Re-imagining discharge summary training through artificial intelligence

Area of Science:

• Medical Education
• Artificial Intelligence in Healthcare
• Clinical Documentation

Background:

• Discharge summary (DS) writing is crucial for junior physicians, but current training methods face quality, accuracy, and timeliness issues.
• Existing faculty-intensive, small-group teaching models are not scalable or sustainable for widespread training.
• There is a critical need for efficient, high-quality training in DS writing with personalized feedback.

Purpose of the Study:

• To develop and evaluate a novel educational model integrating artificial intelligence (AI) for DS writing training.
• To assess the feasibility and potential benefits of a human-in-the-loop AI system for providing feedback on DS.
• To address the limitations of current DS training programs in terms of scalability and resource efficiency.

Main Methods:

• A proof-of-concept evaluation compared AI-generated feedback quality against human feedback using a standardized rubric.
• Development of an asynchronous e-learning module using customized generative-AI (cGen-AI) for draft feedback generation.
• Inclusion of human moderation as a crucial step for quality assurance and safety in the AI feedback loop.

Main Results:

• AI platforms demonstrated potential in generating feedback comparable to human trainers.
• The developed model integrates cGen-AI for scalable, individualized feedback generation.
• The human-in-the-loop approach ensures quality and safety while reducing faculty workload.

Conclusions:

• The proposed AI-integrated model offers a scalable and sustainable solution for DS writing education.
• This approach can significantly reduce faculty burden and improve the consistency of feedback.
• The model's success could pave the way for similar AI applications in other medical training areas.