Machine learning algorithms using natural language processing: A new screening approach for inpatient violence based on clinical notes 

Project Type(s):

Principal Investigator(s):
  • Jacqueline Chipkin
  • Julia Siar
Co-Investigator(s):

Despite increasing violence, verbal abuse, and threats against healthcare workers by hospitalized patients, few methods can accurately and efficiently screen patients for this outcome. Our project aims to create an algorithm powered by natural language processing—a form of artificial intelligence—to “read” clinical notes and identify high-risk patients prior to violent events. To ensure equitable care, we will evaluate our model for biases and design appropriate modifications. A generalizable NLP model will allow rapid identification of patients and their risk factors for earlier interventions and better outcomes.

Project Period:
September 1, 2024 — June 30, 2025

Funder(s):
UW Medicine’s Patients are First Innovation Pilot grant, 2023; Clinician Scientist Training Program

Research Program(s):
Related Fellowship(s):