While most sectors are rapidly getting autonomized, the healthcare industry is one of the sectors lagging behind and for obvious reasons — virtually no one would like to entrust their lives to a 'robot doctor', especially considering the fact that machine learning models still lack true intelligence.
Now a group of researchers is looking to bridge this gap with a bold new development. The team developed a new AI system that is paving the way for autonomous surgery. Their robot, named SRT-H, successfully completed a critical phase of a gallbladder removal, achieving a 100% success rate across multiple experiments. The team conducted 8 ex vivo tests using pig organ tissue.
The researchers trained the AI model on videos from surgeries performed by human surgeons, which they reinforced with text captions. This allows the AI not only to execute tasks but to understand the surgical procedure and respond to voice commands, much like a surgical resident learning from a mentor.
This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures — Axel Krieger, a medical roboticist at Johns Hopkins University.
To test the system’s resilience, the researchers introduced unexpected challenges. They added blood-like dyes to obscure the surgical field and altered the robot's starting position. In every instance, the SRT-H system successfully adapted to the changing conditions and corrected its course without human intervention.
While the robot currently operates more slowly than a human, it achieved results comparable to that of an expert surgeon. This new achievement could pave the way for autonomous robotic surgery in humans, a development that could revolutionize patient care by making elite surgical skills more consistent and accessible.
Source(s)
Science Robotics via The Guardian and New Atlas