Could this iPhone app transform how we diagnose autism?October 22, 2015
It's 11:30 am on a Monday, we're on the 30th floor of a fashionable New York City hotel, and Ricky Bloomfield is getting excited.
And given what he and his colleagues just showed me, I can't blame him.
"We [took] an entire medical study and put it into an app," says Bloomfield, an energetic pediatrician and the head of mobile technology strategy for Duke Medical Center.
The specific app that Bloomfield has previewed for me is called Autism & Beyond. It's the result of a years-long project led by a team of Duke doctors, researchers, and programmers, including Helen Egger, who's head of child psychology, and Guillermo Sapiro, an engineering professor. It's one of several new apps that Apple is making available through its ResearchKit platform today.
And if the Duke team — and Apple executives, who hand-picked them — get their wish, the app could help transform how autism is diagnosed and treated around the world.
Autism & Beyond works like an elaborate, interactive selfie. The app is set up to play 20-minute videos while using an iPhone or iPad's built-in camera to scan viewers' facial expressions, analyze their microreactions, and then indicate if there's a potential risk of autism.
It's intended for parents to use with their children, who see videos of lights, sounds, and storytellers. The demo I get is far less comprehensive. But it's a good example of how Autism & Beyond is designed to work.
When I smile, the dots that line the video version of my face turn green. When I frown, they shade red.
After the app collects enough of those visual patterns, it's able to offer real-time and evidence-based feedback, such as whether a parent should seek out a doctor based on the child's indicators for autism.
The app was created to provoke the same instinctual responses that a psychologist like Egger wants to gauge in a clinic, as she tries to diagnose autistic children. And the Duke researchers hope that parents will start using Autism & Beyond to build up a video library of a child's reactions, which could help doctors prioritize the most at-risk children and bring them into the office for in-person diagnosis.
(Parents who don't want to have their child's face and features captured on video can opt to just record the patterns of dots.)
"Our goal is to develop a screening, like hearing or eyesight at schools," Sapiro said in a statement. "They don’t get glasses; they get a referral."
Expediting autism diagnoses could have important ramifications for our health-care system: As the total number of confirmed and suspected autism cases has skyrocketed, parents often wait months or years to see a clinician. Egger told me that Duke's waiting list can last a year.
And although autism can be diagnosed in children as young as 18 months, the average age of diagnosis in the United States is more than 4 years old. Overseas, it can be far longer. Those delayed diagnoses have a real cost, from the emotional toll on families to the child's stunted intellectual development.
"I'm often asked if autism is being overdiagnosed," says Steve Silberman, author of NeuroTribes, the new bestseller that traces the history of autism. "For the most part, I think it's still being underdiagnosed, particularly among women and people of color."
Silberman adds that we still don't know about the prevalence of autism in adults.
The Duke researchers hope to go further than speeding up diagnosis: Mining the massive amount of video data collected through Autism & Beyond could reveal new insights and patterns about the behaviors of children with autism.
So far, Duke researchers say that preliminary results are promising; they're presenting a study at a mobile health-care conference this week on the technical feasibility of using an app to measure autism risk behaviors. And every Wednesday morning, a cross-collaborative team pulls up to ensure that their computer software is flagging the same behaviors as human observers.
Apple's involvement in health care
While Duke's facial recognition software is exciting, it's worth noting that a number of companies are developing similar technologies, albeit for different purposes. A firm called Emotient, for example, recently touted its ability to read the reactions of viewers to Super Bowl TV ads.
What makes the Autism & Beyond app especially notable is that Apple is throwing its not-insignificant weight behind the project. The company's executives were "immediately interested" when they first learned about the software six months ago, the Duke researchers say, and asked the team to start developing a ResearchKit app.
Apple's focus on health care is nothing new. The company has spent several years steadily trying to position itself as a player in America's $3 trillion health-care system, primarily by launching three new products. Apple's year-old HealthKit software collects data from a range of health and fitness apps, and transmits it to doctors and hospitals. The Apple Watch is being promoted as a health-care "game changer" for its ability to track users' mobility and basic vitals.
And Apple's ResearchKit portal is designed to help the company carve out a new, unprecedented role: as the go-to platform where medical researchers can post their studies and millions of iPhone and iPad users can come and sign up.
So far, the strategy is working. After Apple launched ResearchKit in March and highlighted a new Stanford heart study, more than 10,000 people signed up for the corresponding ResearchKit app overnight.
"To get 10,000 people enrolled in a medical study normally, it would take a year and 50 medical centers around the country," Stanford cardiologist Alan Yeung told Bloomberg News. "That’s the power of the phone."
But to be perceived as a force in health care, Apple needs more than sheer volume. The company needs results — evidence that its studies and apps are uncovering provocative insights, and leading to real health-care improvements.
Could Autism & Beyond be one of Apple's first big health-care wins?
The Duke researchers are thinking big, at least. They're partnering with researchers in China and South Africa, in hopes of collecting international data. And assuming Autism & Beyond is successful, they plan to launch similar initiatives like "Anxiety & Beyond" and "Temper Tantrums & Beyond."
"Our ambition is to transform how, where, and when we identify, treat, and monitor young children’s development, mental health, and well-being," says Egger. "Our work emerges from our conviction that information science and engineering are the keys to creating transformative change."
Bloomfield says that the project is personal for him: His own autistic daughter wasn't diagnosed until she was nearly 7 years old — which meant years of unnecessary stress, guilt, and lost opportunities for his family.
"We're not going to cure autism," Bloomfield concludes. "But we can help put the structure in place to help manage it and move [treatment] forward. And that's why we're all so passionate about this."