ENACT: Translating Health Informatics Tools to Research and Clinical Decision Making

点击翻译按钮获取中文摘要

Several challenges exist in the conduct of EHR-based translational research. First, CTSA hubs vary substantially in their capacity to address challenges in EHR data collection, data quality, data harmonization, methodology for deep phenotyping, maintaining patient privacy, variability in ontology, and limited ability to transfer data beyond institutional firewalls. Second, there is an unmet need for readily available, easily accessed informatics tools that facilitate EHR-based research and can be rapidly disseminated and implemented across all CTSA hubs. Third, CTSA hubs seek guidance on the complicated data use agreements (DUAs) and governance needed to enable data sharing and analysis of shared data. With funding from NCATS, we created a federated system, the ACT Network, that crafted a broad DUA and stood-up local clinical data warehouses (CDWs) at 57 CTSA hubs, created an information superhighway to query the CDWs that include >142M patients, and democratized data access for cohort discovery to all CTSA hub investigators. We initially developed ACT to support the planning and design of multisite clinical trials, which it did well and additionally highlighted the potential value of EHR data for deeper analysis. While the ACT Network has limited analytic capacity in its present form, we will now address this opportunity to fully leverage the research potential of EHR data from almost half the US population through Evolve to Next-Gen ACT (ENACT). We will create a user-friendly collaborative research and computing environment with cutting edge analytical methods. We will start with tools and a dashboard to monitor data quality, provide guidance to individual sites to improve data quality, and provide contextual reports that help investigators interpret their data. We will also apply natural language processing to extract clinical concept data from reports and notes in the EHR, provide user- friendly interfaces that are interoperable with common data models (i2b2, OMOP, PCORnet), expand ontologies (lifestyle factors, genetic variants, retired codes), and provide other sophisticated informatics tools, including those developed by our team and by others. In parallel, we will create a platform and provide statistical and machine learning capacity that clinical and translational scientists can apply to EHR data, either through federated analyses or, for more complex compute-intensive analyses, in a temporary enclave. We envision leveraging these informatics tools and EHR data to enable clinicians to generate evidence that can be applied to improve patient care. With every step, we will design for dissemination and sustainability to foster a learning informatics system. We will prioritize unmet needs among stakeholders, solicit input on the desired features, and ensure that ENACT satisfies the needs of targeted end users. We will leverage the I-Corps@ NCATS program for customer discovery, beta testing, and business model development for sustainability. We will collect data through pilot trials of each tool and resource that will be used to create marketing materials and to develop sustainability models that include cost-recovery based on real-world time and effort required.