Computer models could yield more accurate ADHD diagnoses

January 21, 2021

Behavior observation is one traditional way to diagnose ADHD. Now there may be a better way. (AP Photo/Daniel Cole)

Introducing computer programs to a clinical setting may provide clinicians with faster, more accurate diagnoses for attention deficit hyperactivity disorder, a new study out of The Ohio State University suggests.

The research, published in the December issue of the Psychological Bulletin, reviewed 50 studies from a range of cognitive testing for ADHD in order to assess insights provided by computational models. 

The introduction of those models, which would enable doctors to compare a computer-generated simulation of a typical brain to the dysfunctional behavior seen in the testing, could refine clinical understanding of ADHD symptoms and provide more broad insights into the nature of the disorder in individual patients. The addition of such modeling would complement existing diagnostic methods of cognitive testing and clinical interviews, researchers argue.

Typically, mental health disorders of all kinds are diagnosed based on patient interviews and cognitive testing. However, for ADHD in particular, this type of diagnostics is often unable to capture some of the more subtle symptoms, as it requires a level of “good insight,” according to Nadja Ging-Jehli, the lead author of the article and a graduate student in psychology at Ohio State.

“The children need to be able to think on how they feel, they need to be able to communicate their feelings, and the parents need to be able to observe,” Ging-Jehli added. “Many clinicians also sometimes don’t have the time to observe the children in their usual settings.”

The researchers stressed that computational models would be a “valuable add-on” to the existing process, rather than a replacement, by providing a straightforward method through which clinicians can observe a patient’s typical behavior.

“It gives a clinician a nice complementing instrument. When you ask those children to do such a task, you can see how they are doing it,” Ging-Jehli said. “You can expose them to a situation where they are now showing those clinical characteristics, by maybe making decisions that are typical for them, or you can see how they interact with the other children.”

The researchers argued that existing cognitive tests, which may measure responses such as reaction time, might not be doing enough to determine the underlying reasons as to why that response might be faster or slower. The addition of computational modeling would be able to better capture the range of reactions and provide better information as to the process behind a person’s decision-making or reaction time.

A further benefit of computational modeling for ADHD specifically is the potential to better see and understand the full spectrum of ADHD symptoms. Some symptoms of the disorder, such as a lack of attention or inability to focus, can go unnoticed in childhood if the individual is able to adhere to a schedule and otherwise does well in school, Ging-Jehli said. 

The Diagnostic and Statistical Manual of Mental Disorders, 5th Edition definition of ADHD requires that a patient have symptoms of the disorder before the age of 12, but, "Many of those symptoms may not be observable until the age of 12,” Ging-Jehli noted.

“Even though the neurodivergent characteristics may occur before the age of 12, it’s a different thing to actually observe them before that age,” Ging-Jehli added. “Those people may struggle for years … and many of them are also misdiagnosed.”

Computational modeling would also be able to set parameters that would better be able to quantify ADHD symptoms and create a continuous measure of severity, the researchers said.

“With computational studies, we can disentangle how much does the severity of hyperactivity plays a role and how much inattention plays a role,” Ging-Jehli said. “One of our studies actually suggests that we can use those model parameters to get the spectrum, to measure how severe someone is affected, and where on this spectrum this person would fall.”

Though this research specifically focused on ADHD, Ging-Jehli said that computational modeling would be useful for other types of mental disorders, both in terms of reaching a proper diagnosis and for determining the correct treatment path. Ging-Jehli said that the research team is currently looking into whether cognitive tests used with computational modeling can predict a beneficial treatment plan, which both patients and clinicians can struggle with after a diagnosis.

“Many people who are diagnosed have a hard time finding the right treatment, and that leads to people becoming frustrated, because they try too many different treatments and each might have different side effects,” Ging-Jehli said. “It can be bothersome, time-consuming, costly and frustrating for patients. That, I think, would be a huge benefit — if we can make more informed decisions of which treatment to try first.”

The article, “Improving neurocognitive testing using computational psychiatry—A systematic review for ADHD,” was published in the Psychological Bulletin on Dec. 28, 2020. It was authored by Nadja Ging-Jehli, Roger Ratcliff, and L. Eugene Arnold, all of The Ohio State University.

We use cookies to improve your experience on our site and to show you relevant advertising.