Preserving bedside teaching: An AI-supported system for capturing spontaneous clinical teaching

Area of Science:

• Medical Education Technology
• Artificial Intelligence in Healthcare

Background:

• Bedside rounds offer unique spontaneous teaching opportunities.
• Current methods for capturing bedside teaching are inadequate, leading to lost learning moments.
• Learners face challenges documenting teaching in real-time during patient care.

Purpose of the Study:

• To develop an AI-supported system, Rounds Copilot, for automatically capturing and structuring bedside teaching.
• To convert spontaneous spoken teaching during rounds into usable learning materials.
• To enhance the preservation and accessibility of clinical teaching content.

Main Methods:

• Developed Rounds Copilot, an AI system integrating speech transcription and natural language processing.
• The system automatically extracts teaching points, classifies them (evidence vs. pearls), and generates assessment questions.
• Utilized a simulated 15-minute teaching encounter for initial processing.

Main Results:

• The system processed a simulated encounter in under three minutes, generating seven teaching kernels and five multiple-choice questions.
• Evidence classification helped learners prioritize information, and assessment questions aided knowledge consolidation.
• Informal resident feedback highlighted the utility of the evidence-versus-pearl distinction and assessment questions.

Conclusions:

• Rounds Copilot effectively transforms dynamic bedside teaching into structured, accessible learning resources.
• The system offers a cost-effective and user-friendly solution for preserving clinical education.
• Future work will focus on real-world validation and clinical deployment pathways.