Clinical Documentation in the Age of Artificial Intelligence: A Critical Assessment of ChatGPT-4.0’s Operative Notes for Cesarean Births [ID 1201]

INTRODUCTION: ChatGPT-4.0, the most readily available model to date, has shown promise in generating medical documents, but its capacity to meet clinical documentation standards in obstetrics remains underexplored. This study assesses the completeness of cesarean birth operative notes generated by ChatGPT-4.0 based on standard surgical documentation requirements. METHODS: Twenty cesarean birth operative notes were generated using ChatGPT-4.0. Each note was evaluated for the presence of essential elements, including date of surgery, patient identifiers, surgeon and assistants’ names, anesthesia provider, preoperative and postoperative diagnoses, procedure, and additional procedural details. RESULTS: All 20 notes contained the date of surgery, names of the surgeon and assistants, the name of the anesthesia provider, the procedure performed, and estimated blood loss. A majority of notes included type of anesthesia (95%), method of abdominal access (95%), postoperative diagnosis (80%), skin preparation and draping (80%), patient identifiers (75%), preoperative diagnosis (75%), and specimens sent (70%). However, fewer notes included dressings (45%) and patient position (35%). Significant gaps were found in the documentation of date of dictation (20%), irrigation (5%), and sponge/instrument counts (0%). CONCLUSIONS/IMPLICATIONS: Although ChatGPT-4.0 successfully generated cesarean birth operative notes containing much of the essential documentation elements, notable deficiencies were observed, particularly in dictation dates, irrigation, and sponge/instrument counts. These findings highlight the need for further refinement to ensure that artificial intelligence–generated notes meet clinical documentation standards in obstetrics. Further research is warranted to address these gaps and enhance its utility in medical documentation.