Leveraging data-science for discovery in chronic TBI

点击翻译按钮获取中文摘要

Chronic traumatic brain injury (TBI) is one of the most prevalent neurological disorders in both military and civilian populations, impacting up to 5.3 million people in the US and costing $76 billion in healthcare and loss- of-productivity. Yet relatively little is known about the precise neurobiological features of chronic TBI leading to dysfunction and disability. This lack of knowledge limits the reliability of therapeutic development in animal models and limits translation across species and into human patients. Part of the problem is that chronic TBI is intrinsically complex, involving heterogeneous damage to the most complex organ system. This results in a multifaceted syndrome spanning across heterogeneous data sources and multiple scales of analysis. This multi-scale heterogeneity makes chronic TBI difficult to understand using traditional analytical approaches that focus on a single endpoint for testing therapeutic efficacy. Single endpoints reflect a small portion of a complex system of changes that describe the holistic syndrome of chronic TBI. In this sense, complex chronic TBI is fundamentally a ‘big-data’ problem requiring pooled information and analytics to evaluate reproducibility in basic discovery and cross-species translation. The proposed project will develop novel applications of cutting edge multidimensional analytics to integrate preclinical chronic TBI data on a large scale. The goal of the proposed project is to develop an integrated workflow for preclinical discovery, reproducibility testing, and translational discovery both within and across chronic TBI types. The project team is well-positioned to execute this project given that with prior federal funding it built one of the largest multicenter, multispecies repositories of neurotrauma data to-date, housing detailed multidimensional outcome data on nearly 4000 mice, rats, pigs, and monkeys. The proposed VA merit award will expand these data with new data-donations collected from 5 preclinical TBI research laboratories across the US, including chronic (>1 month) TBI models of penetrating injury, closed head injuries, repeated mild injuries, acceleration/ deceleration, lateral fluid percussion, and blast injuries. The project will harmonize these existing data resources into a single data pool, enabling application of recent innovations from data science to render complex multidimensional endpoint data into robust syndromic patterns that can be visualized and explored by researchers in a user-friendly manner. The project will accelerate data-driven-discovery, scientific reproducibility, hypothesis-generation, and ultimately precision medicine for chronic TBI.