A digital healthcare startup is using an artificial intelligence (AI) platform to help aid in the detection of autism. Palo Alto, CA -based Cognoa said FDA has classified the algorithm-powered solution as a Class II diagnostic medical device.
The FDA designation gives the company a path to get full clearance as a medical diagnostic for autism.
“The goal actually is this summer to submit to the FDA to get full clearance under a de novo as a medical diagnostic,” Sharief Taraman, vice president of medical at Cognoa, told MD+DI. “I think we should be able to get that at the end of the year or by 2019.”
The company said the average age of autism diagnosis is 4.1 years, which falls after the critical early intervention window when therapy has its greatest impact. Cognoa's first application has been clinically validated to identify autism in children as early as 18 months of age.
“Getting that diagnosis early is really important because it really does change the trajectory of what’s going to happen with the patient,” Taraman said. “What Cognoa is doing is moving this waiting game and referral game for most of the cases … to [the point of] getting the diagnosis right away and not a year later.”
Here’s how Cognoa’s solution works. Parents would answer a questionnaire and submit a video sample of the child. That information then comes to the pediatrician and primary care physician to determine if the child has autism. The information is processed through Cognoa’s app-based machine learning software.
Using data from the diagnosis, the algorithm-powered application recommends personalized, evidenced-based activities such as behavioral interventions parents can do at home to best support their child’s development.
“We’re enabling the pediatrician to do what they need to do, without increasing the burden on them,” Taraman said. “We do it in a highly sensitive and highly specific way. Our most recent version is around 90% and is in the mid-80s for specificity, which is very good.”
Cognoa was founded in 2014 and its technology was originally developed at Harvard and Stanford’s medical schools. The app was built on compiling data from about 10,000 children.
The company raised $20.4 million to date and plans to take up funding for additional rounds.
“Our hope as we move forward and raise additional funds is to expand out that algorithm to other developmental or behavioral conditions,” Taraman said. “Our hope is to also submit our first digital therapeutic intervention for [autism patients] as well.”
The machine learning and AI space have seen an increase in regulatory activity lately.
Earlier this month, FDA cleared stroke detection software from San Francisco-based Viz.ai. The company’s Contact application is a type of clinical decision support software designed to analyze CT results that could notify providers of a potential stroke in their patients.