New approach may help detect and predict mental health symptoms in adolescents by analyzing brain-environment interactions

Most mental disorders manifest during adolescence and are related to a multiple interplay of neurobiological and environmental factors. Rather than considering these factors in isolation, a newly developed multiple learning technique can model brain-environment interactions, greatly improving the detection of existing mental health symptoms and the prediction of future symptoms compared to current methods.

The study in Biological psychiatry: cognitive neuroscience and neuroimaging underscores the importance of considering the adolescent brain in relation to the environment in which it develops.

There is an increasing need to build more complex, yet nuanced, models of human biology and behavior, particularly regarding the development of mental health symptoms. Despite the importance of this problem, most work still treats the brain and environment in isolation or as univariate and linear interactions.

May I. Conley, MS, MPhil, Ph.D. candidate, Yale University, Department of Psychology, co-lead author of the study, says, “Developmental scientists have long faced the challenge of testing theories that in many ways lie hidden in plain sight. From the neighborhood to the family, we recognize that young people’s experiences in their environment and neurobiology both influence emotional and behavioral development. Yet we lack methods that accurately capture the complexity of this interplay.”

To address this, the researchers turned to manifold learning, a promising class of algorithms for revealing latent structure from high-dimensional biomedical data such as functional magnetic resonance imaging (fMRI). They developed the exogenous PHATE (E-PHATE) algorithm to model brain-environment interactions. Using the Adolescent Brain and Cognitive Development (ABCD) dataset, they used E-PHATE embeddings of participants’ brain activity during emotional and cognitive processing to predict individual differences in cognition and emotional and behavioral symptoms, both cross-sectionally and longitudinally.

One of the most notable findings of the study was the effect of combining additional environmental variables into the exogenous representation of E-PHATE. Researchers saw a greater correlation between brain activity and mental health symptoms by modeling the neighborhood or family environment in E-PHATE, but combining those metrics with others continued to improve the model’s representation. However, this was specific to adding environmental information, rather than an effect of the number of variables (which was tested with additional analyses). This finding underscores the need to consider the diverse environments that youth navigate in combination with how their brains process information from those environments.

Erica L. Busch, MS, MPhil, Ph.D. candidate, Yale University, Department of Psychology, first author of the study, continues, “I was excited to see that the principles of modeling neuroimaging data that I had developed for basic science questions could be quickly adapted for clinical applications and yield such striking results and mechanistic insights. It also underscored how fruitful interdisciplinary collaborations can be; my fellow student May Conley and her advisor Dr. Baskin-Sommers are experts in biopsychosocial models of mental health symptoms, and combined with my computational experience, we each played a key role in defining the question and approaches of this project.”

The work highlights the clinical applications of novel machine learning and signal processing approaches. In particular, it underscores the importance and complexity of the relationship between adolescent brains and environments as they relate to emotional and behavioral symptoms. The researchers present a general method with broad applications in both clinical and nonclinical domains.

Editor-in-chief of Biological psychiatry: cognitive neuroscience and neuroimaging Cameron S. Carter, MD, University of California Irvine, notes, “Decades of developmental work suggests that both neurobiology and environmental context shape the development of mental health symptoms. This study demonstrates the suitability of more computational methods, such as manifold learning, for modeling complex multimodal developmental data, and they have great potential to improve research on the neurobiology of emotional and behavioral symptoms in adolescents.”

The current research is novel in three important areas:

1. By characterizing both neural and environmental data as multivariate measurements.
2. By considering the interaction between them as nonlinear and lower-dimensional (i.e., existing along a latent manifold, like most real-world data).
3. By enabling simultaneous hypothesis- and data-driven discovery of a meaningful representation of these signals.

Lead author Arielle Baskin-Sommers, Ph.D., Yale University Department of Psychology, concludes, “It is important that we as a field improve our ability to capture the complex transactions between individuals and their environment. However, estimating these transactions requires new methods to process multiple types of data and estimate their interactions within individuals. The method resulting from this interdisciplinary collaboration is an example of how we can estimate these complex transactions.”