Researchers have developed a new artificial intelligence model designed to identify patterns in brain activity associated with obsessive-compulsive disorder (OCD). By analyzing brain scans from over 1,700 people with OCD and 800 healthy controls across 23 global sites, the study used deep learning to look for differences in how brain networks communicate.
The study found that the AI could identify widespread patterns of reduced connectivity in the brains of those with OCD. Specifically, the model noted lower activity in certain brain networks, such as the default mode and salience networks, compared to healthy individuals. The researchers also found that using a much larger dataset for initial training helped make the model's predictions more reliable and less overconfident.
While these results are promising for creating more trustworthy clinical tools, there are important limitations. The model's accuracy varied significantly depending on which site's data was used, and the large-scale training did not fully close this gap. Because the findings are based on this specific AI framework, more research is needed to see if these patterns can be used reliably in everyday medical practice.