Collaborative Platform for Developing Sepsis Products by Leveraging Sepsis Endotypes Developed Using a Unified Biomarker-Clinical Dataset

主要研究者/项目负责人（最后，第一，中间）：Reddy，Jr.，鲍比 项目摘要： 脓毒症是一种知之甚少的临床综合征，其特征在于免疫系统的免疫调节异常。 对感染的反应。它是导致死亡的主要原因，也是美国医院治疗费用最高的疾病， 2011年，这一负担为203亿美元，占全国医疗保健系统总成本的5.2%。的一个主要 临床医生面临的挑战是识别和识别患有脓毒症和即将发生的器官功能障碍的患者。 脓毒症的临床表现是高度可变的，并且感染和器官功能障碍的体征可被诊断为是严重的。 微妙的，可以模仿其他条件。脓毒症也是高度时间关键的。抗生素每延迟1小时， 急诊科（艾德）分诊或器官功能障碍或休克的发生与3-7%的增加相关 结果不佳的几率这些条件创造了一个环境，医生必须诊断 一个复杂的，异质的条件下，在很短的时间内与有限的信息。目前迫切需要 寻找一种可以快速评估患者是否有败血症风险的工具。 Prenosis是一家专注于阐明宿主对感染反应失调的复杂性的公司。在 我们与4家医院合作，建立了世界上最大、增长最快的数据集，数据丰富 将联合收割机时间序列生物标志物数据与疑似医院感染患者的临床数据相结合的生物库 环境.该数据集和生物库目前有> 2，000名患者，> 70，000种专有生物标志物 测量，> 1，200，000个电子病历（EMR）参数和> 25，000个样本库（所有 伴随着完整的时间序列EMR数据）。目前，我们已经为6家医院签订了合同。 合作伙伴关系，如果我们的管道满负荷运行，每年有可能将数据集扩展至> 65，000例患者 容量 在这个拟议的项目中，Prenosis将最终确定NOSISTM平台的第一版， 专有数据集和生物库从目前的约2，000名患者到超过10，000名患者（Aim 1）。使用目前的2,000例患者数据集，我们已经证明了最初有希望的败血症内型， 可以用于各种重要的临床问题。当我们将数据集扩展到10，000名患者时，我们将使用 无监督机器学习算法在大约一半的患者（5,000人）身上进行了训练，以明确证明 这些内型的稳健性和有用性。另一半患者（另外5，000名）将用作 ML算法确定的内型的多中心验证队列（目标2）。我们还将最终确定 NOSISTM产品的实际软件平台（目标3），包括数据安全、限制访问 合作者培训和共同开发产品，以及具有潜力的商业伙伴关系的模板 合作者（最初专注于HIT公司和制药公司）。

Principal Investigator/Program Director (Last, first, middle): Reddy, Jr., Bobby Project Summary: Sepsis is a poorly understood clinical syndrome characterized by a dysregulation of the immune system’s response to infection. It is the leading cause of death and is the most expensive condition treated in U.S. hospitals, exerting a $20.3 billion burden in 2011, 5.2% of total costs to the healthcare system nationwide. One of the major challenges facing clinicians is to identify and recognize patients with sepsis and impending organ dysfunction. The clinical manifestations of sepsis are highly variable and the signs of infection and organ dysfunction can be subtle and may mimic other conditions. Sepsis is also highly time critical. Every 1-hour delay in antibiotics after emergency department (ED) triage or the onset of organ dysfunction or shock is associated with a 3–7% increase in the odds of a poor outcome. These conditions have created an environment where physicians have to diagnose a complex, heterogeneous condition in a short timeframe with limited information. There is currently a dire need for a tool that can quickly assess if a patient is at risk for sepsis. Prenosis is a company focused on elucidating the complexity of dysregulated host response to infection. In partnership with 4 hospitals, we have built the world’s largest and most rapidly growing dataset & data-rich biobank that combine time series biomarker data with clinical data for patients suspected of infection in hospital environments. This dataset & biobank currently have >2,000 patients, >70,000 proprietary biomarker measurements, >1,200,000 Electronic Medical Record (EMR) parameters, and >25,000 samples banked (all with accompanying full time series EMR data). We currently have executed contracts for 6 total hospital partnerships, with the potential to expand the dataset by >65,000 patients per year if our pipeline were at full capacity. In this proposed project, Prenosis will finalize the first version of the NOSISTM platform by growing our current proprietary dataset & biobank from its current size of about 2,000 patients to over 10,000 total patients (Aim 1). Using the current 2,000 patient dataset, we have demonstrated initial promising endotypes of sepsis that could be useful for a variety of critical clinical problems. As we grow the dataset to 10,000 patients, we will use unsupervised machine learning algorithms trained on roughly half of the patients (5,000) to definitively prove the robustness and usefulness of these endotypes. The other half of the patients (other 5,000) will be used as a multi-site validation cohort for the endotypes determined by the ML algorithms (Aim 2). We will also finalize the actual software platform for the NOSISTM product (Aim 3), including data security, restricted access by collaborators to train and jointly develop products, and templates for business partnerships with potential collaborators (with an initial focus on HIT companies and pharma companies).