Analysis of hundreds of scans using artificial intelligence techniques revealed differences that were specific to autism and not found in typically developing boys or girls. According to the scientists, the study may pave the way for better diagnostics for girls with autism and explain why their symptoms differ from those of their male counterparts.
Symptoms of autism can range from mild to severe. Social and communication deficits, restricted interests, and repetitive behaviors are common in children with autism. There was a gender bias in Leo Kanner’s original description of autism, which was published in 1943. Autism affects boys four times more than it does girls, and as a result, the majority of research on the disorder has focused on males.
The study’s lead author, Kaustubh Supekar, PhD, a clinical assistant professor of psychiatry and behavioral sciences, said, “When a condition is described in a biased way, the diagnostic methods are biased,” “A new way of thinking may be required, as this research suggests.
The British Journal of Psychiatry published the study online on Feb. 15th.
According to senior author Vinod Menon, PhD, professor of psychiatry and behavioral sciences and Rachael L. and Walter F. Nichols, MD, Professor, “We detected significant differences between the brains of boys and girls with autism, and obtained individualized predictions of clinical symptoms in girls,” “We know that diagnosing autism in girls is difficult because of the difficulty in distinguishing symptoms, which leads to delays in both diagnosis and treatment.
Girls with autism are less likely to exhibit overt repetitive behaviors than boys, according to the researchers.
An assistant professor of psychiatry at the university who was not involved with this study said that “Knowing that males and females don’t present the same way, both behaviorally and neurologically, is very compelling,” he said in a press release.
At Stanford Children’s Health, Fung treats people with autism, including girls and women who have been diagnosed late in life. Many autism treatments are most effective in the early years of life, when the brain’s motor and language centers are developing, he noted.
For example, children on the autism spectrum who receive early language intervention will have a better chance of developing language like everyone else and won’t have to keep playing catch-up as they grow up,” said Fung. “When a child is unable to express themselves clearly, they fall behind in a wide range of subjects. If they don’t get diagnosed early, the consequences can be dire.”
773 children with autism, 637 of whom were boys and 136 of whom were girls, were scanned using functional magnetic resonance imaging (fMRI). According to Supekar, it was difficult to gather enough data to include a significant number of girls in the study because of the historically low number of girls included in autism research. The study’s findings were based on brain scans collected at Stanford University and in public databases from around the world.
A new mathematical problem was created by the overwhelming presence of males in the brain-scan databases. Groups must be roughly equal in size in order to use standard statistical methods for determining differences between them. If one group is four times as large as the other, these methods, which underlie machine-learning techniques that can be trained to find patterns in very large and complex datasets, cannot be applied to the real world.
Algorithms would tell Supekar that “When I tried to identify differences [with traditional methods], the algorithm would tell me every brain is a male with autism,” when he tried to identify differences using traditional methods. “It was over-learning, with no distinction made between autistic men and women.
As a co-author on the study, Tengyu Ma, an assistant professor in computer science and statistics at Stanford, discussed the problem with Supekar. A new method had recently been developed by Ma that could reliably compare complex datasets, such as brain scans, from different-sized groups of people. The scientists finally had the breakthrough they were looking for thanks to a new technique.
We were fortunate, according to Supekar, that this new statistical approach was developed at Stanford.
Brain scans from autistic children were used to develop an algorithm that correctly identified boys and girls 86% of the time. On the remaining 95 brain scans from autistic children, the algorithm maintained the same level of accuracy in distinguishing between boys and girls.
The algorithm was also tested on 976 brain scans from normal-development children and teenagers. The algorithm was unable to tell the difference between the sexes, indicating that the gender differences discovered by the researchers were specific to autism.
The motor, language, and visuospatial attention systems were all found to have different patterns of connectivity in autistic girls than in boys. It was found that the most significant differences between sexes were found in one group of motor areas, including the primary motor cortex, secondary motor area, parietal occipital cortex, and middle and superior temporal gyrus. There was a correlation between the severity of motor symptoms in girls with autism and differences in their brain patterns, meaning that girls whose brain patterns resembled those of boys with autism had the most severe motor symptoms.
As previously noted, boys with autism are more likely to have language difficulties than girls, according to the new research.
“When you see that there are differences in regions of the brain that are related to clinical symptoms of autism, this seems more real,” Supekar said.
It is recommended, according to the researchers, that the findings be used to help guide future efforts to improve the diagnosis and treatment of female patients
Precision psychiatry in autism is now possible thanks to Menon’s research, which uses artificial intelligence-based techniques.
“It’s possible that the tests for men and women should be tailored to each other. We developed artificial intelligence algorithms that may aid in the better diagnosis of female autistics “Supekar made the statement. However, he added, interventions for female patients could begin earlier.
The study’s other Stanford Medicine co-authors are Carlo de los Angeles, a scientific data analyst; Srikanth Ryali, a senior research scientist; and Kaidi Cao, a graduate student. It is a collaborative effort involving Stanford University faculty from the Wu Tsai Neuroscience Institute and the Human Performance Alliance, as well as members of the Stanford Maternal & Child Health Research Institute and the Stanford Institute for Human-Centered Artificial Intelligence.
