identification.
What did the researchers do?
In the study, researchers Geraldine Dawson, Ph.D. , Guillermo Sapiro, Ph.D. , and colleagues at the Duke Center for Autism and Brain Development tested a tablet-based app called SenseToKnow. The app uses the tablet’s camera to capture a variety of child behaviors, including gaze patterns, facial expressions, head movements, blink rate, and whether the child responded to their name. According to the researchers, this multimodal approach allows them to capture the range of behavioral variations that children with autism may show.
During routine health care visits, toddlers watched specially designed video clips while the device recorded their behaviors and quantified them using computer vision, a type of artificial intelligence. The app then used machine learning to analyze the behavioral data, providing a diagnostic classification and a prediction confidence score indicating the reliability of that classification. The app also produced a quality score that indicated whether the app was administered correctly.
Study participants included 475 toddlers, ages 17 to 36 months. Of these toddlers, 49 later received an autism diagnosis and 98 later received a diagnosis of developmental delay and/or language delay without autism.
What did the researchers find?
Overall, the app showed high accuracy for classifying children with autism compared to neurotypical children, and even higher accuracy when the analyses included only the results that had high prediction confidence scores. Classification accuracy remained high when the analyses included data from children with developmental delay and/or language delay.
The app correctly classified nine children with autism who were not correctly identified using a standard early autism screening tool, the Modified Checklist for Autism in Toddlers (M-CHAT-Revised with Follow-Up). Classification accuracy increased further when the researchers combined the app analyses with input from the M-CHAT screening tool.
Importantly, classification accuracy was consistent regardless of the child’s sex, race, ethnicity, and age. According to the researchers, these initial findings suggest that objective digital screening tools may help reduce existing disparities in early autism screening, although more work is needed to establish the app’s performance across diverse groups.
What do the results mean?
Advantages of the SenseToKnow app include its usability in real-world settings and the fact that it provides actionable information. For example, a low quality score indicates the app wasn’t administered correctly and may need to be re-administered. On the other hand, a high prediction confidence score lends weight to the classification results and can help identify toddlers who are likely to benefit from further screening and evaluation.
Dawson and colleagues are now evaluating SenseToKnow in a variety of contexts. In another NIMH-funded study, the researchers are examining accuracy when parents administer the app at home on their own devices. They are also exploring whether the app can be used to detect early behavioral signs of autism in infants as young as 6-9 months.
The researchers emphasize that they do not intend for SenseToKnow to be the only data source for diagnosis. Rather, they envision autism screening as a multi-part process that includes parent-report questionnaires, objective digital screening tools, and other data sources such as electronic health records. They also note that screening is one part of a broader clinical pathway that includes provider training, careful implementation, and built-in links to services, supports, and interventions.
“We conclude that quantitative, objective, and scalable digital phenotyping offers promise in increasing the accuracy of autism screening and reducing disparities in access to diagnosis and intervention, complementing existing autism screening questionnaires,” Dawson and colleagues write.
Reference
Perochon, S., Di Martino, J.M., Carpenter, K.L.H. 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