Autonomous robotic surgery has taken a giant leap forward with the successful procedure performed by an international team of engineers and surgeons: for the first time, a robot has independently performed a gallbladder removal in a realistic simulated environment. This project, led by Johns Hopkins University in collaboration with Stanford University, highlights how artificial intelligence can transform surgical medicine as we know it.
Far removed from traditional assisted surgery, the new SRT-H (Hierarchical Surgical Robot Transformer) system analyzes, decides, and executes every part of the procedure with precision comparable to that of an experienced surgeon . The machine learns from video recordings of real surgeries and, during the procedure, is able to respond to verbal commands or adjust its behavior to unexpected situations . All of this is achieved without losing its ability to self-improve during the intervention.
How the autonomous surgical robot learns and responds
The development of SRT-H relies on machine learning algorithms similar to those used by large language models, such as ChatGPT. The robot was trained using hours of video footage of gallbladder surgeries performed by experts . Each recording included step-by-step explanations, allowing SRT-H to identify ducts and arteries and know when to grasp, clip, or cut delicate tissue.
One of the key innovations is its ability to interact with the medical team using voice commands . SRT-H can execute simple orders like “grasp the neck of the gallbladder” or correct movements if a specialist tells it to “move the left arm a little.” This active feedback loop allows the robot to adapt and correct in real time, rather than simply following rigid instructions. Thus, the machine is not just a tool, but a true surgical assistant with its own judgment.
Intervention results and challenges for autonomous surgery
In the tests performed, the robot successfully completed a 17-step surgical sequence , from identifying anatomical structures to precisely separating the gallbladder. Accuracy reached 100% in artificial models, even when researchers complicated the operation with unexpected tissue changes or visual alterations similar to those encountered in real emergencies.
Although the system still takes longer than a human surgeon , experts emphasize that the quality and safety of the procedure are entirely comparable to that of a senior professional. The SRT-H also demonstrated a remarkable ability to adapt to unforeseen circumstances, outperforming conventional surgical robots, which require a controlled environment and a rigid plan.
