Toward building neuroimaging biomarkers to capture the cognition-mental health relationship across the lifespan

Abstract:

The NIMH Research Domain Criteria (RDoC), a leading transdiagnostic framework in mental health, identifies cognition as one of the core functional domains underlying psychopathology across diagnoses. RDoC conceptualises the link between cognition and mental health as spanning multiple neurobiological levels of analysis—from genes to brain systems — from normal to abnormal in normative samples. However, recent studies have raised concerns about the robustness of brain MRI in capturing individual differences in cognition, casting doubt on its utility as a neuroimaging biomarker for RDoC’s cognitive domain. To address this challenge, we proposed a machine learning-based multimodal fusion approach that integrates diverse brain MRI modalities—including task-based fMRI contrasts, functional connectivity during both task and rest, and structural MRI—into a unified predictive model. Leveraging large-scale datasets across the lifespan (n > 2,100, age 22-100), we demonstrated that this multimodal fusion consistently enhances the psychometric properties of brain MRI in two key areas: a) Predictive validity—the ability to accurately predict individual cognitive performance out-of-sample, and b) Test-retest reliability—the consistency of predictions over time. We further evaluated the method’s utility in elucidating the relationship between cognition and mental health using large-scale data from the Adolescent Brain Cognitive Development (ABCD) Study and UKBiobank. Our findings revealed that neuroimaging accounted for the majority of the shared variance between cognition and mental health, higher than polygenic scores. These results suggest that multimodal fusion offers a promising pathway for developing robust neuroimaging biomarkers aligned with RDoC’s cognitive systems.

Biography: 

Narun Pat, PhD, is a Senior Lecturer (above the bar) at the University of Otago in Dunedin, New Zealand. He earned his doctorate in Brain, Behavior, and Cognition from Northwestern University in the United States. Following his PhD, he undertook research fellowships at the National University of Singapore and the U.S. National Institute of Mental Health. His research focuses on understanding and predicting neural-cognitive mechanisms underlying health and mental illness. His current work aims to develop predictive markers of cognitive functioning using multimodal neuroimaging, machine learning and large-scale data analytics. His works have been funded by Ministry of Business, Innovation & Employment in New Zealand, Health Research Council of New Zealand and Neurological Foundation of New Zealand.