This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American
Those keeping abreast of the latest medical developments may be aware of the buzz surrounding applications of artificial intelligence (AI) to medical tasks. To date, these have mainly involved application of computer algorithms to clinical data such as x-rays, images or text-based medical records, to diagnose disease. The sensationalism has largely arisen due to the fact that in some instances, these algorithms have met or exceeded capabilities of a specialist physician for particular diagnostic tasks.
With these early accomplishments, a question arises as to how the introduction of clinically viable AI may affect the role of human physicians in the future. Being at a primitive stage and lacking widespread real-world application, the topic remains speculative at present. Much of the debate, however, concerns the effect of AI upon non-interventionalists,* whose primary role is to diagnose and treat diseases non-invasively. An interventionalist (i.e., a surgeon) may therefore wonder how the “AI revolution” may affect him or her; after all, an algorithm cannot perform a heart bypass or remove a brain tumor.
In 2016, the Smart Tissue Anastomosis Robot (STAR), an autonomous surgical robot, underwent experimental trials in animals. The robot, which utilized “smart sensing” apparatus including cameras and mechanical sensors, along with AI-control algorithms, outperformed human surgeons at certain tasks, including joining intestine in a living animal without direct human intervention.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
This was not the first time that AI-controlled robots have experimentally outperformed humans at single surgical tasks such as knot-tying. The closest things to commercially available “autonomous” surgical robots are devices that use external radiation beams to treat cancers (not in direct contact with human tissue) or those offering reduced human input for tasks involving rigid, fixed tissues (i.e., bones) for joint replacements.
A mammal’s abdomen is soft, deformable, delicate and contains blood vessels and organs at great risk of damage during surgery. STAR had to perform multiple real-time tasks simultaneously, while minimizing risk of collateral damage: “seeing” the environment which it was working in, “sensing” the features of the tissue upon which it was operating and “reacting” to environmental changes as they occurred, effectively mimicking human surgeons’ “judgment” in addition to their physical skill.
In their Science Translational Medicine paper, the authors noted: “The intent of this demonstration of feasibility in soft tissue surgery was not to replace surgeons but to expand human capacity and capability through enhanced vision, dexterity, and complementary machine intelligence for improved surgical outcomes, safety, and patient access.”
Surgical robots have existed for over 30 years and have become widely used by certain specialties because of their technical benefits, which augment human capabilities. These include fatigue resistance, increased range of motion and resistance to shaking. Commercial surgical robot designs utilize a “master-and-slave” arrangement; a human surgeon (master) controls the robot (slave), situated near the patient and a few feet away from the operator, in real time. The technical benefits of using surgical robots have translated into improved outcomes and reduced complications for certain procedures such as prostate surgery.
At present, the costs of purchasing ($1 million to $2.5 million) and maintaining robotic surgical devices are cited as a barrier to their widespread use. Robotic surgical procedures are often more expensive than traditional ones, given the specialized equipment required for the robot. Institutions purchasing these devices must maintain a high volume of cases to recover outlays. Nevertheless, because of the technical advantages of such systems, which include a lower complication rate and better removal of diseased tissue, studies have demonstrated a reduction in overall costs to health care systems for robotic surgery for certain diseases such as prostate cancer.
It is therefore reasonable to predict an evolving symbiosis between the benefits already demonstrated by surgical robots, with the nascent advantages of medical AI. Given further development, we may soon see autonomous surgical robots with capacity to perform complex soft-tissue surgical procedures faster, more accurately and with fewer complications than human surgeons.
If these devices are tested by trial in humans, and the advantages are confirmed, it is also not unreasonable to envision a scenario where continuous technological refinement driven by clinical demand leads to drastic cost reductions for purchasing autonomous surgical devices.
A phenomenon where technological advancement produces real-life cost reductions was described by Gordon Moore, of Fairchild Semiconductor and Intel in the 1960s: With advances in semiconductor technology, Moore predicted that the number of transistors held per square inch of circuit board would double every two years. A medically related example of Moore’s Law is provided by the Human Genome Project: In 2001, cost to sequence a single genome was U.S. $100 million.
With improving sequencing technology, costs began to drop in line with Moore’s law until 2008. That year, “next generation” sequencing technology was introduced, resulting in cost decreases betraying Moore’s law. Sequencing cost per genome plummeted to U.S. $1,000, or 1/100,000 of its initial cost by 2017. In 2018, companies such as 23andme offer personalized genome sequencing kits over the internet for under U.S. $100.
Technological refinement and cost reductions may therefore enable widespread adoption of autonomous surgical robots. A “technological singularity” is a hypothesis that the invention of an artificial superintelligence will result in exponential technological growth, producing unfathomable changes to human civilization. As clinically capable, AI-controlled surgical robots may soon offer technical and economic advantages over human surgeons, a “surgical singularity” may therefore occur when capable, cheap and technically superior autonomous robots are introduced.
It is possible that these devices will transform the technical practice of surgery by enabling optimized removal of diseased tissues, faster operative times and better access to hard-to-reach body areas. These technical benefits will translate into improved survival and fewer complications for patients.
Debate regarding the future role of AI in health care should therefore be extended to include AI-enhanced interventional devices. The issue of how these devices will affect the role of surgeons should be considered and incorporated into any future operating frameworks. An ideal outcome is the augmentation of human surgical capability using AI and robotics, resulting in significantly improved patient care. Aside from encouraging the technical development of these devices, further work should be done regarding the potential impact in terms of workload, regulatory frameworks and the ethical and insurance implications that these devices may have upon surgical practice.
* Physicians can roughly be categorized as non-interventionalists or interventionalists: Non-interventional specialties include diagnostic radiology, pathology, oncology, psychiatry and neurology among many others concerned with the diagnosis and/or pharmaceutical treatment of medical conditions. Interventionalists are surgeons whose task is the physical removal of pathology, or reconstruction of tissues via cutting and suturing, underpinned by their manual dexterity.