Eliminating Nonvalue-Added Tasks in Health Care: Artificial Intelligence in the Orthopedic Office

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Documentation Improvement

Background:

• Physician burnout is a significant concern, partly due to administrative burdens like clinical documentation.
• Efficient and accurate documentation is essential for patient care and healthcare system operations.
• Generative artificial intelligence (AI) offers potential solutions for streamlining documentation processes.

Purpose of the Study:

• To evaluate the efficiency and proof-of-concept of a generative AI documentation prototype in an orthopedic practice.
• To assess the usability of AI-generated clinical notes.
• To estimate the time advantage of AI-assisted documentation compared to traditional physician documentation.

Main Methods:

• A comparative study analyzed 68 new patient encounters, comparing AI-generated notes (using ChatGPT) with physician-generated notes.
• Documentation time, editing time for AI notes, visit complexity, and note quality were measured.
• Notes were generated from unedited audio recordings of encounters without identifiable patient information.

Main Results:

• Editing AI-generated documentation took a median of 55.1 seconds, versus 81.2 seconds for physician-generated documentation.
• 91% of AI-generated notes were rated 4 or 5 on a 5-point Likert scale for quality.
• Critical errors were found in 8.8% of AI-generated documents; physician documentation time varied with complexity, while AI documentation time showed minimal variation.

Conclusions:

• Generative AI can produce clinically usable notes across different encounter complexities.
• Physician editing of AI-generated documentation is faster than independent documentation.
• AI-assisted documentation holds significant potential for enhancing workflow efficiency in clinical settings, with human oversight remaining essential.