Industries / Health / Healthcare
Surgery
Operating on the body to remove, repair or replace a part, and the work that makes an operation safe: anesthesia, clean technique and checks.
7
Milestones achieved, from 1846 to 2018.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1846
Ether in the operating room
On 16 October 1846, in the operating theater of Massachusetts General Hospital in Boston, the dentist William T. G. Morton gave ether to a patient, Edward Gilbert Abbott, while the surgeon John Collins Warren operated on a mass under his jaw. Henry Bigelow published four such operations on 18 November 1846, and by the end of December ether was in use in Britain and France. A 2022 study of the hospital's records found that Abbott moved and cried out late in the operation, so the anesthesia was incomplete, and that later operations at the hospital made the stronger case.
LessonThe best-known demonstration was only a partial success, and the operations after it carried the argument.
Credit
William T. G. Morton, John Collins Warren
1867
Killing germs in wounds
Joseph Lister took up Louis Pasteur's work on microbes in the air, reasoned that they caused wound infection, and treated wounds and dressings with carbolic acid (phenol) to kill them. He published six articles in The Lancet from March to July 1867 and read "On the Antiseptic Principle in the Practice of Surgery" to the British Medical Association in Dublin on 9 August 1867. Of 11 patients with compound fractures (broken bones through the skin) he treated this way from 1865 to 1867, 9 stayed free of infection, by the count of a 2022 review.
LessonThe method failed for surgeons who used it carelessly, because it worked only when every step was followed.
Credit
Joseph Lister
1985
Gallbladder out through keyholes
On 12 September 1985 Erich Mühe of Böblingen, Germany, removed a gallbladder through a laparoscope he had designed. A laparoscope is a viewing tube passed into the belly through a small cut. The German Surgical Society rejected his work in 1986, gave him an award for it in 1992, and in 1999 the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) recognized him for performing the first laparoscopic gallbladder removal. Philippe Mouret performed the first in France, in Lyon, in 1987.
LessonA new operation can be turned away by its own specialty before it is accepted.
Credit
Erich Mühe; Philippe Mouret
2000
A robot holds the instruments
On 11 July 2000 the US Food and Drug Administration (FDA) cleared Intuitive Surgical's Endoscopic Instrument Control System, the da Vinci, as a computer-controlled surgical system. The surgeon sits at a console and moves instruments that robot arms hold inside the patient through small cuts.
Credit
Intuitive Surgical
2001
Surgery across an ocean
In September 2001 Jacques Marescaux and Michel Gagner, at a console at Mount Sinai Medical Center in New York, removed the gallbladder of a patient at Strasbourg Civil Hospital in France, about 4,300 miles (6,900 km) away, with a robot called Zeus. Signals took about 155 milliseconds to cross the Atlantic over fiber-optic lines, by The Scientist's report.
LessonRemote surgery worked in 2001 but stayed rare afterwards; a 2025 review says "technological and economic challenges" limited it.
Credit
Jacques Marescaux, Michel Gagner and colleagues
2009
A checklist for every operation
In eight hospitals in eight cities, from Manila to Seattle, deaths after surgery fell from 1.5% to 0.8%, and complications in hospital from 11.0% to 7.0%, after teams began using the World Health Organization's Surgical Safety Checklist (3,733 patients before, 3,955 after, from October 2007 to September 2008). When the province of Ontario, Canada, later urged all its hospitals to adopt checklists, a study of 101 hospitals and about 216,000 operations found no significant change in deaths (0.71% before, 0.65% after) or complications (3.86% and 3.82%).
LessonA checklist adopted by policy did not by itself cut deaths, and the checklist's authors later wrote that a team's safety culture is essential to its effect.
Credit
A. B. Haynes and the Safe Surgery Saves Lives Study Group; D. R. Urbach and colleagues for the Ontario study
2018
Smaller cuts, worse survival
A trial of 631 patients with early cervical cancer compared removal of the uterus through small cuts (minimally invasive surgery) with open surgery. At 4.5 years, 86.0% of the small-cut group were alive without the disease, against 96.5% after open surgery, and at 3 years fewer were alive (93.8% against 99.0%).
LessonA gentler operation has to be tested against the old one on survival before it replaces it.
Credit
P. T. Ramirez and colleagues, in the New England Journal of Medicine
Watch for
Gallbladder steps without a surgeon
A research team trained a robot by imitation, from recorded demonstrations, to carry out the long run of steps in gallbladder removal. A planner gives the robot instructions in words and corrects its mistakes as it goes. The team reports success on all eight gallbladders it tried, on tissue outside a living body (ex vivo), working "fully autonomously without human intervention".
Achieved whenanother group repeats the result, or the system completes the same steps in live animals under outside review.
Claimed by
J. W. B. Kim, A. Krieger and colleagues, in Science Robotics, July 2025
Remote surgery reaches a trial
Vipul Patel, a surgeon in Florida, operated with a robot on a man with prostate cancer in Luanda, Angola, about 7,000 miles away, over fiber-optic lines. ABC News reported that the patient was the first in a human trial of transcontinental robotic surgery approved by the FDA, and that a team in Angola stood ready to finish if the link failed. In 2024 a team including Patel had operated on live pigs between Orlando and Shanghai, about 13,000 km apart, with a median signal delay of 139 milliseconds.
Achieved whenthe trial's results appear in a peer-reviewed journal, or a regulator clears a robot for surgery from a distant site.
Claimed by
Vipul Patel and team, June 2025
Open
01
A ledger of robot faults
A public table of every report about a surgical robot in MAUDE, the FDA's database of reports about medical device problems, from 2014 to the present, for every maker. Each row gives the date, maker and model, kind of operation, whether a patient was hurt or died, what failed (instrument, arm, camera, software or user), and whether the operation was converted to open surgery or abandoned. Each label quotes the words in the report that support it. The table refreshes itself every week.
The briefCredit · unclaimed02
Stitch a practice pad alone
A low-cost, open-source robot arm that picks up a curved needle, passes it through a silicone suture practice pad at marked points, and ties the knot, all by itself, after learning from recorded demonstrations. No tissue and no patients.
The briefCredit · unclaimed
Open 01 · the brief
A ledger of robot faults
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of every report about a surgical robot in MAUDE, the FDA's database of reports about medical device problems, from 2014 to the present, for every maker. Each row gives the date, maker and model, kind of operation, whether a patient was hurt or died, what failed (instrument, arm, camera, software or user), and whether the operation was converted to open surgery or abandoned. Each label quotes the words in the report that support it. The table refreshes itself every week.
Why it is still open
The last study that covered every surgical robot in MAUDE, by H. Alemzadeh and colleagues, stopped at 2013. It found 10,624 reports, including 144 deaths, 1,391 injuries and 8,061 malfunctions, and said reported rates are likely too low. The later studies found each cover one maker or one specialty, and none publishes its labeled reports.
Built with what exists
the openFDA device adverse event API, starting from product code NAY, the FDA's code for computer-controlled surgical systems (the code on the da Vinci's 2000 clearance); a language model such as Claude or Gemini to label each report and quote its evidence, with a person checking a sample; Python and DuckDB for the table. The advance that makes this a small-team job is language models that read thousands of free-text reports and fill fixed fields (see the Note "Seven more flagged" and the Note "A thousand questions, one pass").
First step
Download one year of reports for product code NAY from openFDA, label 100 by hand, and compare a model's labels with yours.
Done looks like
All reports since 2014 for at least five robot makers are labeled. A second person checks 200 random rows and agrees with at least 95% of the labels. The same pipeline, run on 2000 to 2013, comes within 5% of Alemzadeh's counts of deaths, injuries and malfunctions. The table and code are public and refresh weekly for three months.
Open 02 · the brief
Stitch a practice pad alone
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A low-cost, open-source robot arm that picks up a curved needle, passes it through a silicone suture practice pad at marked points, and ties the knot, all by itself, after learning from recorded demonstrations. No tissue and no patients.
Why it is still open
The authors of SutureBot, a 2025 benchmark of 1,890 suturing demonstrations on the da Vinci Research Kit, write that "a fully autonomous suturing pipeline has yet to be demonstrated on physical hardware". The one study found on an open-source arm, from May 2026, has the robot assist a surgeon by holding the thread after each stitch, with 50% to 75% task success under ideal conditions.
Built with what exists
Hugging Face's LeRobot library and its low-cost SO-101 arm, with a 3D-printed needle-holder tip; open robot policies such as ACT, SmolVLA, π0 or NVIDIA's GR00T models; a silicone suture practice pad. The advance that makes this possible is robot models that learn a task from a few dozen demonstrations (see the Note "Five demonstrations per task", the Note "No download" and the Note "Open robot model weights").
First step
Mark 30 stitch sites on a practice pad and record 50 stitches done by hand through the arm's leader arm.
Done looks like
20 attempts in a row, filmed and published, each one stitch from needle pickup to finished knot with no human touch. The published tally counts how many needles pass within 2 mm of both marks and how many knots hold when the two tails are pulled. A surgeon who did not build the robot scores the videos.