Skip to main content
AAO-HNSF 2023 Annual Meeting & OTO Experience Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Facial Plastic and Reconstructive Surgery

Use of Open Source Artificial Intelligence in Writing Patient Instructions

Tuesday, October 3, 2023
1:36 PM - 1:42 PM East Coast USA Time
Location: 205ABC

    Presenting Author(s)

  • CW

    Christina N. Ward, MD, MS

    Resident Physician
    University of Tennessee Health Science Center
    Memphis, Tennessee, United States

    • Co-Author(s)

  • Travis J. Clarke, MD

    Resident
    University of Tennessee Health Science Center
    Memphis, TN, United States

  • Sanjeet V. Rangarajan, MD, MEng

    Associate Professor of Otolaryngology - Head and Neck Surgery
    University of Tennessee Health Science Center
    Memphis, Tennessee, United States

  • Anas Eid, MD

    Assistant Professor - Facial Plastic & Reconstructive Microsurgery
    Department of Otolaryngology - Head and Neck Surgery University of Tennessee Health Science Center
    Memphis, TN, United States

Disclosure(s):
Christina N. Ward, MD, MS: No relevant relationships to disclose.
Sanjeet V. Rangarajan, MD, MEng: No relevant relationships to disclose.
Anas Eid, MD: No relevant relationships to disclose.
Introduction: Artificially intelligent computer systems (AI) are increasing in availability and becoming more accessible to the general public and medical community. Our study aims to explore the potential of AI in writing postoperative patient instructions and informative materials for patients.

Methods: A dataset of patient instructions for various surgical procedures in facial plastic surgery and reconstructive surgery was created using OpenAI ChatGPT. The AI commands were tested, and different command formats were executed to find the best method to produce reliable postoperative instructions materials. The models were evaluated for clarity and accuracy and for how long it took to build the materials, including human editing.

Results: The OpenAI could generate instructions comparable in terms of clarity and accuracy to those written by human authors. A specific format of commands produced more reliable and detailed instructions. Although the model was able to adapt to different treatments and generate instructions with a high degree of accuracy, human editing was required in all instructions, with an average of 5 minutes needed to edit a complete set of procedure instructions. Editing was mostly done to provide better surgical clarification, adjust timeframes or nature of restrictions, or medications used postoperatively.

Conclusions: This study's results demonstrate AI's potential in writing patient instructions for surgical treatments. Specific commands generated better material and more comprehensive patient instructions in ChatGPT. Although AI-generated instructions appear to require some human editing, it is a valuable tool for physicians to ensure that patients receive clear and accurate instructions for their treatments and generate patient education material in a short time but accurately.