RECOVER-EHR-PCORnet Research Program

Project: Research project

Project Details

Description

The NIH Researching COVID to Enhance Recovery (RECOVER) initiative aims to study the public health impact of post-acute sequelae of SARS-CoV-2 (PASC) infection that emerge and evolve over time. This project will leverage the Patient-Centered Clinical Research Network (PCORnet®) for PASC data science research in adults and children. Northwestern will participate as one of 40 institutions contributing structured data from the existing local PCORnet common data model (CDM) data mart (including clinical data such as diagnosis codes and laboratory results, and geospatial data) to a single, centralized study analytic data repository within a secure virtual private cloud. The repository is modeled on the existing Amazon Web Services cloud infrastructure that is leveraged by the INSIGHT network within PCORnet. In addition to PASC cases, we will also construct a control cohort using propensity score matching with COVID-19 patients not meeting the initial RECOVER cohort screening definition. The study team will supplement the CDM standardized data within the repository by linking to vaccine and death data. We will use advanced machine learning methodologies to develop PASC phenotypes, subphenotypes, risk factors and predictive models, and validate in our local data. Based on the screening phenotypes developed, in a second aim we will identify a narrow group to analyze for feature recognition in clinical notes, in order to better account for functional and symptomatic manifestations of PASC that are not well-captured in discrete data. This natural language processing analysis will be limited to patients at risk for syndromic PASC and matched controls.
StatusActive
Effective start/end date10/1/21 → 5/23/25

Funding

  • Children's Hospital of Philadelphia (EHR-02-21 // OT2HL161847-01)
  • National Heart, Lung, and Blood Institute (EHR-02-21 // OT2HL161847-01)

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.