Would you trust a robot surgeon powered by algorithms?
While robots are increasingly common, full automation remains a challenge
Article by Hasan Chowdhury | The Telegraph
Robotic surgery has come into greater focus as the pandemic has introduced wide-scale disruptions
In 2016, a study published in Science Translational Medicine, a peer-reviewed journal, contained a radical proposal for surgery that involved sutures, a robot and pig bowels. The researchers behind the study detailed how the robot, a semi-autonomous machine named STAR, was able to stitch together parts of the pig’s intestines with little supervision.
“The operating room may someday be run by robots, with surgeons overseeing their moves,” the paper’s abstract noted, hinting that the way forward for surgery in the future could depend on “removing the surgeon’s hands”.
The critical element of these robots for the researchers was that it had a degree of autonomy – able to serve as a valuable team member by using machine learning and artificial intelligence to learn how to perform suturing tasks based on information it processed through an algorithm.
Four years on, and robotic surgery has come into greater focus as the pandemic has introduced wide-scale disruptions. A study led by researchers at the University of Birmingham in May found 28.4m surgeries could be cancelled or postponed in 2020 globally as Covid-19 hit staff availability and resources.
There is little doubt that in theory, a robot is better suited to the most delicate surgical procedures. A robot can enter body parts that human fingers cannot touch, and more delicately cut and snip without the risk of hand tremors.
Earlier this year, CMR Surgical, a Cambridge robotics start-up, debuted its Versius robot on the NHS. This robot can perform intricate keyhole surgeries. This robot mimics a human's hand and wrist movements, but is still controlled remotely by a surgeon from a computer or using a control stick.
Today, a fully intelligent, robotic surgeon remains a piece of blue-sky thinking. But it is clear robotics and automation are being used for tasks that once required the steady hand of a surgeon with years of training.
Prokar Dasgupta, a consultant urological surgeon at Guy's and St Thomas' NHS Foundation Trust in London, thinks AI robots will have a greater place in the operating theatre, though sees them first being eased into a role in repetitive tasks like stitching.
“Stitching one part to another has the potential to be automated,” he says. “Actually the robot in those situations might very well perform better than a human being. It's more repetitive, more precise and doesn't get tired unlike a human.”
But could AI robots undertake more complex procedures on their own? Dasgupta doesn’t think this could happen just yet, but he is unwilling to rule out the possibility of robots working in tandem with surgeons in a more involved way depending on certain advances.
For one, current machine learning algorithms, which usually learn and develop by observing patterns in data – in this case, the sequence of decision-making in key stages of surgery – will need to make monumental advances to achieve more complex tasks.
To illustrate, Dasgupta explains the difference between two tasks undertaken in prostate cancer surgery, in which decision-making is “individual to every patient”.
In addition to stitching done around the bladder, a relatively simple task, a surgeon will also remove the prostate gland while working to protect bundles of nerves that run all the way up to the brain. One slip has the potential to be fatal. In terms of complexity, the tasks are worlds apart.
“Current machine learning and AI systems have a long way to go before they can be trained to make that human judgement of where to cut and where not to cut,” he says. “Can they be taught to do that? Yes, but that's not ready for prime time yet.”
To date, robots used in the NHS have been controlled by surgeons sitting at a computer console who use toggles and buttons to control the movements of a robot, usually equipped with arms, in a precise way to perform laparoscopy, or what's known as minimal invasive surgery.
The use of these particular robots is about precision, allowing movements to be made within a degree of millimetres rather than centimetres, while reducing issues that even the best of surgeons risk facing such as hand tremors.
Surgical robots have been in clinical operation in the UK since 2001, with a robot known as Da Vinci being used by doctors in more than 70 hospitals. The robot is now so precise it can suture a wound cut into the skin of a grape.
But critically, the robots lack machine learning. They are instead as sophisticated tools by surgeons.
A number of other firms are trying to muscle in on the market. Google and Johnson & Johnson are working together on a joint venture, Verb, to develop a "next generation" surgical robot. While Digital Surgery, a British start-up, had been exploring the use of AI in surgical training apps.
Mark Slack, a consultant gynaecologist and chief medical officer at CMR Surgical, sees it as highly unlikely that an AI robot could emerge that leads procedures. He says:
It would have to interpret thousands of minute anatomical differences, the only way you can do that is data inputs. Somebody would have to sit and put in videos of thousands and thousands of laparoscopies, I can't even see it in a thousand years.
Even if autonomous robots were to emerge that could operate alone or alongside a doctor on an equal footing, there are ethical questions to be had as to whether or not they should be allowed to conduct surgeries, or if patients will even want them to.
“To tell a patient that I will press a button and everything will be done by a robot while I have coffee may not be acceptable to many,” Dasgupta says. Surgeons have a “gut feeling” that guides them in high-pressure situations. But that will not stop people from exploring applications for AI robots.
In 2018, the Commission on the Future of Surgery delivered by the Royal College of Surgeons, showed technology would not be shunned if it meant better service for patients, highlighting AI as one of four key areas that could transform the field in the next two decades.
“In this future preconceptions must be left outside the operating theatre; what matters is who has the right skills to treat a patient, not whose ‘right’ it is to provide that treatment,” Professor Stephen Powis, medical director of NHS England said in the report.
According to Richard Kerr, a neurosurgeon and chair of the Commission on the Future of Surgery, treatments in the future are going to be “team-based in their delivery”, in which different individuals play a role in the success of an operation. How artificial intelligence factors into this team approach to surgery, though, is uncertain.
“What I don't see, at least right now, is that leap where you introduce the robot as one of the competencies in your team to deliver in isolation aspects of care,” he says.
Despite all the advances in technology, Kerr notes, the crucial thing is the relationship between a surgeon and a patient. “That relationship is absolutely fundamental,” he says. While robots may be increasingly common tools in a surgeon's box, they won't be handing them the knife just yet.