A study has demonstrated that AI can create more accurate operative reports than surgeons.
Published in the Journal of the American College of Surgeons, it is the first report on fully automated, video-based AI surgical documentation.
The research highlights the potential of AI-driven solutions to reduce administrative burdens and improve surgical documentation.
Surgeons frequently regard the creation of operative reports as essential yet time-consuming. These reports are inevitably subjective and may contain inaccuracies or incomplete information.
The administrative task of documentation has also been recognised as a potential factor in physician burnout.
Recent advancements in AI, especially in computer vision, have allowed automated systems to accurately detect surgical steps from video footage.
Researchers aimed to create a platform that automates the generation of video-based AI surgical operative reports for robotic-assisted radical prostatectomy (RARP).
Using an AI-powered algorithm, surgical steps were automatically identified in video recordings and mapped to pre-specified text to generate narrative AI operative reports.
The accuracy of these AI-generated reports was then compared to traditional surgeon-written reports using an expert review of raw surgical video footage as the gold standard.
The study analysed 158 cases from a tertiary referral centre. It found that:
• 53.2% of surgeon-written reports (84 cases) contained at least one discrepancy compared to the surgical video. Of these, 27.2% (43 cases) had clinically significant discrepancies.
• AI-generated reports showed fewer discrepancies, with 29.1% (46 cases) containing at least one inconsistency.
• Clinically significant discrepancies in AI reports were lower, at 12.7% (20 cases).
• Overall accuracy was significantly higher for AI-generated reports than surgeon-written reports (87.3% vs. 72.8%, p=0.001).
The findings suggest that AI-driven operative reporting can enhance accuracy, reduce the documentation burden, and improve transparency in surgical procedures.
Researchers emphasise the need for further studies to explore the wider implications of this technology, including its potential to standardise documentation practices and reduce subjectivity in operative reporting.
