Description

Abstract: Computable phenotypes (CP) enable researchers to assemble complex and highly specific cohorts of patients by leveraging multiple clinical variables present in electronic health record data. Here, we describe our process of developing and validating CPs for the pediatric chronic conditions of Crohn’s disease, glomerular disease and type 2 diabetes from the PEDSnet clinical data research network. Our results demonstrate how such CP algorithms can enable better case identification than by using diagnostic codes alone.

Learning Objective 1: To discuss how to use computable phenotypes to assemble patient cohorts from electronic health record data.

Learning Objective 2: To describe methods of internal and external validation of computable phenotypes.

Authors:

Levon Utidjian (Presenter)
Children's Hospital of Philadelphia

Ritu Khare, Children's Hospital of Philadelphia
Hanieh Razzaghi (Presenter)
Children's Hospital of Philadelphia

Amanda Dempsey, Children's Hospital of Colorado
Michelle Denburg, Children's Hospital of Philadelphia
Chris Forrest, Children's Hospital of Philadelphia
Charles Bailey, Children's Hospital of Philadelphia

Presentation Materials:

Tags