I am currently Assistant Professor and a licensed clinical psychologist in the Department of Psychiatry at the University of Vermont. My long-term career goal is to become an independent investigator using novel strategies in developmental neuroimaging to study mood and anxiety symptomatology from birth to maturity. Although I have been trained in the analysis of longitudinal structural MRI, I require further training in the processing and analysis of state-of-the-art multiband neuroimaging data that allows for more sensitive measures of brain connectivity. I am also lacking expertise with regard to more sophisticated analytic methods for more fully leveraging large-sample multimodal datasets. Such approaches will enable me to move beyond conventional univariate statistical analyses and prepare me for future Big Data initiatives. During the proposed K08 period, my overarching goal is to develop expertise in the application of machine-learning approaches to multimodal data in order to characterize the most salient psychosocial and brain-based predictors of youth internalizing psychopathology. To achieve these goals, I am pursuing career development and training activities in the following areas: 1) assessment and characterization of psychosocial risk factors; 2) theory and implementation of Big Data methods, including machine learning algorithms and cross-validation strategies; 3) analysis of multiband multimodal brain imaging data using Human Connectome Project pipelines with the aim of more comprehensively assessing aspects of cortico-limbic connectivity; 4) independently running my own neuroimaging research study; and 5) developing and submitting a competitive R01 application. In order to obtain this expertise, I am proposing training activities at several institutions, including the University of Vermont, Harvard Medical School, McGill University, and Oregon Health and Science University. The research project in this K08 proposal aims to produce risk algorithms for a transdiagnostic dimension of psychopathology, using novel machine learning approaches to leverage two of the largest longitudinal neuroimaging samples in the world (IMAGEN and the Adolescent Brain Cognitive Development study). These risk algorithms will subsequently undergo refinement using a new sample of clinic-referred youths that I will recruit from an outpatient psychiatric clinic in Vermont. As part of the project, I will also test the degree to which these algorithms predict treatment response. These data will be used as pilot data for my planned R01 application. Given the methods that I am proposing, this project will be able to detect complex non-linear interactions involving risk factors from a multitude of domains. As a result, this work will inform, and help to delineate, various etiological pathways that ultimately result in internalizing problems. Most importantly, this project could inform early identification and targeted intervention strategies during a critical period for the development of internalizing symptomatology. !
This study aims to produce risk algorithms for a transdiagnostic dimension of mood and anxiety psychopathology using novel machine learning approaches to leverage two of the largest longitudinal neuroimaging datasets in the world. These risk algorithms will subsequently undergo further refinement using a new sample of clinic-referred youths that the applicant will recruit from an outpatient psychiatric clinic. This project has the exciting potential to characterize early determinants of future internalizing problems, and set the stage for more individualized interventions and targeted prevention efforts.
