OHDSI Phenotype Library is an open community resource maintained by the OHDSI community to support phenotype development, evaluation, sharing and re-use. The Phenotype Library is maintained by community librarians, and volunteer collaborators who curate the content. The content that includes cohort definition specificatons (e.g. rules, algorithms, code set), interpretation of the performance characteristics of each cohort definition is stored in a version controlled manner in the OHDSI Phenotype Library GitHub repository. The entire library of cohort definition specification.
This application is the Cohort Diagnostics tool - an Open Source software part of OHDSI HADES that takes an instatiated cohort and runs a set of data diagnostics on it (e.g. incidence rate, temporal characterization, visit context, co-occurrence matrix) to allow for empirical evaluation of the performance characteristics of individual cohort using descriptive data and visualization. It is also a tool that allows for easy exploration, and extraction of cohort definitions that are stored in the OHDSI Phenotype Library Github repository.
The OHDSI Phenotype work group is responsible to facilitate the generation and maintenance of the content in the library and the application. The workgroup page is in OHDSI MS Teams environment . Its mission is To improve the quality and reliability of the evidence generated from observational data, by advancing the science of phenotype development and evaluation. Please see the workgroups channel on Mission, Area of focus and priorities here.
For rule based cohort definitions: To be included in the OHDSI Phenotype Library GitHub repository OHDSI Phenotype Library GitHub repository, every cohort definition should have at least one full result set from Cohort Diagnostics executed on at least one data source, with output contributed to the Phenotype library. All rule based cohort definitions are expressed in JSON and SQL (OHDSI SQL) instructions that are compatible with OHDSI analytic tools and OHDSI OMOP CDM v5.0+. For phentoype, literature review when available are organized using a standardized template
For probabilistic cohort definitions: The repository will have a binary classification model to find people based on statistical threshold. Artificats may include R file to build the model and noisy labels used.
The 2021 areas of focus and Objective Key Results for the phenotype workgroup are available for review.
How to download the library contents: To download the full set of phenotypes and cohort definitions please go to the OHDSI Phenotype Library GitHub repository.
How to contribute a full set of phenotype library diagnostics across the full library: Please execute the Phenotype Library Diagnostics study package and submit the results to the coordinating site.
How to run diagnostics on your cohorts using Cohort Diagnostics: You can develop your own cohort using ATLAS and evaluate it using CohortDiagnostics, by following the instructions at OHDSI Cohort Diagnostics..