The algorithm assessed brain scans from a sample of approximately 1,100 patients, then identified a group of patients whom human clinicians had diagnosed with autism, with 82% accuracy.
Credit: immpile/iStock
Stanford University researchers have created an algorithm that may help detect autism from analyzing functional magnetic resonance imaging (fMRI) brain scans, and that also can forecast the severity of autism symptoms in individuals.
The explainable artificial intelligence (XAI) algorithm maps neural activity visualized in fMRI scans over time to produce "fingerprints," which can be sorted and classified based on common elements.
The algorithm applies a simple mathematical model that assesses brain regional interactions and interconnectivity, focusing on three regions exhibiting substantial interconnectivity differences in a groupable portion of the dataset; those three regions have been previously associated with autism pathology.
The researchers found the XAI algorithm evaluated brain scans from a sample of roughly 1,100 patients, extracting a group whom human clinicians had diagnosed as autistic with 82% accuracy.
From Stanford Institute for Human-Centered Artificial Intelligence
View Full Article
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA
No entries found