Data gathered using a smartphone game created by Stanford University researchers might help to diagnose autism spectrum disorder earlier and improve treatment.
One in 44 children in the US is diagnosed with an autism spectrum disorder (ASD), according to the Centers for Disease Control and Prevention (CDC) data from 2018.
ASD can be detected in some children as early as 18 months old. By the age of two, the diagnosis is typically considered reliable. However, most children are not diagnosed until they are age four.
Such delay might have long-term negative impacts on children with ASD because early diagnosis and treatment help them develop their full potential.
Using data collected from a smartphone game, Stanford University researchers are training AI models that could eventually allow semi-automated remote diagnosis of ASD.
How does it work?
In the game Guess What? an adult caregiver holds a smartphone to their forehead with an image on the screen and asks a child to mimic it. The adult then guesses what the child is acting out and registers correct answers by tilting the phone forward; incorrect by tilting it back.
Guess What? game sessions can be video recorded and submitted (with appropriate consent and privacy protections) to Stanford researchers.
"If we switch the camera on, and we can give useful prompts to the child, we can challenge them, help them, and capture information as we go," says Dennis Wall, professor of pediatrics, psychiatry and behavioral sciences, and of biomedical data science at Stanford Medicine and an affiliated faculty member with the Stanford Institute for Human-Centered Artificial Intelligence.
For a few years now, Wall and his colleagues have been gathering Guess What? video recordings and using them to develop new ways to diagnose ASD remotely, improve emotion-recognition datasets, track children's progress in recognizing emotions, and ultimately improve ASD treatments.
In their earlier work, researchers found that non-experts were as good as clinicians at reviewing videos of children and labeling meaningful features of ASD, such as certain types of repetitive speech, a lack of eye contact, particular head movements, etc.
Using these labeled videos, researchers trained a model to predict with high accuracy (above 90%) which children are neurotypical or have ASD in another set of videos. Researchers hope their models will eventually be smart enough to diagnose a child with ASD without human assistance.
In a recent paper, researchers used audio parts of video recordings to directly predict ASD without relying on humans. Applying audio-based deep learning methods to 850 audio clips of 58 children, including 20 with ASD, the team achieved 79% accuracy in distinguishing neurotypical children from those with ASD.
No cure, but treatments available
According to the Autism Science Foundation, a non-profit organization, ASD has no cure, but there are several treatment options available:
- Applied behavioral analysis (ABA) therapy focuses on reinforcing behaviors that form the foundation of skill development. It can help promote a child's social development, help them develop daily life skills, and redirect behaviors that can be harmful, like self-injury.
- Speech therapy helps improve a child's communication skills, allowing them to express their needs or wants better.
- Occupational therapy is used to treat the sensory integration and motor deficits associated with ASDs. It can help to teach life skills that involve fine motor movements, such as dressing, using utensils, cutting with scissors, and writing.
- Physical therapy improves gross motor skills and sensory integration issues, particularly those involving the individual's ability to feel and be aware of his body in space. The therapy can teach and improve skills such as walking, sitting, coordination, and balance.