Robo-surgeons, self-driving cars face similar legal, ethical headaches
Like transportation, medicine is becoming more automated, for better or worse
As drivers contemplate computer controlled cars, physicians get to ponder self-driven surgery tools.
In a research paper published on Wednesday, "Robot Autonomy for Surgery," UC San Diego assistant professor Michael Yip and PhD student Nikhil Das explore the growing role played by surgical robots and the issues raised by systems that are increasingly autonomous.
Just as self-driving cars have six levels of autonomy, ranging from human control to complete surrender, surgical robots can be grouped into similar categories.
The da Vinci Surgical System, which debuted in 2000, was the first FDA-approved robotic system for laparoscopic surgery, the paper explains, and represents a remotely controlled tool.
While it is capable of autonomous behavior, such use hasn't been approved.
Further along the directed-automated continuum, there's the Steady-Hand Eye robot, which was developed at Johns Hopkins University. Designed for a particularly delicate procedure – retinal microsurgery – it combines human control with robotic force correction to cancel a surgeon's inadvertent hand movements.
Under the category of supervised autonomy, Accuray's CyberKnife system, developed at Stanford University, creates and carries out a surgical plan to deliver radiation therapy to tumors based on a preoperative blueprint.
"The human surgeon adjusts the automatically generated plan prior to execution and ensures the system performs the task safely during the procedure," the paper says.
At the end of the spectrum, the paper's authors consider fully autonomous surgery far-fetched from technological point of view at present, to say nothing of the social, legal or ethical issues involved. The situation is similar for vehicles capable of SAE Level 5 automation: the technology isn't there yet.
But the authors observe that complex operations can often be broken down into discrete tasks, such as suturing or resection, some of which may be automated. And doing so, they argue, offers safety advantages similar to those cited by advocates for autonomous vehicles.
"With many of these surgeons overworked to high levels of fatigue, they are less attentive and more prone to human error," the paper states. "Autonomous surgery provides a consistency and quality to a treatment unaffected by these issues."
Beyond being untiring, robot surgeons are more precise, can work in hazardous environments, and can synthesize multiple sensory inputs, the paper says. However, they're not without downsides: bedside manner aside, they lack judgement, their sense of touch isn't as refined as a human surgeon, they're expensive, and they lack the ability to improvise.
A mostly automated system that could see use one day is DARPA's Trauma Pod, which aspires to be a way to delivery battlefield telesurgery.
Robot surgery devices are benefiting from the rapid advancement in machine learning and related fields like computer vision. The technology has not yet reached the point where robo-docs can learn operations through observation, but the paper says such work is underway.
"Some autonomous systems have successfully performed isolated surgical tasks based on a human-provided exemplary dataset," the paper says. "Trajectory smoothing of human-provided motion examples enabled faster and smoother trajectory executions on suture knot-tying tasks on an in-house laparoscopic workstation compared to a human."
Moving toward greater autonomy will raise legal and ethical obstacles, the paper predicts, similar to those that accompany automated vehicles.
"While current practices of autonomy in surgery utilize robotics more as an intelligent tool than as an entirely independent agent, increasing the level of autonomy of the surgical system raises the question about who is in charge of the surgical operation," the paper says. "In the case of surgical errors, the human surgeon who used or authorized the use of the robot, the hospital, the robot designer, and the insurer are included as potential culpable entities."
Who pays when the euthanasia bot malfunctions? In which jurisdiction can you sue when your botched telesurgery involved people on three different continents? What happens when region-locked abortion machines get hacked for unauthorized usage or sabotaged?
To forestall such scenarios, the paper's authors advocate supervised autonomy over full autonomy, a position most doctors presumably support, for the health of their patients if not for their continued employment. ®