A tablet-based AI app screens for autism by measuring several behavioral signs
SenseToKnow could help eliminate disparities in early autism diagnosis and intervention
Researchers at Duke University have developed an AI-driven app called SenseToKnow, which can run on a tablet to screen for autism in children by assessing various behavioral indicators.
Here are a few key points:
The app analyzes behaviors like facial expressions, gaze patterns, head movements, blink rate, and motor skills through tablet sensors.
Algorithms compare a child's biomarkers to population-level data to detect autism and reports which biomarkers are most predictive.
The app provides scores that evaluate data quality, result confidence, and the probability of autism spectrum disorder. It offers detailed information to healthcare providers for further action.
SenseToKnow is user-friendly, has few hardware limitations, and is accurate across different demographics, potentially reducing disparities in early autism diagnosis and intervention.
In a study, SenseToKnow showed 87.8% sensitivity and 80.8% specificity for autism detection. Combining the app with a standard questionnaire increased the probability of a positive screen leading to diagnosis.
Snippets of videos and activities meant to provoke specific responses in youngsters while the tablet they are playing on captures their behaviors are shown in this video. To detect complicated indicators of autism, the SenseToKnow app uses AI to assess their responses and compare them to a baseline built from a vast database. Credit: Duke University.