AI is now being used to spot signs of schizophrenia in brain scans

A breakthrough study has found that AI can be used to identify schizophrenic traits by studying the flow of blood in a person’s brain.

The research, by the University of Alberta, used IBM’s AI to analyse fMRI images from 95 participants. This group was divided into two groups – one control group and one group of people diagnosed with schizophrenia. By measuring the ‘connectivity’ of each participant’s’ brain, this AI accurately diagnosed patients 74% of the time.

The machine determined diagnoses by tracking blood as it flows around the brain. When scans show abnormal blood flow in the brain, it’s often a sign of schizophrenia. Anatomical disruptions were first identified as an indicator for schizophrenia in the early 1900s by German neuroanatomist Carl Wernicke, who often studied the brains of hospital and asylum patients for his work.

The AI compared the same physicalities in the brains of the control group and schizophrenic subjects to learn whether it was a positive or negative diagnosis.  

Once diagnosed, the researchers applied statistical analysis to the patient’s “whole brain connectivity” — the first time this method has been used — to identify features that can be used to determine the severity of symptoms. Schizophrenic symptoms vary, but generally cause patients to seem like they’ve lost touch with reality and act abnormally.

This isn’t the first time IBM has used AI technology to better understand medical conditions. Watson for Oncology suggests cancer care plans, while data presented at the American Society of Clinical Oncology’s annual meeting last month showed that the AI’s suggestions often aligned with physicians’ recommendations. The technology is currently being used by nine new medical centres.

Using technology to diagnose patients and determine care leads to more “objective measurement-drive characterisations” of diseases, according to IBM’s and the University of Alberta’s study. Avoiding subjectivity is also stressed in the Research Domain Criteria by the US National Institute of Mental Health.

In addition to promoting objectivity, this study uses methods that could lead to better care for schizophrenic patients.

“This unique, innovative multidisciplinary approach opens new insights and advances our understanding of the neurobiology of schizophrenia, which may help to improve the treatment and management of the disease,” Dr. Serdar Dursun, a Professor of Psychiatry & Neuroscience with the University of Alberta, said in a statement. The study was published in the journal Nature.

Images: M Gheiratmand et al./Nature