Diane Joy Galos

Surgery has always required years of study and a steady hand for a human, but robots may soon be able to pick it up faster, thanks to advanced artificial intelligence (AI) technology.



Researchers at Johns Hopkins University (JHU) and Stanford University have successfully trained a robot to perform surgical tasks by simply showing it videos of those procedures.

Using a da Vinci Surgical System, typically remote-controlled by a surgeon, the team taught the robot to handle delicate tasks like dissection and suturing. 

The system, which costs over $2 million, gives surgeons precise control and a closer view of the patient.

By using imitation learning, the researchers taught the robot three key surgical tasks: manipulating a needle, lifting body tissue, and suturing. 

The robot not only performed them with human-level skill, but it also learned to fix its own mistakes. “If it drops the needle, it will automatically pick it up and continue,” Axel Krieger, an assistant professor at JHU, said.

To make this happen, the researchers used the same machine learning technology that powers chatbots like ChatGPT, but instead of text, it outputs kinematics to guide the robot’s movements. 

Hundreds of surgical videos, recorded with wrist cameras on da Vinci robots, were used to train the system.

The team believes this method could train a robot to perform any surgical procedure much faster than traditional methods, which require hand-coding each step.

Krieger sees this breakthrough as a step closer to making automated surgery a reality. "What is new here is we only have to collect imitation learning of different procedures, and we can train a robot to learn it in a couple of days," he said. 

"It allows us to accelerate to the goal of autonomy while reducing medical errors and achieving more accurate surgery."

This method could lead to significant advancements in robot-assisted surgery. 

While some devices, like Corindus's CorPath, are already used in complex surgeries, they only assist with specific steps.

Krieger also emphasized that hand-coding each step of a surgery is time-consuming. "Someone might spend a decade trying to model suturing," he said. 

"And that's suturing for just one type of surgery."

In 2022, Krieger's team developed a robot called STAR that autonomously sutured a pig's intestine. 

Now, they’re working on using their imitation learning method to train robots for full surgeries.

It may take years before robots fully replace surgeons, but with advancements like this, the future of more accessible surgeries is closer than ever.