Talk of how artificial intelligence (AI) can improve medicine is ubiquitous. But providers are hesitant to trust AI – despite claims that it can improve diagnostics, detect diseases earlier, and help create personalized treatment plans.
Perhaps this is because technology in the field often outstrips humankind’s capacity to control, monitor, debate, and absorb innovations and new applications. And, while providers are seeing more and more patients, they need to understand AI enough to be able to use it and trust it.
Layered onto concerns about the technology itself are worries about job security and ethics. The combination means that many providers hesitate to use AI.
AI Earned Some Caution
Most people use AI daily via our cellphones, tablets, laptops and other devices. Apps quickly learn our likes and dislikes, pointing us to targeted ads and collecting more information than we would like.
This is a type of AI, built-in machine learning (ML), that allows computers to gather large amounts of data, and then use algorithms to sort information and recognize patterns, which the machine continues to build on (or learn).
ML has also brought unwelcome threats via seemingly innocuous, daily use of our smart machines. At a time when Facebook algorithms are targeting users to sway global elections, Internet identity theft runs rampant, kids are becoming addicted to their smartphones, and driverless cars are killing both drivers and pedestrians, AI might seem to be more of a liability than an asset. Health care is no exception:
- MD Anderson Cancer in Texas had to cancel its contract with the Watson supercomputer’s Oncology Advisor after it’s algorithms failed to find the best cancer treatments for patients.
- In a 2018 British Medical Journal Quality and Safety article, Challen et al. warned “Developing AI in health … is a fertile area of research, but the rapid pace of change, diversity of different techniques and multiplicity of tuning parameters make it difficult to get a clear picture of how accurate these systems might be in clinical practice or how reproducible they are in different clinical contexts.”
- And, as highlighted in a recent opinion piece in the Hippocratic Post, machines are in danger of introducing their own biases into patient care.
Add to this Silicon Valley’s overzealous promotion of the utility and infallibility of new tech where aI-driven robots and computers have a tendency to be overhyped. Advertisers, the media, and Big Tech all seem to conspire to sell us “The Next Big Tech Thing” that we can’t live without.
Nevertheless, there are enormous risks in dismissing new, unfamiliar technology out of hand.
What’s So Intelligent About AI?
Yes, the Internet can be a repository of misinformation and manipulation; however, AI-driven machine learning built into the web provides countless benefits for physicians. A web-based PubMed search can yield vast amounts of clinical information in just a few clicks, data that 20 to 30 years ago would have taken days or months to retrieve.
And now that data – thousands of diagnoses and treatments and position statements – can be accessed at lightning speed because of ML built into millions of reputable websites.
Similarly, ML algorithms built into robots and computers in medicine allows computers to use algorithms to:
- Improve diagnoses by recognizing and sorting disease and treatment patterns.
- Draw on thousands and even millions of national and international scientific studies and policy statements.
- Reference enormous amount of data from patient records within and outside of their institutions.
- Access accurate diagnostic data and predictive treatment plans.
- Keep patients’ electronic health records updated in real time.
- Tailor treatment plans to each patient’s individual needs.
- Share outcomes with national and international peers.
- Reduce or even eliminate medical errors and minimize malpractice claims.
- Alert local/national health agencies about emerging threats and pandemics.
- Accurately track and perform routine administrative tasks.
- Analyze patient data, and alert institutions to ineffectual and outdated treatment paradigms.
In addition to those benefits, leveraging AI-driven data has the potential to ameliorate or even eradicate certain vexing health care issues, including provider shortages and burnout, standardization of electronic health records, drug/medical supply shortages, resurgent diseases, antibiotic resistance, plus administrative inefficacy and rising care costs.
AI technology has the unique ability to reach people throughout the United States via innovative phone apps that can track progress, monitor disease, allow patients 24/7 access to providers and their staffs, and ameliorate health care deserts in underserved rural and urban areas of the country and even the world.
Telemedicine is a burgeoning business that promises to level the playing field by directing services to the most needy individuals with the least access to care. New AI tech allows patient and provider to stay in touch and keeps patients from slipping through the cracks.
Will All Doctors Be Robots?
Robots and computers will never replace physicians and other health care staff. Although it may be odd to think of Mr. Roboto as a partner, the technology has limitations. The “human-like” qualities of robots and computers should not be confused with characteristics and tasks that are uniquely human.
As Sunnie Southern, health and life sciences lead at Onix pointed out, “AI let’s physicians be physicians again by having to spend less time entering and interpreting data, and spending more time doing what they love – caring for patients.”
As far as AI taking over decision-making, Southern said, “AI delivers analytics that help physicians make highly complex decisions and choose from two or three options in some cases.”
In addition, The Medical Futurist says that physicians and surgeons have almost no chance of their professions being automated and “the medical community should not fall for the fear mongering around AI.”
The Futurist delineates five uniquely irreplaceable human qualities that simply are not programmable, including:
- Empathy: As providers and patients make the emotional journey through difficult treatment and life-altering diagnoses and sort through complicated data to make those decisions, the human mind and touch will continue to be needed.
- Treatment has a non-linear quality: Data and quantitative analysis are just one part of the equation. Diagnosis and treatment require “creativity and problem-solving skills that algorithms and robots will never have,” noted the Futurist.
- Humans will always be required to leverage the tools AI provides. For example:
- Specially trained surgeons are needed to operate the complex robotics of the Da Vinci surgical system.
- Although the revamped and widely used IBM Watson oncology application provides evidence-based treatment suggestions, doctors, along with their patients, must discuss and choose the appropriate treatment option.
- As the Futurist points out, “No robot or algorithm [can] clearly interpret complex, multi-layered challenges involving the psyche. While [AI] … provides the data, interpretation will always remain a human territory.”
- There are tasks that algorithms and robots cannot perform. Those include the Heimlich maneuver, administering emergency CPR, or even selecting when to take a patient’s hand (e.g., when subtle signs of anxiety are manifested).
- “Us versus them” is the wrong view of emerging AI technologies. The Futurist pointed out “No matter whether it’s A.I., robotics, augmented or virtual reality, we should accept that they have a massive influence on the way health care operates, and then start utilizing their power.”
Several studies and expert commentaries support these views and provide evidence that AI is not meant or able to replace humans in health care. A recent study by the Massachusetts Institute of Technology, for example, concluded that “By analyzing doctors’ written notes on intensive care unit patients … the doctors’ ‘gut feelings’ about a particular patient’s condition played a significant role in determining how many tests they ordered for the patient.”
The researchers added, “This intuition plays an even stronger role during the first day or two of a patient’s hospital stay, when the amount of data doctors have on patients is less than on subsequent days.”
Major Medical Societies and Health care Organizations Endorse AI
As AI’s utility has become apparent and its use increases in myriad fields of medicine, nearly every U.S. medical association, society, and public health agency has strongly advocated for the use of AI in health care as a transformative clinical tool. For example, a 2018 statement from the American Medical Association included the following from board member Dr. Jesse M. Ehrenfeld:
“As technology continues to advance and evolve, we have a unique opportunity to ensure that augmented intelligence is used to benefit patients, physicians, and the broad health care community,” Dr. Ehrenfeld said. “Combining AI methods and systems with an irreplaceable human clinician can advance the delivery of care in a way that outperforms what either can do alone. But we must forthrightly address challenges in the design, evaluation and implementation as this technology is increasingly integrated into physicians’ delivery of care to patients.”
The website of the National Institute of Standards and Technology recently agreed. “Remarkable surges in AI capabilities have led to a number of innovations including autonomous vehicles and connected Internet of Things devices in our homes,” NIST wrote. “AI is even contributing to the development of a brain-controlled robotic arm that can help a paralyzed person feel again through complex direct human-brain interfaces.”
The organization is working with key government health agencies to make emergent AI technologies “more secure, usable, interoperable and reliable.”
A recent feature article from the American Academy of Neurological Surgeons extolled the virtues of AI, including that the technology speeds patients toward time-critical interventions; reduces costs by accurately predicting who will benefit from interventions; rapidly identifies unusual cases that require urgent, additional testing; and supports neurosurgeons in designing optimal interventions.
The American Gastroenterological Association (AGA) published a commentary on the organization’s take on AI. “In gastroenterology, AI can now identify bleeders, colorectal polyps, and angioectasias at impressive rates of accuracy and can be deployed for pathological evaluation of lesions at the microscopic level as well,” it said.
The AGA commentary added, “The AGA Center for GI Innovation and Technology recognizes AI’s potential and we are committed to exploring how it can help us improve patient outcomes and potentially reduce health care costs, physician workload and complications/miss rates.”
Jason is an independent group of elite scientists, many of them Nobel-prize winners, that advises the United States government on matters of science and technology. It recently released a report commissioned by the Agency for Health care Quality and Research that detailed how the use of AI in health and health care is promising, and actionable. Specific recommendations include:
- Capturing smartphone data to enable patients to monitor health and integrate that information to support AI databases.
- Capturing and integrating social and environmental data from communities to understand differing health outcomes affected by these data points.
- Supporting competitions to further advance data generation and AI development.
Endorsements of the use of AI to improve medical care keep rolling in. An April of 2019 FDA statement from Commissioner Dr. Scott Gottlieb affirmed, “Artificial intelligence and machine learning have the potential to fundamentally transform the delivery of health care.”
The commissioner added, “As technology and science advance, we can expect to see earlier disease detection, more accurate diagnosis, more targeted therapies and significant improvements in personalized medicine.”
Major U.S. Hospitals Are Already Using AI
Without a doubt, specific safeguards and rigorous review — along with input from physicians, hospital executives and other stakeholders in the field — is required to ensure that AI remains safe and effective in health care. However, turning from the technology could prevent important, time- and life-saving AI-driven applications from being widely applied.
AI’s positive influence already is being used at several renowned U.S. hospitals for:
- Predictive analytics: Monitoring patients to prevent emergencies by analyzing data for key indicators
- Chatbots: Automating provider inquiries and pointing patients to appropriate specialists
- Predictive health tracking: Tracking patients’ status using real-time data collection
The Mayo Clinic is partnering with a tech company to develop molecular sequencing for 1,000 patients in studies related to immunotherapy for lung cancer, melanoma, bladder cancer, breast cancer and lymphoma. Mayo is also investing in AliveCor, an AI-driven platform that monitors patients for early atrial fibrillation, and Kardia Mobile, a mobile phone-enabled EKG.
The Cleveland Clinic is collaborating with Microsoft to help medical centers identify at-risk patients in ICU care, and to tap into advanced analytics and predictive care through the AI digital assistant Cortana, used by 126 million members of the public.
Massachusetts General has installed a super-computer that contains 10 billion medical images for radiology and pathology and will soon add electronic health records to the mix. The idea is to make medical data collection, storage, and retrieval consistent by creating a nationally integrated health care data system that provides fully accurate information on each patient while saving health networks and providers time and money.
Johns Hopkins Hospital, partnering with GE in 2016, launched a hospital command center that uses predictive analytics to support a more efficient operational flow. The Command Center receives 500 messages per minute and integrates data from 14 Johns Hopkins systems across 22 high-resolution, touchscreen enabled computer monitors.
And that’s just a smattering of AI’s use in major medical centers. AI has already provided marvelous advances in our everyday lives, in the tiny AI-driven devices we carry around with us daily: our smartphones. AI in medicine can make a difference, a life-and-death one, for patients worldwide.