The research was able to predict instances of schizophrenia with 74% accuracy. The team, working out of the IBM Alberta Centre for Advanced Studies, also discovered the ability to predict the severity of specific symptoms in schizophrenia patients, something that wasn’t possible before.
Schizophrenia doesn’t currently have medical testing that can provide an absolute diagnosis, which can cause a significant delay before a symptomatic person is properly diagnosed.
The chronic neurological disorder affects seven or eight of every 1,000 people and those with the disorder can experience hallucinations, movement disorders and cognitive impairments.
These findings can be used to help doctors more quickly assess and begin treatment for patients, as well as measure the progression of the disorder and the effectiveness of treatment, Gheiratmand said.
For the study, researchers analyzed brain functional magnetic resonance imaging (fMRI) data, which Gheiratmand explained is basically a movie of the brain while it’s in action, of patients with schizophrenia, as well as a healthy control group without the disorder.
The team used machine learning techniques to examine brain scans of the 95 participants to develop a model of schizophrenia that identifies the connections in the brain most associated with the disorder.
“We were actually interested to see how people with schizophrenia are different from (the control group) and basically use that to predict who has schizophrenia and who does not,” she said. “Basically be able to discriminate between these two groups based on brain images.”
Further, the research showed a similar model can be used to determine the severity of symptoms, including inattentiveness, formal thought disorder and lack of motivation.
This discovery could lead to a “spectrum” characterization of schizophrenia and not just a binary label of simply having it or not.
“It actually enables us to give a more precise, more measurement-based characterization of the disease,” Gheiratmand said.
These are still the early stages, Gheiratmand said, of using these technologies in practice. But they show promise in objectively and precisely diagnosing conditions early.
Researchers are also looking into using the same approach for other psychiatric disorders, such as depression and post-traumatic stress disorder.
“If you have a model that can predict the disease at earlier stages, then you can intervene earlier,” she said. “And it’s very critical to intervene early for much better outcomes for patients.”