Robotic Surgeons Are Showing Hints Of One Day Outperforming Humans

When it comes to fields that are considered the most complex of human endeavours, the most typically cited are those of rocket science and brain surgery. Indeed, to become a surgeon is to qualify in a complex, ever-changing, and high-performance field, with a pay scale and respect to match.

The tools of surgery have changed over time, with robotic assistants becoming commonplace in recent decades. Now the latest robots are starting to outperform human surgeons in some ways. Let’s look at how that’s been achieved, and what it means for the future of medicine.

Robots With The Moves

Robotic surgery has been in development since the 1980s, when basic robots were used to help out with simple surgical tasks. Robots offer plenty of advantages that make them ideal for surgical purposes. They can hold tools with excellent positional accuracy, don’t suffer from fatigue, and their manipulators can work on much smaller scales then human hands. This is of particular benefit when doing keyhole surgeries, where tiny robotic implements can be inserted into a patient along with a camera. This allows surgeons to operate while making minimally-sized incisions in a patient, improving recovery outcomes and reducing the likelihood of any infections.

Moving pegs on a board is a typical way human surgeons practice with robotic surgical manipulators. However, one recent study has shown that the robot can achieve the task faster under autonomous control than with a human making the moves. Credit: Research paper, IEEE Transactions on Automation Science and Engineering

Traditionally, robotic surgical apparatuses, like the well-known Da Vinci robot, are used under direct control from human surgeons. They have the benefits of filtering out tiny tremors common in human movement, and also offer control of surgical instruments with more range of motion than is possible with the human hands alone. Fundamentally, though, the robot is under full remote control, with a human directing the exact movements of the tools that operate on the patient.

However, recent research has explored adding autonomous control to robotic surgery systems. One study explored using a surgical robot to achieve a simple task, where triangular pegs were transferred from one side of a pegboard to another. The pegboard is a common training tool used by human surgeons, which teaches fine motor control, precision, and speed. The study explored three variations on the peg movement task, a one-handed method, a two-handed method, and a method involving transferring pegs from one hand to the other before placing them back on the board.

A human surgeon controlling the Da Vinci robot system was able to achieve the one-handed task faster than the fully automated system created by the research team. However, the two-handed tasks were more quickly achieved under fully autonomous control system.

Of course, moving pegs around a board is simple compared to surgery on humans. Tracking rigid objects of known size and shape is relatively simple given today’s machine vision systems. Still, it shows that the processing power and control systems are now available to best a human at such a task, and that it’s a fruitful area for experimentation and research.

If you’re familiar with the DaVinci robot surgery system, it may be from the classic grape surgery video that went viral online. 

However, other teams are going beyond simple standardized training tasks. A research group from John Hopkins University trialled autonomous robot surgery on live pigs. This involved rejoining severed intestines, in what is called an intestinal anastomosis procedure. It’s an incredibly challenging process, requiring surgeons to carefully suture two sections of intestine back together without damaging the incredibly soft tissue.

Early tests involved performing the surgery outside the body, with the robotic system able to achieve the procedure with the assistance of human doctors. Further upgrades enabled the surgery to be achieved via keyhole methods inside the pig’s body. The Smart Tissue Autonomous Robot, or STAR, was able to complete two-thirds of the required stitches by itself, with the remaining completed with guidance from the the human surgical team. It’s an impressive feat for a robot to achieve, and it suggests that future developments could allow robots to work alone in future decades.

Right now, it’s our superior eyes and brains that are keeping human surgeons ahead of the game. The fact is that many surgeries are already achieved with robotic instruments. Once the robots get superior vision systems and intelligence, it may be that they outcompete human surgeons entirely. As always, time will tell.


31 thoughts on “Robotic Surgeons Are Showing Hints Of One Day Outperforming Humans

  1. I’m alive today because of the daVinci robot and brilliant surgeons at the world’s highest rated surgical center. It’s way, way beyond the grape example for minimally invasive surgery. They have cutters, pliers, cauterizers, staplers, with everything operating through 3 or 4 10mm holes between the ribs and in the abdomen with the only surgeons’ visibility being from remotely inserted cameras. The more you know about this the more incredible it is. Truly meeting the Clarke standard of sufficiently developed technology indistinguishable from magic.

      1. Standard U.S.A. ACA compliant employer’s health insurance in a big city university surgical center, although one of the highest rated ones. Other than annual deductible and co-pays, nothing extraordinary. I did agree to permit student observation of the procedure and recovery.

        1. I can’t tell if you meant that as a joke, but many hospitals are already using robots for sterilizing rooms. They rove from room to room, LIDAR scan to ensure the room is empty and has been straightened by a human, and then sweep the room with UV-C for a few minutes. Apparently they do a better job than humans at preventing staph outbreaks.

      1. Poorly paid personnel, usually… I did my civil service in a hospital and I was kind of shocked to see that there’s specially “trained” but poorly paid people who do stuff like sterilizing surfaces and even holding abdominal hooks for hours..

        1. In our society, broadly speaking, the more fundamentally important a role is, the less the people doing it are paid.

          Just look at the pandemic and who the “essential workers” were, who were forced to come to work to keep society running, and who were allowed to stay home.

  2. It’s about time they started turning the machines into surgeons instead of mere tools for surgeons.

    I look forward to the day when humans are merely a supervisory or directive part of surgeries because machines outperform humans in so many ways. In addition to radically reducing the number of staff needed, being able to perform and coordinate complex simultaneous surgeries will help save rapidly deteriorating patient severe to extreme injuries that currently have a negligible or even non-existent chance of survival.

    I’m really hoping this technology gets pushed to the point where there is something akin to the “auto-doc” from various sci-fi movies that can perform scans to identify issues and then perform corrective surgeries, all with minimal human guidance. This would save many many lives.

    1. Why? How do you know they will outperform humans? Article of faith?
      I look forward to having to solve captchas which train AIs to recognize tumors and clots instead of fire hydrants and traffic lights. Self-driving cars which outperform humans are still vaporware.
      It will probably be a bit of a liability nightmare once one of these has a glitch and returns to its base pose while the tools are still inside a body or something.

    2. Doing a DaVinci case, at least now, approximately doubles the amount of OR staff. You need the regular staff then the proper to tend to the robot. Plus a medical
      student to hold a thing for 6 hours.

    3. “I look forward to the day when humans are merely a supervisory or directive part of surgeries because machines outperform humans in so many ways.”

      Yes, it will be so much easier and quicker for tyrannical regimes to perform targeted lobotomies and sterilizations on those who question their rule! No need to lock them in prisons for years at a time, or even shoot them, when they can be made slaves for the spice mines efficiently and quickly!
      [Tongue in cheek]

  3. Couple of points. First, operating faster is not necessarily “better,” I would say technical accuracy is more important. That said, and this is something most lay public people and even non surgical medical people don’t understand. The surgery itself is obviously important but it is very, very rare for technical errors to be made along the lines of “cut the wrong thing.” I mean, they do happen but they are very rare. For major operations the peri operative and post op medical management is a lot more important. That is also where the vast majority of malpractice lawsuits originate. Surgeons operate like one or two days a week the rest of the time is clinic and post op patient management.
    The DaVincini is rad though. I got to demo one of the first versions before stereo vision and haptic feedback and tremor filtering and even at that it was incredible.
    Finally, an intestinal anastomosis is stock and trade of any general surgeon. I’m not saying it isn’t hard but that’s like saying an instrument landing is some miracle for an average airline pilot. Plus for gut work, mostly they just use a big stapler they aren’t sitting there sewing forever (usually).
    Oh and one more thing. Train robots to do mechanical tasks all you want. That’s “easy.” If and when to do the task you need a human doctor for the foreseeable future.
    Cool article. Thanks.

    1. “Plus for gut work, mostly they just use a big stapler they aren’t sitting there sewing forever (usually).”
      Stapler is incredibly crude though, and using better suturing ways would be badly needed. One of them is based on special glues, but staples do not create the nicest scars.

      The aim is to get better, not to say it’s all good enough and doesn’t matter either way, since what they do in practice is crude anyway.

      Operations do and can be improved. Just because post-op care needs more attention doesn’t negate this fact. The more precise and less invasive an operation is the better. Good “enough” is not how progress is made.

      1. The timing signal on my pumping bus was out of kilter, so they tried to reset me a few times (8) with the paddles and glitch it back to synchronous, but that didn’t work and they put me in for rework with shaving the crystal with a laser and improving the isolation between conductors.

  4. Craig says, “Doing a DaVinci case, at least now, approximately doubles the amount of OR staff.”

    Not sure where your information comes from, but as an OR nurse who works with the da Vinici surgical robot on a daily basis, we do not have any more personnel in the room than with a standard laparoscopic procedure.

    My background is electronics in automation and controls. Decisional intelligence has never been a strength in the field. As with most computer controlled environments, programmed intelligence never trumps human intelligence. It can just process information faster.

    Bottom line, right tool for the right job. I cannot see automated robotic surgery in any near future on humans.

    1. I suspect automated surgery might be rather closer than that, as I can see it being in the near future – but very very limited in scope. The relatively simple tasks of putting screws and plates on the easier to work on bones or just basic stitching folks with surface flesh wounds back together first perhaps. On which it probably spends many times longer than the human would just going through the problem analysis and treatment planning stages that the computers are not as good at.

      Carefully triaged as they come in the door so the stuff where it doesn’t matter if a robot is slower for studying and processing every move its going to make in excruciating detail first with the one surgeon supervising, maybe even actively hitting the ‘good plan, execute’ button for many patients.

      It won’t be real world day to day mass deployment any time soon I don’t think – as the surgical machines are just too damn expensive. Or make much sense to go fully automatic in places without huge populations, and so huge numbers of trivial surgeries. But that central hospital in a big metropolis will probably be able to gain more staff time on the complex stuff and so serve more patient if the simple stuff that comes in the door can be dealt with automatically. So its going to be worked towards.

    2. My information comes from being an academic physician and PhD scientist for a decade. Maybe in a svelte private group nothing changes but a typical case of a surgeon one nurse and an anesthesiologist is not sufficient for a robotic case. It also adds like an hour to set up. Again, my personal experience across 3 large institutions. We can all be correct.

      1. Yes we can all be correct!
        In my 14 years in the OR, robotic teams have always been anesthesia, surgeon, first assist, scrub and circulator. In a laparoscopic case, there might not be an assist. Only difference.
        As for turnovers, we average the same 28 minutes that all the other cases do. Even for the most complicated case setups.
        My experience is only two hospitals, but one has a surgeon who consistently performs the most cases in the country yearly. His partner comes a close second now and then.

    3. Semi automatic would be very good already. You can give the general idea and instructions as a surgon and the robot will do the detailed precision tasks.

      I am not sure that fully automated is really desirable, since the robot will not be able to communicate with a patient like a person can.

  5. As a physician/surgeon (with a prior 13 years in EE and CS), I’ve used laparoscopy for 25 years and am in the process of getting DaVinci certified now. When comparing autonomous robotic driving to autonomous robotic surgery, I think the latter is a couple orders of magnitude greater complexity task than the former. Still, progress is progress.

    1. The requirements are wildly different for sure. Surgery requires dexterity and good vision. But you can also control the environment better and can create a list of all likely causes.

      So you could focus on control engineering and computer vision and other highly precise mechanical problems. It’s more of a close world problem than the open world problem of driving.

      I actually believe that you have more chances to make a really well designed surgical robot that gets better and better with time and incrememental well engineered improvements, because the human body can be well described formally, as well as its physical properties.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.