Leveraging data-science for discovery in chronic TBI

慢性创伤性脑损伤（TBI）是军事和军事领域最常见的神经系统疾病之一， 平民，影响美国多达530万人，造成760亿美元的医疗保健和损失， 生产力。然而，对慢性TBI的精确神经生物学特征知之甚少， 功能障碍和残疾。这种知识的缺乏限制了动物治疗开发的可靠性。 模型和限制跨物种和人类患者的翻译。部分问题在于慢性TBI 内在复杂，涉及对最复杂器官系统的异质性损伤，通常伴有 四肢多处创伤和心理创伤这导致了多方面的综合症 跨越异构数据源和多个分析尺度。这种多尺度的异质性 使得慢性TBI难以使用传统的分析方法来理解， 用于测试治疗功效的终点。单个端点反映了一个复杂系统的一小部分， 描述慢性TBI的整体综合征的变化。从这个意义上说，复杂的慢性TBI是 从根本上说，这是一个“大数据”问题，需要汇集信息和分析来评估 基本发现和跨物种翻译。拟议的项目将开发切割的新应用 边缘多维分析，以大规模整合临床前慢性TBI数据。的目标 拟议的项目是为临床前发现、再现性测试和 在慢性TBI类型内和跨慢性TBI类型的翻译发现。项目团队有能力执行 这个项目是因为在联邦的资助下，它建立了一个最大的多中心、多物种的储藏库 迄今为止的神经创伤数据，包含超过10000只小鼠，大鼠，猪， 猴子拟议的退伍军人管理局优秀奖将扩大这些数据与新的数据捐赠收集的几个 美国临床前TBI研究实验室，包括慢性（>1个月）穿透性TBI模型 损伤、闭合性头部损伤、重复性轻度损伤、加速/减速、侧向液压冲击、冲击波 损伤以及多发性创伤和压力模型。该项目将把这些现有的数据资源统一到 一个单一的数据池，使数据科学的最新创新的应用，使复杂的 多维终点数据转化为强大的综合征模式，可以可视化和探索， 研究人员以用户友好的方式。该项目将加速数据驱动的发现、科学 重复性，假设生成，并最终为慢性TBI提供精确的药物。

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, often accompanied by polytrauma to extremities as well as psychological wounds. 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 over 10000 mice, rats, pigs, monkeys. The proposed VA merit award will expand these data with new data-donations collected from several 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, blast injuries as well as polytrauma and stress models. 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.