Identification of an ANCA-associated vasculitis cohort using deep learning and electronic health records

Open Webpage

Liqin Wang, John Laurentiev, Claire Cook, Shruthi Srivatsan, Yining Hua, Jie Yang, Eli Miloslavsky, Hyon K. Choi, Li Zhou, Zachary S. Wallace

Published: 2025, Last Modified: 15 Oct 2025Int. J. Medical Informatics 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Traditional methods for identifying AAV cases rely on clinical registries or billing codes, which may miss key subgroups.•Free-text clinical notes contain valuable information on diagnoses and symptoms that can aid case identification.•We found that a deep learning approach outperforms rule-based algorithms by identifying more AAV cases at the patient level.