PCORsearch: A scalable, user-centric platform for self-service cohort discovery and feasibility analysis of PCORnet data.

Regulatory and institutional restrictions on access to electronic health records (EHR) result in prolonged delays, often extending several weeks or more, when these records are accessed to assess the feasibility of proposed research. Investigators are restricted from directly querying EHR data servers for feasibility analysis to protect patient privacy, and this responsibility is shifted to independent data analysts. To address this bottleneck, we developed the Patient-Centered Outcomes Research Search Tool (PCORsearch), a web-based set of applications that enables investigators to independently analyze deidentified EHR-derived data while maintaining regulatory compliance. PCORsearch allows users to interactively explore medical terminology codes and data definitions from the PCORnet Common Data Model (CDM) to construct custom cohorts and retrieve feasibility counts. Cohorts are defined using a custom grammar tailored for querying PCORnet CDM data, with attention to usability and security. In addition to cohort discovery, the platform supports the generation of summary statistics and visualizations to further characterize the identified cohorts. Cohorts can be saved and shared between users to aid in collaborative research. The platform ships complete with utility applications to manage and request access to the platform, as well as a companion application where analysts can make use of the web interface and custom grammar to get line-level data access. By streamlining investigator-analyst feasibility assessment workflows, PCORsearch may substantially accelerate early-stage research planning. It enables broader access to large-scale clinical data resources in a modern, secure, compliant manner.