A rich resource for Health IT (HIT) events is in urgent need to improve patient safety. We assessed multiple individual and hybrid deep learning models to identify HIT events from the FDA MAUDE database. 48,997 HIT event reports were identified by the optimal model to establish the first database for HIT events. The strategy and the database could help healthcare providers describe, understand, and integrate the events connecting to a fuller spectrum of HIT events.

Learning Objective: To identify and discuss the rich resource for HIT events and promote patient safety through shared learning.

Authors:

Hong Kang, UTHealth
Yang Gong (Presenter)
UTHealth

Presentation Materials:

• Abstract/Manuscript - 1