Early diagnosis of autism is essential for developing successful early interventions, and in the last few years, researchers have made significant progress in comprehending the disorder’s development. But since autism spectrum disorder, or ASD, is more of a behavior-based than a biologically-based condition, it’s still difficult to determine which child will acquire autism in the future.
ASD is a neurodevelopmental disease typified by a pattern of behavior and a lack of social skills. At the moment, a diagnosis is made on average after two years. To determine when this condition first appeared, diagnostic techniques have been employed. Autism As part of the Diagnostic Observation Schedule (ADOS), a kid and an examiner evaluate the child’s behavior by watching videotapes.
Electroencephalography is a widely used technology that involves the identification of brain functions and activity. It has been used by Boston Children’s Hospital and has been shown to be effective in anticipating when autism in children may manifest.
Researchers from the University of North Carolina at Chapel Hill and Washington University School of Medicine claim that they can now predict with a 96 percent accuracy rate which child will develop autism before they turn 24 months old in a paper published in Science Translational Machine.
The study’s senior author, Joseph Piven, a professor of psychiatry, psychology, and pediatrics at the University of North Carolina, Chapel Hill, told CNBC, “In the field we are always trying to detect autism at younger ages, so we can start treatment earlier, but we hit a wall around 2 to 3 years of age, because the symptoms don’t show up until around that.”
230 brain areas were mapped from 59 high-risk, 6-month-old infants’ brain scans in order to produce functional connectivity matrices from the MRI data for each child. A total of 26,335 pairs were created and subsequently examined to investigate the functional organization of an infant’s entire brain. After the kids reached two, the researchers looked at their behavior and discovered that 11 of the kids had autism symptoms.
We created a completely cross-validated machine learning algorithm and trained it with the 6-month-old infants’ images. Once more, the algorithm was developed to examine the brain scans of the 6-month-old babies. It had a sensitivity of 81.8% and successfully predicted 9 out of 11 babies who were given an autism diagnosis at 24 months. Furthermore, the 48 6-month-old babies who were not given an ASD diagnosis were appropriately categorized.
This discovery can also be compared to early detection of brain problems in Parkinson’s and Alzheimer’s patients, before the conditions cause irreversible impairment. Autism is a spectrum condition whose symptoms can range from minor impairments to severe ones. This strategy, which is data-driven and shows promise as a decent predictive measure, can be improved to determine if it can also forecast the disorder’s severity.
The findings of this study imply that artificial intelligence (AI) and machine learning may someday accurately diagnose illnesses, provide therapies for them, and perhaps even stop the disorders’ progression.
