Machine learning method helps predict mental health symptoms in adolescents

0
Machine learning method helps predict mental health symptoms in adolescents

In recent decades researchers have shown that an adolescent’s neurobiology and environment interact to shape emotional and behavioral development, but to date this work has struggled to capture the complexity of this interplay.

In a new study, Yale researchers modeled these brain-environment interactions using a machine learning technique known as manifold learning, vastly improving the ability to detect mental health problems and to predict future problems compared with existing methods.

The findings were published in the journal Biological Psychiatry: Cognitive Neuroscience and Neuroimaging.

The study was led by Erica L. Busch and May I. Conley, who are both Ph.D. candidates in the Department of Psychology, and supervised by Arielle Baskin-Sommers, an associate professor of psychology and of psychiatry at Yale.

Nearly 75% of all mental health disorders appear during adolescence, with about half occurring by age 14. Given the myriad risks associated with these mental health problems, developmental scientists have long sought to understand how neurobiological and environmental factors are connected to the emergence of emotional and behavioral problems. Yet most of this research has considered these factors in isolation or as simple interactions.

For the new study, the researchers used a technique known as multi-view manifold learning, a class of algorithms capable of uncovering structure within complex, multimodal biomedical data, such as from functional magnetic resonance imaging (fMRI). Specifically, they developed an exogenous PHATE (E-PHATE) algorithm (a method that helps visualize complex data) to model brain-environment interactions.

The researchers used data from the National Institutes of Health-supported Adolescent Brain and Cognitive Development (ABCD) Study — the largest long-term study of brain development and child health ever conducted in the United States (which includes one data collection site based at Yale, led by Baskin-Sommers). By applying E-PHATE to bridge and visualize two different types of data — brain images collected via fMRI and information on the environments in which adolescent participants lived — the researchers were able to predict individual differences in cognition and emotional and behavioral symptoms, both at a snapshot in time and two years later.

The brain–environment manifolds of certain brain regions — for example, frontoparietal and attention networks — vastly improved detection and prediction of mental health issues relative to prior state-of-the-art approaches,” said Busch, the first author of the study. “This underscores the importance of considering the adolescent brain in conjunction with the environment in which it develops.”

The new results demonstrate that manifold learning techniques are well-suited for the complexity of multimodal developmental data and have great potential to enhance research on the neurobiology of emotional and behavioral problems in adolescents.

For a long time, developmental scientists have faced the challenge of testing theories that, in many ways, are hiding in plain sight,” said Conley, co-lead author of the study. “From the neighborhood to the family, we recognize youth’s experiences in their environments and neurobiology both influence emotional and behavioral development. Yet, we haven’t had methods that capture the complexity of this interaction precisely.”

The researchers noted the remarkable effect of combining additional variables which characterize adolescents’ environments into the exogenous view of E-PHATE. They found a greater correlation of brain activity with mental health symptoms through modeling either the neighborhood or familial environments in E-PHATE. But by combining those metrics along with others, they said, the model kept improving its representation.

These findings, researchers say, reinforce the need to consider the multiple environments that adolescents navigate in conjunction with how their brain takes in information from those environments.

The work also highlights the clinical applications of new machine learning and signal processing approaches, they said. Specifically, it underscores the significance and complexity of the relationship between adolescent brains and environments as they relate to emotional and behavioral symptoms.

It is important that, as a field, we improve our ability to capture the complex transactions between the person and their environment,” said Baskin-Sommers, senior author of the study. “However, to estimate these transactions, new methods are needed to handle multiple types of data and estimate their interactions within individuals.

The method produced from this interdisciplinary collaboration is one example of how we can estimate these complex transactions.”

link

Leave a Reply

Your email address will not be published. Required fields are marked *