Evaluation of a Natural Language Processing Model to Identify and Characterize Patients in the United States With High-Risk Non–Muscle-Invasive Bladder Cancer

Identifying and characterizing patients with non-muscle invasive bladder cancer using unstructured EHR data is methodologically challenging, and NLP approaches had not been validated for this specific oncology application at scale in real-world US data.

Target RWE researchers developed and evaluated an NLP model applied to EHR data from the Komodo Healthcare Map to identify and characterize NMIBC patients, validating model performance for extracting clinically meaningful patient profiles from unstructured clinical notes.

Demonstrating that NLP can reliably identify and characterize NMIBC patients from real-world EHR data expands Target RWE's capability to study rare oncology populations, enabling oncology drug development partners to access previously inaccessible patient cohorts.

NLP-based EHR patient identification for NMIBC oncology population characterization, Unstructured data extraction methodology for rare oncology patient cohorts in pharma partnerships