Project Summary Traumatic brain injury (TBI) has been termed the “signature injury” of the recent military conflicts in Iraq, with incidence estimated between 9-28%. While the vast majority of TBIs are mild, long- term effects can persist years post-injury. TBI can impact psychological health, physical health, functioning, and quality of life, but predicting outcome post-injury is challenging. Large sample sizes are necessary to accurately characterize heterogeneity in outcome, and two on-going longitudinal studies have collected in-depth data with the goal of improving Service Member and Veteran (SMV) health. Long-term Impact of Military-relevant Brain Injury Consortium (LIMBIC) and Translational Research Center for TBI and Stress Disorders (TRACTS) collectively have gathered information on thousands of SMVs around the United States, and combining these datasets will give us unprecedented power to examine factors that influence outcome post- injury. This Proof of Concept (POC) project will demonstrate the feasibility of combining these massive samples and will show the power of large datasets. Together, TRACTS and LIMBIC have information on TBI severity, physiologic functioning, fluid, and imaging biomarkers, behavioral factors, functional outcomes, and mental health outcomes on approximately 3,500 individuals, and our team has significant experience in collating and organizing large amounts of data and applying harmonization methods to imaging, fluid biomarker, and clinical endpoint data. With the expertise of our project team, we will pool and harmonize data for integrated analyses. Combining and harmonizing data from different sources has many challenges, but the project team has extensive, interdisciplinary expertise, particularly in the area of big data analytics. Neuropsychological data will be combined by domain. Crosswalks and common analytical methods will be created for both multimodal magnetic resonance imaging data (MRI) and fluid biomarker data. MRI data will be harmonized using ComBat-GAM to minimize site effects. Fluid biomarkers will be combined utilizing harmonization methods to integrate multi- omic markers using newer methods including deep learning (DL) algorithms. Using the biomarker data as independent variables and the cognitive and function data as dependent variables, risk factors associated with brain disorders and recovery will be identified. Given large, rich datasets, machine learning approaches can help outline symptom profiles, but careful curation of datasets is critical to ensuring validity. Beyond the massive dataset generated by this POC project, by creating and validating methods for combining multi-site data, a critical deliverable will be the infrastructure for future cross-study analyses. The end goal of this project is to leverage the vast amount of information available regarding TBI and SMV health to identify generalizable biomarkers of outcome that can inform patient-tailored care and thus decrease TBI-...