AI-Generated Draft Replies Integrated Into Health Records and Physicians' Electronic Communication

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Workflow Optimization

Background:

• Physician workload is a significant concern, with patient messaging contributing to administrative burden.
• Generative artificial intelligence (GenAI) offers potential solutions for streamlining communication and reducing physician burnout.
• The impact of GenAI on physician time and communication efficiency requires empirical investigation.

Purpose of the Study:

• To investigate the association between GenAI-drafted replies for patient messages and physician time spent on answering messages.
• To assess the impact of GenAI on the length of physician replies to patient messages.
• To evaluate physician perceptions of GenAI tool utility in clinical communication.

Main Methods:

• A randomized waiting list quality improvement (QI) study was conducted in an academic health system.
• Primary care physicians were randomized into immediate and delayed GenAI activation groups.
• Mixed-effects models analyzed changes in message reading time, reply time, and reply length.

Main Results:

• GenAI use was associated with a 21.8% increase in message read time (P=.008).
• No significant change was observed in physician reply time (P=.33).
• GenAI-drafted replies significantly increased message length by 17.9% (P<.001).

Conclusions:

• GenAI-drafted replies in patient messaging led to increased reading time and reply length, without altering reply time.
• Physicians perceived value in GenAI tools but identified areas for improvement.
• Further rigorous studies are essential to validate GenAI's performance and explore its impact on patient experience and different AI models.