Can AI Help With Autism Diagnosis?

cognoa-forbes-horizontal, Image From COGNOA

Image credit: Cognoa

Source: COGNITIVE WORLD on FORBES

An Interview with Cognoa’s CEO Brent Vaughan

In the US, one in five children are living with a diagnosable behavioral health disorder; however, only 21% of those children who are diagnosed receive needed treatment. Often, parents have to wait one to three years to receive a proper and accurate diagnosis and an individualized treatment plan for their child. For the last 15 years, the average age of one particular group of patients, those with Autism Spectrum Disorder (ASD), the diagnosis has remained unchanged at 4 years, 4 months. In the first years of life, we are missing the window when treatments have the greatest impact on brain development. In the age of innovation, Artificial Intelligence can help clinicians improve the speed of Autism Diagnosis and allow children and families who live with ASD to begin treatment programs during the critical period of early childhood.

This is why Cognoa started with a tool to accurately and more quickly diagnose Autism. Parents often see the signs and symptoms that end up being diagnosed as autism, Oppositional Defiant Disorder (ODD), Obsessive Compulsive Disorder (OCD) or Attention-Deficit/Hyperactivity Disorder (ADHD) within 18 months. Half of the parents see signs even before this, as early as 12 months of age. The average age of diagnosis for autism, is 4 and ½ years of age, which is when many children begin to receive services. The average age of diagnosis for ADHD is 7 and the average age of diagnosis for anxiety is in the teens. If you look at the data on treatment over the last twenty years, most of these kids can see dramatic improvement if you start to treat them earlier, during the key windows of time when their brains are developing rapidly.

Brent Vaughan, CEO of Cognoa

In early childhood, under the age of 3 years old, children’s brains will establish 90% of the neural connections that they will need over a lifetime. In this critical period of rapid development, our children’s brains have high levels of neuroplasticity. Neuroplasticity is the ability of the brain to change and remodel. For children with ASD, this is a critical period to receive a wide range of therapies that they will need to manage their symptoms. It is estimated that with early diagnosis and treatment during this critical period, over 75% of children with autism can potentially participate in mainstream education.

Currently, pediatricians have limited tools and training to be able to properly evaluate a child with ASD. Pediatricians refer the child to be evaluated by specialists as the first step to receiving treatment. ASD is a spectrum disorder, which means that children often have vastly different symptoms and behaviors that may qualify under the ASD diagnosis. Currently, there is a shortage of pediatric specialists in the US. Parents must often wait for one to three years to receive a proper diagnosis and an individualized treatment plan for their child. 

What happens right now is that the pediatrician has to refer children to specialists nearly 100% of the time. The wait time to see a specialist to be evaluated for a diagnosis can be over a year from the initial pediatrician referral . This hasn’t changed in 15 years but, if we can move the average age of diagnosis down to two years of age, then we can take advantage of the critical early period of child development.

Brent Vaughan, CEO of Cognoa

This week, I had the pleasure of sitting down with Brent Vaughan, the CEO of Cognoa to talk about the state of Autism Diagnosis and what Cognoa is doing to use artificial intelligence to help children who, under the current system, fall through the cracks. 

In the brain, what happens to children during this critical period of early development before the age of 5 years old? What is different about children on the ASD spectrum?

This is the histology of a neurotypical developing brain. What you can see is that kids go through aggressive synaptic proliferation from newborn to 2 years of age. These children are making millions of new connections per day.

Synaptic Pruning, Photo From COGNOA

Synaptic Pruning, Photo From COGNOA

This is what happens between 2 years old and 4 years old. Their connections get pruned down and the child’s dominant social behaviors start to evolve. The brain prunes away things that are not used frequently. Therefore, the connections that are used more become thicker. They become more dominant.

So, kids on the autism spectrum don’t prune like neurotypicals.

Brent Vaughan, CEO of Cognoa

neuron HTTPS://WWW.NCBI.NLM.NIH.GOV/PMC/ARTICLES/PMC4159743/

neuron HTTPS://WWW.NCBI.NLM.NIH.GOV/PMC/ARTICLES/PMC4159743/

If you look at the picture of the ASD Neuron, you can see how the ends have not been pruned away. This is super important. Kids on the autism spectrum have hyperconnectivity. That’s why many of these kids, who are hyper tactile, will use weighted blankets to help provide sensory input and to produce sensations of calmness, relaxation and reduced anxiety.

When a kid on the autism spectrum sees a social cue, they don’t just get a signal back across their brain saying that the person is happy, sad, or confused; they just get an overwhelming number of signals. They don’t know what to do with it all. When they look at changes in facial expressions, they don’t know how to process it. At the same time, they are perceptive. They know that the adult might be getting frustrated. If you have a kid on the autism spectrum and you present them repeatedly with anxiolytic stimuli, they will avoid it.

If you talk to kids on the spectrum, they often don’t look you in the eye. They look away. They can struggle to process that sensory input. Studies have shown that if you can treat these kids in that 2 to 4 years of age window, you can permanently impact their brain development. Although you probably can’t help with pruning, you can repeatedly activate and strengthen certain networks. Once you do that, those networks can end up becoming more dominant and that results in those behaviors becoming more dominant.

It’s not a symptomatic treatment. If you can treat children while they are young, these behaviors can be retained for life.

Brent Vaughan, CEO of Cognoa

How does Cognoa’s therapeutic device help with this problem of Autism Diagnosis?

If we can empower pediatricians to make an accurate diagnosis we can lower the average age of diagnosis, not just shorten the line. Children’s hospitals and specialty centers can have a 1-year waiting list and up to 1000 kids waiting to get diagnosed.

So it’s not about getting kids to the front of the line. It’s about getting rid of the line.

Brent Vaughan, CEO of Cognoa

We figured out that if we could create a medical device that could receive FDA clearance for pediatricians to use, pediatricians could diagnose the majority of the kids in their offices, instead of referring kids onto the waiting lists. Then, these kids can be prescribed treatment during the early development window before they reach five years old. Cognoa is also developing a digital therapeutic that uses augmented reality on the parent’s phone to provide targeted treatment to these kids. Not only will doctors be able to prescribe Cognoa’s app to diagnose autism once it is FDA cleared, but they will also be able to prescribe the therapeutic app once it is FDA cleared to provide individualized treatment. Our goal is to have the pediatricians be able to prescribe our autism diagnostic and digital therapeutic and have them covered by insurance.

How does the diagnostic tool work for pediatricians and parents?

Once the insurance eligibility is determined, then a notification will be sent by the pharmacy to the parents to download the app. Parents use the app to upload the answers to a set of questions, similar to the way they would answer questions at the specialist’s office. Then, they upload a few short videos of the child’s natural behavior at home. We use a combination of human and AI to score the information provided and identify the features to power the diagnosis. For a Pediatrician, they get back a result that is similar to how they use send-out labs. The physician should be able to use their EMR to order the test like they prescribe other things. Once the parents provide the information, if the device determines it has sufficient information to provide an accurate diagnostic result, then the pediatrician will get back a result. The pediatrician makes a final diagnostic decision, combining the result of the diagnostic with their patient context, history and experience.

How does the therapeutic tool work for the child?

The child and parent use the app that utilizes the augmented reality features of the phone, kind of like Pokemon Go. The child uses this app interactively with their parent or another family member. It’s been shown that the autistic brain processes known faces in a different part of the brain than unknown faces. So, we think that using the digital therapeutic app with family members is very important. The child looks through the phone at their parent or family member as the parent makes different faces and expressions. The app uses facial recognition classifiers and emotional recognition classifiers to recognize the emotional state of the parent. Then, the app tells the child the emotional state that corresponds with the facial expression. The use of the app also activates the child’s mirror neuron networks. Studies have shown that if you look at smiling faces, you tend to feel happier because looking at smiling faces activates your mirror neuron network and reinforces the happy emotions. When our digital therapeutic activates and reinforces facial recognition, emotional recognition and mirror neurons simultaneously, it’s quite powerful. In our proof of concept study published in JAMA Pediatrics, we measured improvement in the accredited Vineland Adaptive Behavior Scale (VABS) of children who used the app for 6 weeks. The study showed improvements in VABS scores that were similar to 2 years of standard therapy. This is a therapeutic effect like you would hope to see with medication but without the problems like side effects that may be experienced by giving young kids medication.

What is the difference between digital health and digital therapeutics?

You can think about digital health and digital therapeutics simply. There are lots of digital health products that include the disclaimer, “not intended to be a medical claim.” They are products that engage you and maybe make you feel better. But, for some reason, these companies have never completed the peer-reviewed clinical studies to support a medical claim. I think it helps to refer to these products as digital health products. Other products have completed peer-reviewed outcome studies to support medical claims and these are the products we generally refer to as digital therapeutics. Digital therapeutics can be divided into two groups: those pursuing FDA clearance, which is what Cognoa is developing and which we call Prescription Digital Therapeutics (PDTs) because they are intended to be prescribed by trained healthcare providers and, the second group of products which also have shown that they can improve medical outcomes but have chosen not to go down the path of FDA approval.

Using AI and machine learning combined with a deep understanding of child behavioral development, Cognoa is helping Pediatricians diagnose autism in a timely fashion so that more kids can get the right treatments years sooner, when they can have the greatest impact. Cognoa is also developing digital therapeutics to empower parents and clinicians to help children better engage in therapy and improve children’s lifelong outcomes.


Jun Wu is a Hybrid Journalist for Technology, AI, Data Science. She has a background in programming and statistics.