Table 1.

Sample use cases for large language models in health care

Use Case	Possibilities
Task automation	•Drafting prior authorization requests, insurance appeals letters, work letters, disability paperwork, medical leave requests, and more •Triaging patient messages (in-basket) and drafting a skeleton response for provider review
Patient education	•Translating medical jargon into clinic visit instructions or discharge instructions at a patient-friendly reading level •Drafting lifestyle counseling recommendations such as diet, exercise, basic physical therapy etc. •Creating medication tables with clear instructions for use and side effects
Clinical documentation	•Using prior clinic visit notes, diagnoses, laboratory test results, and imaging results to prepare the outlines of a note •Summarizing long hospital course to prepare an outline of a discharge summary •Pulling out key information from a note to identify billing level
Diagnostic assistance	•Offering a basic differential diagnosis based on a problem representation or the available clinical data in the EHR •Identifying case reports of patients with similar problems or rare diseases
Literature review	•Summarizing key studies and their findings in a particular field •Identifying and summarizing the articles relevant to a new research project •Copyediting academic manuscripts to help bridge equity gaps for trainees and providers for whom English is a second language

Definition of abbreviation: EHR = electronic health record.