Rosota surgical AI trained on 849 demos runs robot appendectomy steps in animals

Korean surgical-data startup Rosota said its AI, trained on laparoscopic surgery data recorded with sensors attached to standard instruments, carried out the instrument movements of an appendectomy on a real robot in a preclinical animal study, Korea's Robot Newspaper reported on Oct 20 citing the company.

Rosota recorded video, instrument motion, instrument state and sensor readings on one timeline, used 849 surgical demonstrations collected on chinchilla rabbits for training, and evaluated the robot on separate animals not used for data collection, covering locating the target, aligning and approaching, grasping and retracting tissue and resection; recovery and complications were observed after surgery.

The company stressed that this is not fully autonomous surgery: surgeons chose each surgical step and the system executed the instrument motion for that step. Rosota described the result as a world first, a claim MEKAWIRE has not verified, and no paper or peer-reviewed publication was named.

It plans an AI surgical copilot and wider data collection across hospitals.