ICEES+ Knowledge Provider: Leveraging Open Clinical and Environmental Data to Accelerate and Drive Innovation in Translational Research and Clinical Care.

ICEES 知识提供商：利用开放的临床和环境数据加速和推动转化研究和临床护理的创新。

• 批准号: 10333478
• 负责人: Stanley Carlton Ahalt
• 金额: $ 98.22万
• 依托单位: UNIV OF NORTH CAROLINA CHAPEL HILL
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-01-23 至 2022-01-22
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10333478
• 关键词: Address; Algorithms; Asthma; Automobile Driving; Award; Bayesian neural network; Clinical; Clinical Data; Complement; Computational algorithm; Data; Data Analyses; Development; Disease; Ecosystem; Ensure; Environmental Exposure; Etiology; Event; Funding; Genetic Polymorphism; Goals; Infrastructure; Institution; Institutional Review Boards; Knowledge; Methods; Modeling; Modification; Multivariate Analysis; National Institute of Environmental Health Sciences; Neural Network Simulation; Pathway interactions; Patients; Pharmaceutical Preparations; Phase; Privacy; Protocols documentation; Provider; Publishing; Rare Diseases; Registries; Research; Schools; Science; Secure; Security; Services; Specificity; Statistical Algorithm; Statistical Methods; Stress Tests; Time; Translational Research; Variant; Visit; Work; adverse outcome; application programming interface; autoencoder; body system; clinical care; cohort; data sharing; design; improved; innovation; insight; liver injury; machine learning method; patient mobility; patient privacy; preservation; programs; prototype; recurrent neural network; research and development; spatiotemporal; tool

As part of the feasibility phase of the Translator program, we have developed a disease-agnostic framework and approach for openly exposing clinical data that have been integrated at the patient- and visit-level with environmental exposures data: the Integrated Clinical and Environmental Exposures Service (ICEES). We have validated ICEES and demonstrated the service’s ability to replicate and extend published findings on asthma, while also supporting open team science, accelerated translational discovery, and integration with the broader Translator ecosystem. This proposal aims to move ICEES from prototype to development via creation of an ICEES+ Knowledge Provider (KP). Specifically, we aim to address three major challenges that we have identified through research and development (R&D) of the prototype ICEES in an effort to improve the quality, value, and impact of query answers and assertions. Specific Aim 1. Advance the rigor of insights and assertions that ICEES provides. Our prototype ICEES currently provides the ability to dynamically define cohorts and conduct simple statistical associations to examine bivariate relationships between feature variables. Recently, we have identified an approach to extend the bivariate functionalities to support multivariate analysis of the data. For the proposed work, we will apply multivariate analyses, including traditional statistical methods (e.g., regression models) and machine learning methods (e.g., bayesian neural network models, variational autoencoder models), and systematically quantify the extent of data loss and analytic bounds when algorithms are imposed on the ICEES+ KP open application programming interface (API) versus the Institutional Review Board (IRB)– protected, fully identified, pre-binned, underlying integrated feature tables. The overall goal is to provide users with more rigorous insights and estimates of the robustness, validity, accuracy, and specificity of knowledge and assertions generated via the ICEES+ KP OpenAPI. Specific Aim 2. Address issues related to space–time and causality. Clinical and environmental data are inherently spatiotemporal, with observations or events that are contingent on space and time and may be causally related. For the proposed work, we will evaluate and implement technical approaches (e.g., ICEES+ design modifications), spatiotemporal statistical algorithms (e.g., conditional auto-regression), recurrent neural network models, and causal inference models. As part of this effort, we will derive insights from and contribute real-world evidence to support Causal Activity Models and Adverse Outcome Pathways. We also will explore approaches for incorporating into ICEES+ nationwide public data on school exposures—data that will allow us to begin to address patient mobility. Specific Aim 3. Evaluate the security of the ICEES+ KP to ensure that patient privacy is preserved as new capabilities are enabled. ImPACT is an NSF-funded package of tools and services that provides end-to-end infrastructure and support for privacy-assured research and computation on sensitive data. Over the award period, we will implement and evaluate ImPACT security protocols, focusing initially on application of the ImPACT secure multiparty computation (SMC) algorithm as a method to support secure multi-institutional sharing of data on rare diseases and events—a functionality that is not currently supported by ICEES. In addition, we will evaluate other ImPACT security protocols, working under the guidance of a security advisor and in the context of driving use cases and capabilities developed under Specific Aims1 and 2. Importantly, the project aims will be driven by three use cases and associated high-value queries designed to complement and extend our asthma-focused work on the prototype ICEES: (1) an asthma cohort from the Environmental Polymorphism Registry (EPR) at the National Institute for Environmental Health Sciences (NIEHS); (2) a primary ciliary disease cohort (PCD) from the UNC PCD Registry; and (3) a drug-induced liver injury (DILI) cohort from the National DILI Network. These use cases will invoke new diseases, new data types, new organ systems, new institutions, and new queries, thereby stress-testing the ICEES framework and approach and moving it from prototype to development as the ICEES+ KP.

作为翻译程序可行性阶段的一部分，我们开发了一种疾病不可能的 框架和方法公开公开已集成的临床数据 带有环境暴露数据的患者和访问级别的数据：综合临床和 环境暴露服务（ICEES）。我们已经验证了冰，并证明了 服务复制和扩展有关哮喘的发现的能力，同时也支持 开放团队科学，加速转化发现，并与更广泛的集成 翻译生态系统。该建议旨在通过 创建ICEES+知识提供商（KP）。具体而言，我们的目标是解决三个主要 我们通过原型的研发（R＆D）确定的挑战 冰块为了提高查询答案和断言的质量，价值和影响。 特定目的1。提高冰冰提供的洞察力和断言的严格性。我们的 原型冰目前提供了动态定义队列并进行简单的能力 统计关联以检查特征变量之间的双变量关系。最近， 我们已经确定了一种扩展双变量功能以支持多变量的方法 数据分析。对于拟议的工作，我们将应用多元分析，包括 传统的统计方法（例如，回归模型）和机器学习方法（例如， 贝叶斯神经网络模型，变异自动编码器模型）和系统量化 当算法施加在冰+ KP上时，数据丢失和分析界限的程度 开放申请编程接口（API）与机构审查委员会（IRB） - 受保护，完全识别，预先固定的基础集成特征表。总体目标是 为用户提供更严格的见解和稳健性，有效性，准确性， 通过Icees+ KP OpenAPI产生的知识和断言的特异性。 具体目标2。解决与时空和因果关系有关的问题。临床和 环境数据本质上是时空的，有观察结果或事件 取决于空间和时间，可能意外相关。对于拟议的工作，我们将 评估和实施技术方法（例如ICEES+设计修改）， 时空统计算法（例如，有条件的自动回归），复发性神经网络 模型和因果推理模型。作为这项努力的一部分，我们将获得并从中获得见解 贡献实际证据以支持因果活动模型和不利结果 途径。我们还将探索将融合到ICEE+全国公众中的方法 有关学校暴露的数据 - DATA将使我们能够开始解决患者的流动。 特定目标3。评估冰+ KP的安全性，以确保患者隐私是 保留为新功能。影响是NSF资助的工具包， 提供端到端的基础架构和支持隐私研究的支持以及 敏感数据计算。在整个奖项期间，我们将实施和评估影响 安全协议，最初专注于撞击安全多方计算的应用 （SMC）算法作为支持稀有数据的安全多机构共享的一种方法 疾病和事件 - 当前尚未得到ICE的功能。另外，我们 将在安全顾问的指导下评估其他影响安全协议 在驱动用例和功能的背景下，在特定的目标1和2下开发了功能。 重要的是，该项目的目标将由三种用例和相关的高价值驱动 旨在完成和扩展我们在原型冰上以哮喘为重点的工作的查询： （1）来自国家环境多态性注册中心（EPR）的哮喘队列 环境健康科学研究所（NIEHS）； （2）原发性睫毛疾病队列（PCD） 来自UNC PCD注册表； （3）来自国家的药物诱导的肝损伤（DILI）队列 DILI网络。这些用例将调用新疾病，新数据类型，新器官系统， 新机构和新查询，从而强调了冰块框架和方法 并将其从原型转移到开发中，如冰+ kp。