Duke University’s tablet-based AI “SenseToKnow app” revolutionises autism screening, surpassing traditional methods by detecting diverse behavioural indicators, elevating accuracy in a mere 10 minutes
SenseToKnow uses artificial intelligence to measure and weigh multiple behavioural indicators, delivering scores that assess data quality and result confidence and specify the indicators contributing to its conclusions.
This comprehensive analysis empowers healthcare providers with invaluable insights for assessing and intervening in referred children.
Tablet-based AI app transforming autism screening
The app’s ease of use and absence of hardware restrictions make it accessible for screening in any setting, including a child’s home. Its remarkable accuracy transcends gender, ethnicity, and race, promising to bridge existing disparities in early autism diagnosis and intervention.
While previous research focused on eye movement tracking for autism diagnosis, SenseToKnow surpasses these limitations.
It identifies diverse behaviours such as facial expressions, gaze patterns, head movements, blink rate, and motor skills through an engaging on-screen game, making it one of the earliest indicators of autism.
AI-driven precision
Leveraging the tablet’s sensors and AI to analyse a child’s responses, the app claims the ability to predict the likelihood of an autism diagnosis. By comparing each child’s biomarkers against population-level data, it identifies unique behavioural markers that other screening tests may miss, offering a novel and essential feature.
Enhanced diagnostic confidence
The app continues beyond identifying behavioural indicators; it also provides scores for data quality and result confidence, ensuring that healthcare providers have all the information they need to guide further evaluations and interventions.
In a study involving 475 children, SenseToKnow exhibited an 87.8% sensitivity for autism detection and 80.8% specificity for excluding developmental delay.
By combining the app with a standard parent questionnaire, the probability of a positive screen leading to diagnosis increased to 63.4%, reducing the chances of children slipping through the cracks.
Changing how we screen and support children with autism
The American Academy of Pediatrics recommends autism screening at 18 and 24 months, but current methods relying solely on parent reporting often miss children, especially girls and those from diverse backgrounds. SenseToKnow’s ability to detect autism consistently across demographics complements subjective questionnaires, offering objective tools to narrow the diagnostic gap.
As researchers explore the app’s potential for measuring a child’s progress and evaluating intervention programs, SenseToKnow promises to transform how we screen for and support children with autism.
It enables clinicians to focus on areas needing assistance while highlighting a child’s strengths, ushering in a new era of precision and inclusivity in autism screening and intervention.
