A new approach to analyzing brain scans could help researchers better understand psychiatric illness using much smaller groups of patients than previously thought necessary, potentially accelerating the development of more precise mental health treatments.
Until recently, scientists believed they needed scans from thousands of people to draw reliable conclusions about how brain function relates to behavior and symptoms – a requirement that put such studies out of reach for most clinical researchers.
However, in a new in Science Advances, researchers demonstrated they could accurately predict cognitive functioning in psychiatric patients using hundreds, rather than thousands, of subjects while maintaining scientific rigor.
"This is very much in the research stage," said Avram Holmes, associate professor of psychiatry at and senior author of the study. "But eventually, these approaches could help identify underlying mechanisms that might be causing patients’ symptoms and help treat the cause as directly as possible."
The researchers used an approach called "meta-matching," which leverages data from large population studies to boost the accuracy of smaller clinical studies. The technique works somewhat like how tech companies use large datasets to improve predictions about individual users.
"It's like a bridge that jumps from those big population samples to the smaller scale clinical collections," said Holmes, who is also a core member of the and the
The team demonstrated they could accurately predict cognitive functioning – a key concern across psychiatric conditions – in three different groups of patients with various diagnoses, including depression, anxiety and schizophrenia. The predictions remained accurate even when tested across different patient groups, scanning sites and cognitive tests.
The findings are particularly significant because cognitive problems are a major concern for psychiatric patients, an issue that often doesn't improve even when other symptoms get better.
"It's a chief complaint amongst many patients. They become ill, and then their ability to think through complex problems and their executive functioning suffer as their symptoms increase," Holmes said.
The research identified specific brain networks involved in cognitive function. They also observed decreased connectivity between these regions and areas handling basic sensory information. This pattern was consistent across different psychiatric conditions, suggesting shared biological mechanisms may underlie cognitive problems in various mental illnesses.
While the current research focused on cognitive function, Holmes said the approach could potentially be expanded to study other symptoms and ultimately might help doctors better match treatments to individual patients. For example, it might help identify which patients with depression might be at risk for medication side effects.
"Individuals could present the same clinically at the moment but have completely different illness courses going forward," Holmes said. "The idea is that some of these algorithms will be able to predict not just current symptoms but eventual illness course."
However, Holmes emphasized that routine brain scans for psychiatric patients remain far in the future, both because the technology needs further development and because of cost considerations.
"The question in my mind is whether there are other things to spend those funds on that would be more beneficial to the patients in the near term," he said. "How do you best weigh future scientific discovery, which could aid in clinical care down the road, versus near-term clinical needs?"
The researchers are now working to expand their approach by studying other aspects of psychiatric illness while also trying to understand the biological mechanisms that link brain function to symptoms. Their goal is to develop more precise and personalized approaches to psychiatric treatment.
RSS