Digital screening app correctly detects autism 88% of the time, NIH-backed study finds

Digital screening app correctly detects autism 88% of the time, NIH-backed study finds

A tablet-based app could help reduce disparities in autism spectrum disorder diagnoses across gender, race and ethnicity, according to a new study funded by the National Institutes of Health.

The app, named SenseToKnow, was developed by researchers from the Autism Center of Excellence at Duke University. As a tool for screening toddlers for autism, it could potentially replace the parent questionnaires that are typically used for diagnosis and which have been found to miss many cases of autism, especially in girls and children of color, the NIH noted in its Monday release about the study.

In contrast to the questionnaires, the SenseToKnow app tasks toddlers with watching short movies and records and analyzes their behavioral responses to the onscreen stimuli—including changes in facial expression, blink rate, head movement, attention span and more.

In the NIH-backed study, which was published in the journal Nature Medicine on Monday, 475 toddlers between the ages of 17 months and 3 years used the SenseToKnow app during a routine well-child visit.

Ultimately, 49 of the children went on to be diagnosed with autism; according to the study results, the app was able to identify those children with 87.8% sensitivity, while correctly ruling out those without autism with 80.8% specificity.

Participants in the study who screened positive for autism spectrum disorder on the app had about a 40% probability of being definitively diagnosed with the condition—compared to the mere 15% probability rate for those who screened positive using the standard questionnaire, per the NIH. That accuracy rate jumped even higher when the two methods were combined: Using the results of the app and parent survey together led to a 63.4% chance that screening positive would lead to an official diagnosis.

Additionally, the study’s findings were consistent among both boys and girls and across a range of races and ethnicities, though specificity was quite a bit lower among Black children than those of other races, at 54% compared to about 85%.

The study’s authors wrote that more extensive research into the app’s performance across race, ethnicity, sex and age differences are already under way, but noted that the early results “[offer] promise in increasing the accuracy of autism screening and reducing disparities in access to diagnosis and intervention, complementing existing autism screening questionnaires.”

“Although we believe that this study represents a substantial step forward in developing improved autism screening tools, accurate use of these screening tools requires training and systematic implementation by primary providers, and a positive screen must then be linked to appropriate referrals and services,” they continued. “Each of these touch points along the clinical care pathway contributes to the quality of early autism identification and can impact timely access to interventions and services that can influence long-term outcomes.”

The Duke researchers aren’t alone in developing new digital tools to improve autism diagnosis rates. Cognoa blazed a new trail last year, earning the first FDA clearance for an app that uses machine learning technology to analyze videos of children performing activities and questionnaires filled out by their parents and doctors to help diagnose the condition in kids between 18 months and 6 years old.

And EarliTec Diagnostics was quick to follow: It racked up its own agency nod this summer, allowing its eye-tracking technology to aid in making diagnoses for toddlers between 16 months and two and a half years old.

Share:
error: Content is protected !!