Loading...
High-quality hospital discharge summaries are crucial for communicating the essential components of a patient's hospitalization and follow-up needs to primary care clinicians and are associated with improved outcomes (e.g., fewer medication errors and readmissions). Nevertheless, many busy clinicians struggle to produce high-caliber discharge summaries and perceive them as an administrative burden. Artificial intelligence (AI) large language models (LLMs) offer an exciting potential solution.
Researchers prompted an LLM (GPT-4 [OpenAI]) to generate discharge summaries for 100 randomly selected patient admissions of 3 to 6 days' duration. Twenty-two blinded physicians reviewed the LLM-generated narratives and the original hospitalist-written …