Autism – the developmental disorder characterized by limited social interaction and communication and repetitive behaviors – is on the rise globally, especially in densely populated countries like China. Recent figures suggest about 1 in 100 children in China have autism, similar to rates in Western countries.
The screening and diagnosis of autism spectrum disorder is a complicated puzzle for the medical community as the causes and development of autism are not well understood. Diagnosis of childhood autism relies on rating scales combined with individual doctors’ experience.
One of the signs of autism are delays in many areas of a child’s development. By the time those delays are apparent and a child has been diagnosed with autism, he or she may have already developed negative social behaviors.
Now, a new detection method based on an artificial intelligence (AI) algorithm co-developed by the Ping An Institute of Artificial Intelligence and Peking Union Medical College Hospital (PUMCH) has shown promise as a diagnostic test for earlier diagnosis.
Scientists and researchers have found that urine tests for children can be analyzed to detect risks of autism. The method, based on an AI algorithm, XGBoost, analyzes urine sample data to calculate the risks of autism. Its diagnostic accuracy has set new standards for the scientific screening and diagnosis of autism.
In addition, the algorithm tests fewer urine organic acids than other algorithms but has achieved greater accuracy, making it more cost-effective.
The research findings of Ping An and PUMCH were published in Frontiers in Neuroscience, a leading international academic journal that publishes rigorously peer-reviewed research. In this project, Ping An and PUMCH compiled the biggest data set ever used in an international research report for a similar task.