Development of an Artificial Intelligence Algorithm for Early Detection of Autism
Russian and Chinese scientists have successfully developed an artificial intelligence algorithm that will facilitate the early detection of autism by analyzing electroencephalogram (EEG) data, according to the Russian Academy of Sciences.
A statement from the Academy’s press service indicates that the research team was able to develop an AI algorithm capable of detecting autism with 95% accuracy.
“The algorithm we have developed specifically aims to search for distinctive features in autistic children and compare their EEG data with that of healthy children,” said Professor Alexander Khramkov of the Baltic Federal University in Russia. “This algorithm will also allow us to use AI machine learning to quickly detect autism symptoms in their early stages.”
He added that the scientists developed the algorithm using a particular type of AI technology called a “variational autoencoder system”. The algorithm was programmed to detect specific patterns in EEG data. To do this, EEG data was collected from 298 children aged 2 to 16, half of whom suffered from different forms of autistic disorder. The data from the children with the disease were then compared with that of healthy children.
Similarly, tests carried out on the algorithm after its development showed that it was able to identify autistic people from EEG data with 95% accuracy, without any false positives.
The algorithm also allowed scientists to identify several distinctive features in autistic people, including a weakness in certain functional connections in the frontal lobe of the brain. The developers therefore suggested using it, as well as EEG data, to develop new methods of detecting the disease.
