There’s a lot of talk these days about computational medicine, which uses massive amounts of data to train a machine to understand even more than experts or, at least, to identify health-related problems more quickly.
Cognoa, a consumer-focused healthcare outfit, is among the developing field’s biggest proponents. The two-year-old, Palo Alto, Ca.-based company claims it can dramatically speed up the time that it takes parents to identity whether their child has developmental issues, and it can do so by assessing far fewer data points than have been traditionally employed toward the same end.
The company’s story centers on the work of Dennis Wall, an associate professor of pediatrics and psychiatry at Stanford who began looking into the complexity of diagnosing Autism while an associate professor of pathology at Harvard several years ago. Specifically, he learned that the process of better understanding whether a child’s development is on track typically means hours of behavioral examinations by certified practitioners who’ve been trained to perform interview-based analyses with parents or with children directly.
As you might imagine, appointments are hard to get as a result. In fact, the process is so slow, says Wall, that the average age of a child being seen by one of these practitioners is 4.5 years old. That’s not good. By that age, a kid has missed a window of brain plasticity when an intervention can have the biggest impact.
Work by researchers at The New England Center for Children — which studied 83 toddlers diagnosed with autism in the school’s Early Intensive Behavioral Intervention program — underscores the problem. According to their findings, there’s an alarming gulf between the impact that 20 to 30 hours of weekly one-on-one therapy can have on a child who’s under age 2 and one who is 2.5 years old or older. While fully 90 percent of the toddlers in their study aged 2 or younger made “significant gains” in social and communication skills, just 30 percent of children who entered therapy at age 2.5 or older made “significant gains.”
Cognoa says it can get children in front of doctors faster with its deceptively simple app, one that asks parents to answer 15 questions that address a minimum viable set of behaviors that indicate whether their child is at risk of Autism.
How can it boil down the process so drastically? The company says much of its power is rooted in the information that Wall has culled over the years, including from research repositories like the Autism Genetic Research Exchange, Cure Autism Now (later subsumed into Autism Speaks), the Autism Consortium, and the National Database for Autism Research, which is funded by the National Institutes of Health.
Collectively, the repositories feature observations about 10,000 children. It’s always been possible to request access to that information, says Wall, but he claims no one before had tried to combine, synthesize, and analyze the data using machine learning.
