Creating a Baseline for Artificial Intelligence–Generated Obstetric Operative Reports: Analyzing ChatGPT 3.5 Generate Cesarean Birth Reports [ID 1189]

INTRODUCTION: While artificial intelligence (AI) is becoming more widely utilized throughout medicine, using ChatGPT to generate obstetric operative reports has yet to be examined. ChatGPT-3.5, an older model of ChatGPT, can serve as a baseline for AI efficacy in generating obstetric operative reports. This study examines completeness of cesarean birth operative reports generated by ChatGPT-3.5 to assess weaknesses and provide a baseline for ChatGPT obstetrical notes. METHODS: Twenty cesarean birth operative reports were generated using ChatGPT-3.5. Each note was evaluated for inclusion and completeness of history of present illness, operative findings, technique of resection, limits of resection, technique of reconstruction, and closure technique using a Likert scale. RESULTS: None of the 20 notes demonstrated completeness in any category. Brief history of present illness had a median score of 0. Operative findings, technique of resection, limits of resection, technique of reconstruction, and closure technique all lacked detail with a median score of 2. CONCLUSIONS/IMPLICATIONS: ChatGPT-3.5-generated cesarean birth operative reports demonstrated weaknesses in documenting all examined variables. The most concerning deficit was history of present illness, which was largely absent in generated reports. Gaps in detailed reports included operative findings, technique of reconstruction, limits of resection, technique of reconstruction, and closure technique. These findings highlight ChatGPT-3.5’s inadequacy in generating complete obstetric operative reports. Further research is needed to examine whether newer ChatGPT models address this gap.