An Innovative Video Game Boasts an 80% Success Rate for Diagnosing Autism – Part II (15:07)
The video game tool was able to correctly distinguish children with autism vs. neurotypical children with a 80% success rate
Broadcast Retirement Network’s Jeffrey Snyder discusses improving autism evaluation and diagnosis with Dr. Bahar Tungenc of the Nottingham Trent University and Dr. Stewart Mostovsky of Kennedy Krieger Institute.
Jeffrey H Snyder, Broadcast Retirement Network
This morning on BRN, part two of our conversation around an innovative one-minute video game that boasts an 80% success rate in diagnosing autism. And we’re going to welcome back to the program Dr. Bahar Tungenc of the Nottingham Trent University and Dr. Stewart Mostovsky of Kennedy Krieger Institute. Bahar, Stuart, welcome back to the program.
Great to see you. Thanks for joining us again. This morning.
Bahar Tungenc, Nottingham Trent University
Hello.
Jeffrey H Snyder, Broadcast Retirement Network
Thank you. Happy to be here. Bahar, yesterday we had a great conversation about some of the challenges in diagnosing autism.
I want to ask, and you brought up technology. And I want to, my first question to you is how important is technology been to diagnosing autism?
Bahar Tungenc, Nottingham Trent University
There has been a lot of research that has gone into using technology in various ways to help diagnose and also treat autism. So millions of dollars have been spent on neuroscience research, for example, trying to understand the brain related differences, which uses very advanced imaging technology. Again, millions of dollars have been spent on genetics of autism, really trying to identify the autism gene or some autism genes that people can pinpoint so that we can predict who is more or less likely to have autism diagnosis when they’re born or when they grow up.
And, yeah, the genetic approach has had very varied success. There really is no such thing as an autism gene. Yesterday, we talked about what a heterogeneous condition this was.
And I think that is reflected in the biological makeup of the condition as well. And so it’s not surprising that there is not a single gene or a small cluster of genes that determine autism. So I would say that probably Stuart can speak more to this as a clinician, but I would say that most of the autism diagnosis today is actually done without a huge amount of aid from technological tools.
Autism is still, by and large, a condition that is diagnosed based on behavioural presentation. And those behaviours are assessed using some standardised questionnaires that the children and or their parents might fulfil. There are some behavioural tests that the children might be subjected to when they come to the clinic, for example, to see how they behave in a communicative, conversational context or how they play with toys, what kinds of choices they make and so on.
And so then the clinician would try to infer whether they behave in a more, quote unquote, autistic way or in a more neurotypical way. So these are usually the methods that are used in order to diagnose autism. So I would say that technology does not really play a huge role in the traditional diagnostic process.
Jeffrey H Snyder, Broadcast Retirement Network
But Stuart, I mean, that’s changing. And technology, in the case of what you and Bahar and the team have been working on, I mean, you’ve developed a video game called Kami to help in this assessment. Do you want to tell us a little bit about Kami and how that technology has helped make these determinations?
Stewart Mostofsky, MD, Kennedy Krieger Institute
Yeah, I’d be happy to. Thank you for the question. I want to first say that I completely agree with everything Bahar said.
I think that we’re in a state that we’ve been in for decades, which is that the diagnosis really is based on observation and history taking around behavioural presentation. There’s limited use of technology, but there is clearly a need to potentially leverage technology for a number of reasons. One of them is that access to resources is a challenge.
There are, for instance, many rural areas in the United States where getting to a clinician with expertise in autism is a daunting prospect. So the ability to potentially start to leverage technology that would allow for, at the very least, some initial remote assessment. Bahar said yesterday, and I think reiterated today, that none of these technologies we feel are necessarily going to replace the careful, clinical, thorough diagnostic process.
But they could provide a window that would help us screen for and then maybe better target how we approach diagnosis of autism. So with that in mind, and to get to your question, we leveraged decades of research in which people have observed, our own lab and others have observed, that people with autism appear to have challenges quite often with imitating other people, with imitating other people’s actions. This makes sense in a general way because imitation is really crucial for how we learn, all of us learn, social interaction and aspects of communication.
From very early ages, from the time we’re born pretty much, or at least within the first year of life, we’re observing what the people around us do, and we’re developing our repertoire of social interaction in large part based on what we observe. And the fact that people with autism have challenges with motor imitation sort of gives insight into, at least for some people with autism, what may be a really crucial contributing factor to why they go on to have the challenges they do have with socialization and communication. So with that in mind, we developed this one-minute video game where there is an avatar, a human video of a human performing a series of actions on the screen.
So a series of movements that they perform. And the instructions for this video game couldn’t be simpler. Just imitate everything you see the person doing.
And so this computerized assessment of motor imitation, KAMI, was something that we developed and then implemented with children with autism. And in papers that Bahar and I have published, have reported that it is fairly good within one minute of time in distinguishing people with autism from people who don’t have autism. Now, the most recent paper that we published together is really fascinating in that it reveals not only is this KAMI video game, this one-minute KAMI video game, able to distinguish children with autism from neurotypical children, it’s also fairly reliable in distinguishing children with autism from those who have attention deficit hyperactivity disorder, ADHD.
People with ADHD often have some motor challenges, but in years of research, we really haven’t observed that people with ADHD have as much difficulty with motor imitation as we see in people with autism. And so the findings were not necessarily entirely surprising, but they were reassuring regarding KAMI, this video game KAMI’s ability to potentially help and assist with diagnosis with autism. We, of course, perform this in the lab, but as I mentioned earlier, remote assessment might be increasingly important and something like this could be leveraged to do remotely.
Jeffrey H Snyder, Broadcast Retirement Network
And Bahar, I mean, 80% success rate based on at least what I read, that’s pretty good. I wish I could be right 80% of the time. My wife, by the way, is right 99.9% of the time. So I think you still have some ways to go, but in all seriousness, that’s pretty substantial. And how do you take the success of KAMI? Does it get rolled out?
How do I access it if I’m a parent? Is it something that a practitioner would have access to that I have to go to the practitioner? Would my primary care physician or my general internist for my family, a family medicine doctor, be able to access it in order to steer the child to the right practitioner?
How would this all get rolled out?
Bahar Tungenc, Nottingham Trent University
Yeah, thank you for the question. So maybe something worth explaining before I answer that more directly is just what KAMI is. So Stuart already explained that part of what KAMI does is it’s a video game, right?
So there’s this avatar that the children are watching and trying to copy the movements of. And we try to make sure that those movements are fun. So they’re dance like movements.
But at the same time, we try to pick them based on the literature that we know of and our personal clinical observations as well. We try to pick the movements so that they are particularly the types of movements that autistic children might find challenging. So that video game and the dance movement, dance element is one side of KAMI.
But the other side of KAMI, so that’s the MI in KAMI, if you will, the motor mutation side. But the CA, the computerised assessment part, is another crucial part of what constitutes KAMI. And that is where we use the technology at its best, really.
So as the children do those movements, we record their bodies, their movements using three dimensional video cameras. And then we apply an algorithm that we have developed together with our computer scientist collaborators, which analyses the accuracy of the children’s imitation performance. So those two things together, the behavioural side and the algorithm together are what make up KAMI.
Now, for this reason, the first part is relatively easy to give to people. I could give you the one minute avatar video game and you could get a child to imitate it if you put it on your laptop or whatever. But that wouldn’t necessarily give you any scores, right?
It wouldn’t necessarily tell you just how well that child has imitated. In order to understand that part, you need the computerised assessment. And so at the minute, that side is a little clunky, I would say.
So it’s still a research tool for the moment that we are working to develop so that we can scale it up further. And there might be multiple steps to getting that, getting there. One of them is, as I mentioned in the papers that we have published so far, we have used three dimensional cameras.
So these are specialised cameras that are relatively costly and that require some degree of technical knowledge in order to do the recordings accurately and extract the data from it and so on. And so it’s not very easy for a parent to just apply it in their home right now or for a clinician to do it for that matter. What we’re trying to do is apply CAMI to two dimensional so-called off-the-shelf cameras that are available to everybody.
You know, this is your phone camera, this is your computer camera, because we have cameras everywhere. But there are some technological or technical challenges related to changing CAMI from three dimensional cameras to two dimensional cameras. And that’s exactly what we are working on at the moment.
So, yeah, in short, to answer your question, I would say at the minute, it is a research tool that’s not easy to just send over to a clinic or to a parent to apply. But I think right now we are at a stage where we would be really interested to get in those conversations with interested parties, especially in clinical places, perhaps not so much in home settings at the moment, where we can look into opportunities for working together to apply this with the two dimensional cameras as we continue to develop CAMI for that sort of scaling up.
Jeffrey H Snyder, Broadcast Retirement Network
Well, I mean, it certainly sounds like you’re headed in the right direction. I mean, 80 percent, you’re boasting an 80 percent success rate. I couldn’t see why someone wouldn’t want to partner because the practical applications are just so robust.
Stuart, Bahar, we’re going to have to leave it there. Great to see you over these past two days, and we look forward to having you back on the program again very soon. Thank you so much.
Stewart Mostofsky, MD, Kennedy Krieger Institute
Thank you.
Bahar Tungenc, Nottingham Trent University
Thank you very much for having us.
Jeffrey H Snyder, Broadcast Retirement Network
And don’t forget to subscribe to our daily newsletter, The Morning Pulse, for all the news in one place. Details, of course, at our website. And your subscription helps support BRN content.
And we’re back again tomorrow for another edition of BRN. Until then, I’m Jeff Snyder. Stay safe, keep on saving, and don’t forget, roll with the changes.