Developing Ideas | Eunah Yang

Neuroimaging-based transdiagnostic approach to mental health variation

We still lack objective biomarkers that can reliably diagnose or predict mental health conditions. Much of this limitation arises from traditional case–control studies, which are built on the assumptions that each condition has a distinct and isolated cause, and that group averages adequately represent the population. However, individuals within the same label often show substantial variability in both symptoms and neural patterns, while those across different labels may exhibit considerable overlap. As a result, categorical definitions tend to oversimplify the complex and continuous nature of cognition and mental health, making it difficult to bridge psychological constructs (top-down) with interpretable neural mechanisms (bottom-up) that vary meaningfully across individuals.

This illustration describes that individuals distributed in a continuous behavioral space cannot be neatly separated by diagnostic boundaries. Clusters based on shared phenotypic features often cut across traditional categories. (left figure: Astle et al., 2021, Journal of Child Psychology and Psychiatry, right figure: generated by me)

How can we model the relationship between brain connectivity and multidimensional behavioral phenotypes in a way that captures individual variation beyond diagnostic categories?

To move beyond these categorical limitations, I propose a brain-based approach that focuses on individual network patterns rather than predefined labels. Instead of predicting a single disorder, each person is represented as a vector of multiple behavioral and cognitive constructs derived from various assessments. By predicting these multidimensional profiles directly from functional connectomes, this idea aims to capture continuous variation in how brain networks support diverse mental processes. This perspective (described as transdiagnostic approach) seeks to reveal shared neural mechanisms that span conventional categories while maintaining sensitivity to individual differences. Ultimately, it enables a more interpretable and mechanistic understanding of how the brain functions to complex behavioral phenotypes.

Using connectome-based predictive modeling (CPM), I plan to test whether patterns of functional connectivity can predict these behavioral vectors, thereby modeling psychiatric features as gradients rather than discrete categories. With this approach, I aim to uncover shared neural dimensions that extend across traditional boundaries and help explain the diverse ways symptoms manifest across individuals. Ultimately, this project may provide a quantitative, individualized, and biologically grounded view of mental health and importance of cognitive process alteration, contributing to a more flexible and integrative understanding that recognizes both common mechanisms and unique individual trajectories.

Using connectome-based predictive modeling (CPM), I plan to test whether patterns of functional connectivity can predict these behavioral vectors, thereby modeling psychiatric features as gradients rather than discrete categories. Through this approach, I aim to uncover shared neural dimensions that extend across traditional boundaries and help explain the diverse ways symptoms manifest across individuals. Ultimately, this idea seeks to provide a quantitative, individualized, and biologically grounded view of mental health. It may also help identify how different cognitive processes interact with symptom expression, offering insights into both common mechanisms and individual variability.