Researchers at Johns Hopkins University have shown that an artificial intelligence-powered robot—trained on videos of previous surgeries—could learn how to perform the procedure itself, without help from a human operator.
According to the university, the autonomous system was able to adapt to the natural variations in anatomy between simulated patients, and react correctly to unplanned events in real-time—such as the introduction of blood-like dyes that obscured the surrounding tissue. It was also able to recover on its own from initially missed instrument placements.
In the controlled, ex vivo study, the robot was able to perform eight separate gallbladder removal surgeries with 100% accuracy—a complex procedure of 17 separate tasks, requiring it to identify and isolate certain ducts and arteries before cutting tissue.
“This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures,” said medical roboticist Axel Krieger, an associate professor in the university’s department of mechanical engineering.
Krieger previously helped develop the Smart Tissue Autonomous Robot, or STAR, which in 2022 performed a laparoscopic surgery on a live pig. However, that operation relied on specially marked tissue and a predetermined surgical plan, the university said.
This time, the team employed a ranked series of AI algorithms to guide the robot’s actions, including large-language models similar to ChatGPT that also allowed it to learn from and respond to voice commands from the team during the procedure.
“This is a critical distinction that brings us significantly closer to clinically viable autonomous surgical systems that can work in the messy, unpredictable reality of actual patient care,” Krieger said.
Described as a hierarchical framework, it includes a high-level, LLM-based approach to plan out tasks and generate instructions, while a low-level AI guides the robot’s movements. The researchers’ work was published this week in the journal Science Robotics.
“This work represents a major leap from prior efforts because it tackles some of the fundamental barriers to deploying autonomous surgical robots in the real world,” said lead author Ji Woong “Brian” Kim. “Our work shows that AI models can be made reliable enough for surgical autonomy—something that once felt far-off but is now demonstrably viable.”
The researchers provided their robot—built on older versions of Intuitive’s da Vinci hardware—with videos of Johns Hopkins surgeons performing gallbladder removals on pig cadavers, with captions describing the tasks. Though the robot took longer to complete each task than a human, its movements were smoother, and the university said the results were comparable to an expert hand.
“Just as surgical residents often master different parts of an operation at different rates, this work illustrates the promise of developing autonomous robotic systems in a similarly modular and progressive manner,” said the study’s co-author, Jeff Jopling, an assistant professor of surgery.