Introduction: Intraoperative errors remain a significant cause of surgical morbidity and mortality, particularly in resource-limited settings. While robotic and navigation technologies have improved precision in specialized centers, their high costs hinder widespread adoption. Artificial intelligence (AI) offers new opportunities, but most existing approaches emphasize automation rather than human augmentation. By reframing AI as augmented intelligence, this study proposes a collaborative framework that empowers surgeons in real time, enhancing precision, reducing preventable errors, and extending surgical safety across diverse healthcare systems. To achieve this, the study aims to design a real-time AI framework for intraoperative decision support that enhances surgical precision, minimizes error rates, and enables scalable deployment in both advanced and resource-constrained operating environments.
Methods: The framework integrates multimodal data streams including surgical video, intraoperative imaging, biosensor feedback, and patient-specific electronic health records. Computer vision algorithms provide continuous anatomical recognition, while predictive models identify high-risk maneuvers and forecast potential complications. Retrospective surgical datasets were used to train the models, and simulated operating environments tested responsiveness under time-critical conditions. As a model case, laparoscopic cholecystectomy was analyzed, where AI-assisted anatomical recognition highlighted the cystic duct and artery to reduce misidentification risk. In neurosurgical simulations, predictive prompts forecasted vessel injury during high-complexity dissections. Comparative evaluation focused on intraoperative error reduction, operative efficiency, and adaptability across infrastructure levels.
Results: In simulation trials, the augmented intelligence system demonstrated significant reductions in error incidence, improved identification of critical anatomy, and enhanced workflow efficiency. In laparoscopic cholecystectomy simulations, AI prompts reduced misidentification events at Calot’s triangle, while in neurosurgical tasks they forecasted vascular injury risk with improved accuracy compared to baseline performance. Surgeons receiving AI-generated prompts achieved shorter operative times and greater precision in high-complexity tasks. The modular architecture allowed seamless adaptation: in high-resource settings it integrated with robotic platforms and advanced imaging, while in low-resource environments it functioned effectively with standard video capture and cloud-based support.
Conclusion: Augmented intelligence redefines the role of AI in surgery, positioning it as a supportive partner rather than a replacement. By combining broad adaptability with concrete specialty-specific applications, this real-time framework demonstrates potential to enhance precision, reduce errors, and bridge global disparities in surgical safety. It exemplifies the interdisciplinary innovation required to achieve equitable and safe surgery worldwide.