Multimodal MRI markers of cognition derived from diffusion-weighted, resting-state functional, and structural MRI collectively explain 48% of the covariation between cognition and mental health in the UK Biobank, demonstrating that neural indicators of cognition extend beyond predicting cognition itself to capture cognition-mental health covariation.
The predictive association between mental health and cognition was r=0.3 in the UK Biobank sample.
Individual neuroimaging phenotypes captured between 2.1% and 25.8% of the cognition-mental health covariation.
Combining neuroimaging phenotypes within each MRI modality substantially improved explanation of the cognition-mental health covariation.
Combining neuroimaging phenotypes across all three MRI modalities explained 48% of the cognition-mental health covariation.
Neural indicators of cognition derived from neuroimaging extend beyond predicting cognition itself to capture cognition-mental health covariation transdiagnostically.
The study used a machine learning approach applied to 72 neuroimaging phenotypes across three MRI modalities to derive neural indicators of cognition.
This research suggests that brain scan measurements can help explain why people who have mental health difficulties often also show lower cognitive performance (such as poorer memory or attention). Using data from over 14,000 participants in the UK Biobank, the researchers first established that mental health and cognitive ability are meaningfully related to each other (correlation of r=0.3). They then used machine learning to extract brain-based indicators of cognition from three types of MRI scans—one measuring brain structure, one measuring brain connectivity at rest, and one measuring the structure of white matter fiber tracts—and tested how much of the mental health-cognition relationship these brain measures could account for. The researchers found that individual brain measures explained between about 2% and 26% of the link between cognition and mental health, but that combining measures within a single scan type improved this to roughly 25–32%. Most strikingly, when all three types of MRI were combined, they collectively explained 48% of the relationship between cognition and mental health. This suggests that different MRI modalities capture distinct but complementary aspects of brain biology that together paint a much fuller picture. This research suggests that brain imaging markers developed to predict cognitive ability are not narrowly useful—they can also illuminate why mental health problems and cognitive difficulties tend to go hand-in-hand. This 'transdiagnostic' finding, spanning many different types of mental health conditions, could eventually help researchers better understand the shared neurobiological foundations of cognitive and psychiatric difficulties, potentially informing future approaches to identifying individuals at risk across multiple conditions simultaneously.
Buianova I, Silvestrin M, Deng J, Pat N. (2026). Multimodal MRI marker of cognition explains the association between cognition and mental health in the UK Biobank.. eLife. https://doi.org/10.7554/eLife.108109