Researchers at Johns Hopkins and Stanford have trained an AI-driven surgical robot to autonomously perform portions of a gallbladder removal (cholecystectomy) on pig cadavers. The system, called Surgical Robot Transformer-Hierarchy (SRT-H), breaks procedures into subtasks, mimicking the hierarchy of human surgical teams. Trained on real surgical videos, the robot identified and clipped ducts and arteries without human voice prompts, showing early proof that autonomous surgery could one day be feasible — though real-world challenges like bleeding and organ movement remain.
Key Takeaways
- System used: Surgical Robot Transformer-Hierarchy (SRT-H), a hierarchical AI framework.
- Performance: Completed 17 surgical subtasks, including clipping ducts/arteries, on 8 pig cadavers.
- Learning: Initially guided by voice commands, but adapted to act independently.
- Limitations: Tested ex vivo — no bleeding, heartbeat, or breathing challenges present.
- Implications: Potential for emergency use when surgeons aren’t available, similar to autopilot in aviation.
- Next step: Scaling model with larger training datasets and testing in more complex, realistic scenarios.