Will artificial intelligence soon play a part in diagnosing autism spectrum disorder?
For children with autism and their families, early diagnosis and interventions can play a big role in improving development and long-term health. An early diagnosis typically results in earlier access to clinical, behavioral, educational, and social services that a child and family may need. But many families—particularly those in rural, low-income, and minority communities—face long delays in receiving diagnoses and services. These delays can greatly affect the progress of a child’s behavioral and brain development with life-long consequences.
A quarter of children under age eight living with autism are currently undiagnosed, according to a recent study by Rutgers University Professor Walter Zahorodny, and most of them are Black or Latino. Meanwhile, despite mounting evidence that autism is more prevalent in girls than once believed, boys are still four times more likely to receive a diagnosis than girls, and boys are also diagnosed earlier than girls. Across race and ethnicity, socioeconomic status, geography, and gender, the story is the same: children from less advantaged and minority groups receive a diagnosis and begin individualized treatment later than Caucasian, male, and wealthier children.
Despite these realities, I am optimistic that the healthcare community is increasingly aware of the existing disparities and is ready to embrace technology designed to tackle these challenges head-on so that we can improve the lives of children and families living with autism.
Behind the disparities
As Boston University-based pediatrician Sarabeth Broder-Fingert points out, most autism research has focused on “white, higher-income children and families”, and primarily young white males. This means that when children, particularly non-white males, are evaluated for autism, assessments are largely based on data that isn’t always representative of all children. This has real-world consequences for non-white children and their families.
Aggravating the challenges are disparities in access to healthcare resources, differing levels of education and broad understanding of autism, language barriers, and more. Figuring out the complex healthcare system is difficult for families, especially those with limited resources.
In an ideal world, pediatricians—who are generally the doctors that children and families see most often—would have more tools and resources to respond to concerns for autism and/or developmental progress of children, in order to coordinate appropriate care in a timely, effective manner.
Click here to find out more
Can AI technology help make this a reality?
There’s been plenty of talk about the potential of artificial intelligence (AI) in a number of industries. At its core, AI is a way to understand and make sense of a vast amount of information.
In this case, that includes different types of data—video of children at play, answers to questions from parents and healthcare providers, clinical data of children with and without autism—that reflect the many characteristics of the autism spectrum, such as eye contact and a child’s response to social cues and emotional exchanges.
AI does more than simply crunch data, however. Its real value lies in its ability to pinpoint subtle relationships between different data points. By simultaneously analyzing hundreds or even thousands of data points, AI algorithms can identify and predict behavioral patterns that point towards or away from autism.
This helps physicians to make more informed and efficient assessments, in contrast to a scenario where they rely exclusively on their own training, experience, and clinical observation of a child. Such observations could be influenced by the child’s gender, race, ethnicity, economic status, and geographic location.
With the help of AI-based diagnostics, physicians can have the opportunity to make earlier and more accurate diagnoses. As the technology grows increasingly sophisticated, it can also be used to assess verbal abilities and differences in a child’s development.
By utilizing AI-based solutions, we can imagine that diagnostic specialists will be able to see children with more complex presentations of autism sooner because those children with more clear-cut autism diagnoses can be diagnosed and supported more immediately by their pediatricians.
Technology can help create a more streamlined care system. This will benefit children and families, who will be able to access appropriate care sooner. Another consequence could be a more responsive healthcare community better able to serve the needs of patients and their caregivers.
Of course, AI systems are only as good as the data they’re built upon. That is why algorithms must be developed in a way that deliberately takes into account the gender, racial, ethnic, and socioeconomic dynamics of a child, recognizing, for example, that girls with autism show different traits than boys with autism.
By incorporating equally nuanced data, AI can scale this knowledge and experience in a way that reflects highly diverse communities of children and families and make it available to every pediatrician. In this way, children and families will finally have more equitable access to more accurate and efficient diagnoses.
Collaboration and education are key
AI-based diagnostics are set to become an increasingly important tool in clinicians’ toolboxes. But technology in autism is not a silver bullet. New solutions in clinical practice must be accompanied by education and training that better informs and connects families and clinicians alike to the networks and services that will provide ongoing care and improve lifelong outcomes.
Fortunately, the healthcare community is increasingly attuned to the many factors that affect the health of children, families, and communities. Together with technology advancements and a shared focus on better outcomes, we can see to it that every child and family living with autism receives optimal, timely, and responsive care and support—regardless of gender, family background, location, or income.
This article was featured in Issue 125 – Unwrapping ABA Therapy