Artificial Intelligence
AI-Enabled Robot Performs Gallbladder Surgery
AI breakthrough advances progress toward fully autonomous robotic surgery.
Posted July 14, 2025 Reviewed by Jessica Schrader
Key points
- Robot-assisted surgery is used today for a variety of purposes.
- In a recent study, researchers used an AI model to guide a surgical robot to perform gallbladder removal.
- The groundbreaking proof-of-concept was performed ex vivo, laying the foundation for further research in vivo.
New artificial intelligence (AI) research brings digital health care a step closer toward fully autonomous robotic surgery. A recent landmark study led by Johns Hopkins University created an AI model with architecture similar to ChatGPT to guide a surgical robot to perform gallbladder removal in a realistic ex vivo setting without help from human doctors.
“Our method achieves a 100% success rate across eight different ex vivo gallbladders, operating fully autonomously without human intervention,” wrote study co-authors Ji Woong Kim, Juo-Tung Chen, Pascal Hansen, Lucy X. Shi, Antony Goldenberg, Samuel Schmidgall, Paul Maria Scheikl, Anton Deguet, Brandon M. White, De Ru Tsai, Richard Jaepyeong Cha, Jeffrey Jopling, Chelsea Finn, and Axel Krieger.
The Rise of Surgery Robots
Surgical robotics is expected to reach a global market size of roughly USD $188.8 billion by 2032 with a 9% compound annual growth rate during 2023-2032, according to Allied Market Research. Examples of robotic surgery systems include the da Vinci by Intuitive Surgical, the Mako by Stryker, NAVIO by Smith & Nephew, Senhance by Asensus Surgical, Hugo RAS System by Medtronic, Maestro System by Moon Surgical, Ottav by Johnson & Johnson, ROSA Knee System by Zimmer Biomet, Galaxy System by Noah Medical, and others, to name a few.
Robot-assisted surgery started in the 20th century and is quickly on the rise. The term robot comes from robota, the Czech word for “forced labor,” and was introduced by the Czech playwright Karel Čapek (1890-1938) in R.U.R. that was published in 1920. Twenty years later in 1941, American writer Isaac Asimov (1920-1992) introduced the term robotics in Liar!, a science-fiction short story. A year later in a different sci-fi short story by Asimov titled Runaround, he introduced the now famed three laws of robotics. But it wasn’t until 40 years later that the concept of robotics left the realm of science fiction and made it into the surgical room. In 1985, the PUMA 200 (Programmable Universal Machine for Assembly 200) performed a neurosurgical biopsy on its first human patient.
Today, robot-assisted surgery is used for a variety of purposes such as urological procedures, prostate surgery, orthopedics, neurology, gynecology, hysterectomies, cardiac procedures, gastric bypass, sleeve gastrectomy, bariatric surgeries, thoracic surgeries, and pancreatic surgeries.
AI and Robotics
Artificial intelligence and robotics go hand-in-hand. In this new research, the team led by Johns Hopkins University researchers created an AI model with natural language processing capabilities with architecture similar to ChatGPT for autonomous step-level surgery called the Hierarchical Surgical Robot Transformer (SRT-H).
The top-down architecture of their AI model breaks down complicated surgical procedures into smaller tasks. Their AI model was trained on labeled videos of human surgeons performing surgery on over 30 ex vivo porcine gallbladders.
The researchers then tested their AI-enabled surgical robot on eight ex vivo porcine gallbladders that were not part of the training data and found it could successfully perform gallbladder removal surgery without human help. However, the robotic surgeon did take longer when compared to an expert human surgeon who was not part of the training data. Interestingly, their robotic surgeon was able to self-correct in real-time as needed, ranging from as low as one correction to 14, with an overall average of six corrections per surgery.
According to the study authors, their AI-enabled robotic surgeon performance was comparable to an expert human surgeon and was able to complete all cases with 100% success, self-correct errors in real-time as needed, identify tissue structures, and work at a reasonable speed. This groundbreaking proof-of-concept was performed ex vivo, laying the foundation for additional research in vivo and for other types of surgical procedures in the future.
Copyright © 2025 Cami Rosso. All rights reserved.
