RAPID: Early Warning Algorithms for Predicting Ebola Infection Outcomes

这项调查涉及开发数学算法，为早期发现埃博拉病毒感染提供快速诊断工具。与目前的方法不同，拟议的方法不要求受试者在检测病毒时出现症状。拟议的方法利用了这样一种观察，即免疫系统的行为就像煤矿中的金丝雀，提供了一个早期预警系统，如果定量地理解，就可以用来识别感染、加快治疗和改善结果。该检测技术包括为例如表征健康免疫系统的标称状态的基因表达数据构建一组数学模型。然后，这些模型被应用于检测受感染对象的免疫系统的新奇或异常行为。最初的模型构建阶段将使用非人类灵长类动物和小鼠的数据来建立该方法的可行性。转录分析已被广泛应用于确定疾病分类、诊断和预后的标志物。已经开发了许多方法来从静态的角度来识别响应不同生物状态之间的变化的信号通路，即健康状态和疾病状态。然而，生物状态之间的转换是一个复杂的动态过程，信息丰富。在流感的初步工作中，使用实验感染流感病毒的健康个体的数据，为400多条途径建立了基因表达的非线性模型。在这些途径中，胞质DNA/RNA传感途径(一种检测病原体相关核酸的系统)首先在大多数出现症状的受试者中显示出基因表达的变化，反映了免疫系统对入侵病原体的初始反应。此外，随着免疫系统反应的进展，有一系列异常的通路信号，反映在基因表达的变化上，这可以为在外部可观察到的疾病症状出现之前很久就发现病原体提供早期预警信号。在这个项目中，将研究哺乳动物细胞对埃博拉病毒的反应的级联通路，目的是描述其疾病特异性进化的特征，这些特征将被用于在可观察到的症状之前识别用于诊断的分子特征。还将探讨建模过程的敏感度和稳健性。

This investigation concerns the development of mathematical algorithms to provide rapid diagnostic tools for the early detection of infection by the Ebola virus. Unlike current methods, the proposed approach will not require that the subject be symptomatic for detection of the virus. The proposed methodology exploits the observation that the immune system behaves like a canary in a coal mine, providing an early warning system that, if quantitatively understood, could be used to identify infection, accelerate treatment and improve outcomes. The detection technique consists of building an array of mathematical models for, e.g., gene expression data, that characterize the nominal state of the healthy immune system. These models are then applied to detect novel, or anomalous behavior of the immune system in infected subjects. The initial model building phase will employ non-human primate and mouse data to establish viability of the approach.Transcriptional analysis has been widely applied to identify markers for disease classification, diagnosis, and prognosis. Many methods have been developed to identify the signaling pathways that respond to the changes between varying biological states, i.e., healthy and disease states, from a static viewpoint. However, the transition between biological states is a complex dynamic process that is information rich. In preliminary work on influenza, a nonlinear model of gene expression was built for over 400 pathways using data from healthy individuals that were experimentally infected with influenza virus. Of these pathways, the cytosolic DNA/RNA sensing pathway (a system for detecting pathogen-associated nucleic acids) was the first to exhibit changes in gene expression in the majority of subjects who became symptomatic, reflecting the immune system's initial response to an invading pathogen. Moreover, as the immune system response progressed, there was a cascade of anomalous pathway signaling, reflected by changes in gene expression, which could provide an early warning signature for detection of a pathogen well before externally observable symptoms of the disease appear. In this project, the pathway cascade of the mammalian cell response to Ebola virus will be investigated with the goal of characterizing the features of its disease-specific evolution, which will be used to identify molecular signatures for diagnosis prior to observable symptoms. The sensitivity and robustness of the modeling procedure will also be explored.