The use of artificial intelligence (AI) could help physicians prepare hospital patient discharge summaries more efficiently and with greater accuracy and completeness, according to a study.
Discharge summaries provide essential information for patients when they leave a hospital. The data includes essential information such as a patient’s treatment, medication needs, changes in medication, unresolved medical issues and follow-up needs as an outpatient.
Primary care physicians rely on discharge summaries to continue their care of patients, and discharge summaries help to provide a patient care roadmap going forward.
This study, published in the journal npj Digital Medicine, focused on longer, complex hospital stays, where there is a larger volume of medical information to summarize. Such in-house scenarios can be a significant burden for providers and add to their time spent on discharge documentation.
The method centered on an approach that used a large language model (LLM) — a form of AI that can process and generate text — to develop discharge summaries for complex patient cases. Clinicians helped design the approach based on the information physicians need after patient discharge. The LLM process organised patient medical records and other information to create a patient summary that provided the core information for a patient discharge summary—all within a few minutes.
The researchers at Stony Brook University compared LLM-generated and human-authored discharge summaries for 60 internal medicine hospitalisations at Stony Brook University Hospital lasting seven to 21 days, with paired assessment by hospitalists and primary care physicians.
They found that the physician reviewers preferred the AI-generated discharge summaries in 95% of the evaluated hospitalisations. Additionally, the reviewers gave the summaries higher scores for overall quality, readability, accuracy and completeness. They also identified fewer missing clinical details compared to the human-authored summaries.
According to the researchers, primary care physicians also found the AI-generated (LLM) summaries easier to understand, easier to explain to patients and caregivers, and more useful for follow-up care.
“These results provide evidence that artificial intelligence can support discharge documentation for complex hospital stays,” said Rachel Wong, co-corresponding author and clinical associate professor at Stony Brook University.
The team also found that both types of discharge summaries contained errors. Yet, physician reviewers found no statistically significant difference in the estimated severity or likelihood of harm associated with those errors. The authors stress that physician review and verification remain essential before any LLM-generated discharge draft is used for a patient record.
