Drafting discharge and nursing summaries is one of the heavier tasks for physicians and nurses. Recently, hospitals have begun using generative AI to auto-draft from chart data and cut authoring time substantially. This article reviews published cases and what to consider when adopting it.
Published adoption cases
Oda Hospital (Saga)
In work by Oda Hospital and OPTiM, the time to create discharge nursing summaries with generative AI was reported to fall by up to 54.2% (from about 16.3 minutes on average to roughly 7.5 minutes once operation matured). Notably, an on-premise large language model was used to protect patient information. A related joint study on nursing summaries reported roughly 40% time reduction in an academic journal.
Fujita Health University Hospital (Aichi)
In a discharge-summary support system developed by Fujita Health University Hospital and FIXER, about 92% of physicians who used it reported it helped shorten time and improve work, with a cumulative reduction of roughly 1,000 hours in the three months after introduction. What used to take 10–15 minutes per summary is said to be completed in a few clicks.
Figures and dates are based on the publishing organizations' materials; please check their primary sources for the latest details.
How it works: why it saves time
Generative AI auto-drafts a summary from the progress notes and order/execution data accumulated in the EHR. The clinician's task shifts from writing from scratch to reviewing and correcting the AI's draft, compressing the time spent composing text.
What to watch for in clinical adoption
- PHI protection: verify the data-handling model, such as on-premise or in-hospital processing
- Review process: clinicians must always check and take responsibility for finalizing the AI draft
- Accuracy and scope: decide which documents to start with and how to measure impact
Summary
AI-assisted summary drafting is an area where multiple hospitals have reported substantial, real time savings. Adopted with safe data handling and a clinician review process, it directly eases front-line burden. Sakigake's EHR aims to integrate this kind of AI document drafting in an AI-native way.