Research Paper Highlights AI as a Developmental Diagnostic Tool

AI Robot Photo by Franck V. on Unsplash

Digital diagnostics firm Cognoa has distinguished itself as a Silicon Valley company founded on improving children's’ lives through developmental data that empowers physicians and families. Using an artificial intelligence-driven platform now available to clinicians and selected employers, Cognoa works to help in the screening of young children for autism and other developmental delays.

The company’s advisory team includes medical professionals, clinicians, and researchers who are capable of utilizing AI and other developing technologies to facilitate earlier and more accurate diagnosis of developmental delays, making crucial early intervention easier.

Members of Cognoa’s executive and advisory teams contributed research resulting in a paper on the use of machine learning in the timely detection of autism, published in the August 2018 issue of the Journal of the American Medical Informatics Association.

The study of 162 children used two algorithms including a series of questions to caregivers and a tagging device to observe behavior using homemade videos. Researchers combined the results and compensated for data imbalances through innovative engineering and coding functions.

The research paper concluded that an AI-powered app is capable of providing reliable data to screen for autism in settings outside a clinician’s office. The Cognoa model allows parents to conduct at-home screenings that can lead to meaningful conversations with physicians about their children’s development.

The paper noted that its conclusions were based on a limited data set, and encouraged further clinical research on this topic using larger population samples.

Cognoa Teams with LEARN Behavioral to Support Earlier Autism Diagnoses

Cognoa’s AI-based app provides parents with a new tool to foster optimum well-being for their families. Cognoa recently became partners with LEARN Behavioral, which will enable the Cognoa Child Development app to reach the hands of more families.

LEARN Behavioral is a network of providers of healthcare services for children on the autism spectrum and for children with other special needs. Every year, thousands of families nationwide have their children’s needs met through a LEARN Behavioral provider. 

With the Cognoa Child Development app now available to LEARN Behavioral, parents can conduct an initial child development screening to help inform them of their child’s developmental health - as well as risks for development delay - and guide them through towards an expedited diagnosis, if needed. Cognoa’s solution has the potential to improve today’s standard of care where most children are diagnosed with autism spectrum disorder after the age of 4 years - past the optimum window of opportunity for intervention programs to achieve the most success.

About Cognoa

One in six children in the U.S. has some form of development delay. The Centers for Disease Control has determined that the average age a parent notices the signs and symptoms of developmental delay is often as young as 12 months, but they are not diagnosed until 3 or 4. Children who consistently miss milestones in key areas of development (social, emotional, physical, cognitive, language) may be at risk for a developmental delay, including autism.

Cognoa's app helps every parent find out if their child’s development is on track and get the right help as early as possible. A parent answers simple questions and uploads two short videos of their child's behavior. Cognoa’s app then determines whether the child is at risk for developmental delay or autism using machine learning algorithms Dr. Dennis Wall developed at Harvard and Stanford Medical Schools. Cognoa provides a personalized report parents can share with their doctor that identifies specific areas where the child is doing well in addition to areas where the child might need early or rapid intervention. Cognoa also provides personalized next steps, including activities recommendations that integrate on-going monitoring.

Our machine learning scoring system has been validated using several large archival repositories containing data on over 10,000 families with different types of developmental delay and/or autism spectrum disorder. The scoring system has also been validated in a prospective clinical validation study, with two additional multi-site studies underway. The sensitivity and specificity values for the system currently range from 90-97% and 80-94% respectively.