QLTI-25. Developing a computable phenotype for primary central nervous system lymphoma

Abstract BACKGROUND Primary central nervous system lymphoma (PCNSL) is a rare intracranial pathology with a poor prognosis. Due to its rarity, large-scale studies are challenging to conduct at individual institutions, and thus it is important to be able to identify patients with this diagnosis easily in electronic records for population studies. However, this can be challenging as diagnostic codes are non-specific. The aim of this study was to create a computable phenotype (CP) for PCNSL from structured and unstructured data to identify patients with this condition in a large electronic health records (EHR) and multi-institutional databases. METHODS We used the University of Florida Health Integrated Data Repository, a centralized clinical data warehouse that stores clinical and research data from various sources within the UF Health system, including the EHR system. We performed multiple iterations to refine the PCNSL-relevant diagnosis codes, procedure codes, medication codes, and keywords through manual chart review of patient data and evaluated the performances of various possible proposed CPs constructed from the relevant codes and keywords. RESULTS Multiple rounds of manual chart reviews were undertaken to refine CP elements and identify the most relevant codes and keywords. The final CP algorithm for identifying PCNSL patients was selected based on the best F1 score. CONCLUSIONS We developed a CP algorithm for identifying patients with PCNSL using both structured and unstructured EHR data from a large tertiary care center. Utilization of CPs to identify patient cohorts from health databases will reduce the need for manual chart review to identify those who meet inclusion criteria while simultaneously minimizing possible biases from misclassification errors, thus improving access to and quality of real-world data for patients with this disease.