We're a step closer to entering an operating theater without any human life besides ours, following the world's first surgery performed by a robot responding and learning in real time. Its precision and skill matched that of experienced surgeons.
Researchers at Johns Hopkins University trained a robot on videos of operations, and then had it conduct a gallbladder removal on its own – with no mechanical help, just voice commands, like a theater team assisting the lead surgeon. Named SRT-H (Surgical Robot Transformer-Hierarchy), the robot absorbed its training and converted it to practice, with the ability to extract the gallbladder time and time again, and adjusting in real-time when needed.
"This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures," said medical roboticist Azwl Krieger. "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."
SRT-H is a breakthrough for the field, demonstrating mechanical precision but also the more challenging ability to adapt and understand in real time, adjusting when needed for success rather than following a linear path or script.
While this study details SRT-H's complete gallbladder removal (cholecystectomy) across eight different surgeries, it's worth noting these were performed on a realistic human-like model, but, understandably, not a human. However, the tissues used in the human-like models closely mimicked ours, and the robot breezed through the operation that required 17 tasks to be performed, each lasting minutes. It was able to identify specific ducts and arteries and grab them precisely, strategically place clips and sever parts with scissors.
"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, a former postdoctoral researcher at Johns Hopkins who's now with Stanford University. "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."
What's more, SRT-H houses the same machine learning architecture that drives ChatGPT, so the robot can respond to spoken commands from observing medical staff, correcting and learning in real time. It could also adapt its moves if the tissue it was working with appeared different to what was expected during the procedure. So even though its foundational training is based on imitation, it has the flexibility to learn and improve, much like a human surgeon.
Previously, the robot also achieved a 100% success rate when it performed the same gallbladder procedure on pig organs (not live pigs but likely cadaveric or synthetic models).
The field of surgical robotics is advancing as rapidly as AI itself; a year ago, the Johns Hopkins team had trained a robot to perform an important trio of surgical tasks: manipulating a needle, moving body tissue and administering sutures. It was able to do those tasks in mere seconds.
And, before that, Krieger's Smart Tissue Autonomous Robot, STAR, performed the first autonomous robot surgery on a live animal – a laparoscopic surgery on a pig – in 2022. But it was much more heavily guided by medical staff, who marked tissue and had the machine follow a blueprint with the real-time learning.
The new model "is like teaching a robot to navigate any road, in any condition, responding intelligently to whatever it encounters," Krieger said,
While the SRT-H took longer to perform its gallbladder surgeries than its human peers, the results were comparable to professionals that would normally be doing such operations. It's not yet ready to be unleashed on actual human patients, however, the team behind this advanced medical robot expect this to happen within the decade.
"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 study co-author Jeff Jopling, a surgeon at Johns Hopkins.
Next, SRT-H is expected to expand its expertise, being trained to conduct a variety of surgeries and making it even more autonomous, to perhaps one day have it performing successful surgeries with no supervision or external feedback.
"To me it really shows that it's possible to perform complex surgical procedures autonomously," Krieger said. "This is a proof of concept that it's possible and this imitation learning framework can automate such complex procedure with such a high degree of robustness."
The study was published in the journal Science Robotics
Source: Johns Hopkins University