Artificial Intelligence: The Future of Medicine?
Will doctors— one of the most highly trained professionals— become obsolete?
Past
Memorial Sloan Kettering Cancer Center (MSKCC) is the oldest and largest private cancer center in the world. The hospital has been delivering exceptional patient care, innovative research, and outstanding educational programs for over 130 years, and differentiates itself from other large hospitals by focusing on cancer care.[1] MSKCC has consistently been ranked as one of the top two cancer hospitals in the US based on its ability to “provide patients with the best care available as it works to discover more effective strategies to prevent, control, and ultimately cure cancer in the future”.[1, 2]
Historically, MSKCC has delivered on this promise by:
- Hiring physicians with an extraordinary depth and breath of experiences in diagnosing and treating all forms of cancer
- Using multidisciplinary care teams that combine the expertise and experience of numerous specialists
- Leading the way in developing new ways to diagnose and treat cancer by maintaining one of the world’s most dynamics programs of cancer research
- Training the next generation of healthcare professionals
With advances in technology, MSKCC has expanded its operating model by engaging in innovative collaborations, one of which has revolutionized the way care is delivered to patients: IBM Watson Oncology.[1]
Present
Watson—a cognitive computing system—“enrolled” in medical school in 2011 after becoming the Jeopardy world champion.[3] Watson “graduated” from its medical residency in 2013, and at that point had already analyzed 605,000 pieces of medical evidence, 2 million pages of text, and 25,000 training cases. 14,700 clinician hours were spent fine-tuning Watson’s decision accuracy.[4]
Watson can retain infinitely more medical knowledge than a doctor and is also more accurate, more consistent (inconsistency is a surprisingly common flaw among doctors), and faster at decision-making.[4] For instance, Watson correctly diagnosed a 60-year old woman’s rare form of leukemia within 10 minutes, while doctors at the University of Tokyo were unable to diagnose her even after assessing her for months.[3] For a time-sensitive disease like cancer, the speed at which patients are diagnosed can have significant implications on a patient’s chances of survival.[5]
Clinicians at MSKCC partnered with IBM to strengthen Watson’s “background” in oncology. Today, Watson is considered a “colleague” that helps physicians provide better care. So what exactly does an appointment with Dr. Watson look like? IBM describes the process as follows:
“First, the physician might describe symptoms and other related factors to the system. Watson can then identify the key pieces of information and mine the patient’s data to find relevant facts about family history, current medications and other existing conditions. It combines this information with current findings from tests, and then forms and tests hypotheses by examining a variety of data sources—treatment guidelines, electronic medical record data and doctors’ and nurses’ notes, as well as peer-reviewed research and clinical studies. From here, Watson can provide potential treatment options and its confidence rating for each suggestion.”[4]
Future
Innovations in healthcare delivery will continue to change how hospitals operate, but will doctors be replaced by technology? Maybe in the future, but certainly not today. I would characterize a physician’s responsibilities as follows:
Learn –> Analyze (i.e, diagnose) –> Evaluate (i.e., put things into context?) –> Decide –> Execute (i.e., deliver care)
Watson’s ability to fulfill these responsibilities is limited. Its learning is dependent upon knowledge produced by physicians and other healthcare providers, while its evaluation and decision-making processes lack the ability to empathize with the patient and take context into account. Finally, Watson completely lacks the ability to execute.
Having said that, the line between humans and technology is continually becoming more blurred. Consider the da Vinci® Surgical System, a robot that aids physicians during surgeries.[6] Or consider Pepper, a Japanese robot that recognizes emotions (based on the human’s tone, facial expressions, body movements, and words used), and adapts its behavior to the human’s mood.[7] Given these capabilities, a future where technology will be the sole provider of healthcare is not unreasonable. Studies have actually shown that humans feel empathy for robots experiencing “pain”, suggesting that as these technologies become more prevalent, patients may become less resistant to being treated by non-humans.[8]
Technology and digitization will continue to provide opportunities for MSKCC to provide exceptional care. I think that MSKCC should focus on finding ways to empower physicians. Watson is certainly an example of such a technology, as it helps physicians make better informed decisions. However, addressing fundamental healthcare inefficiencies should be the hospital’s number one priority. Having worked at MSKCC’s main competitor, I have been shocked by the amount of time that doctors spend on administrative responsibilities that prevent them from focusing on what they do best: deliver care to patients. Discovering and implementing technologies that enable physicians to focus on practicing medicine is, in my opinion, the key for MSKCC to continue to “deliver the best care available”.
Word count: 797
References
- About Us. Available from: www.mskcc.org/about.
- U.S. News & World Report Names Memorial Sloan Kettering Top Hospital for Cancer Care in the Nation. Available from: www.mskcc.org/blog/u-s-news-world-report-names-msk-top-hospital-care-nation.
- IBM’s Watson gives proper diagnosis after doctors were stumped. Available from: www.nydailynews.com/news/world/ibm-watson-proper-diagnosis-doctors-stumped-article-1.2741857.
- IBM’s Watson Supercomputer May Soon Be The Best Doctor In The World. Available from: www.businessinsider.com/ibms-watson-may-soon-be-the-best-doctor-in-the-world-2014-4.
- 5 Ways The IBM Watson Is Changing Health Care, From Diagnosing Disease To Treating It. Available from: www.medicaldaily.com/5-ways-ibm-watson-changing-health-care-diagnosing-disease-treating-it-364394.
- The da Vinci® Surgical System. Available from: www.davincisurgery.com/da-vinci-surgery/da-vinci-surgical-system/.
- Who is Pepper? ; Available from: www.ald.softbankrobotics.com/en/cool-robots/pepper.
- Suzuki, Y., et al., Measuring empathy for human and robot hand pain using electroencephalography. Sci Rep, 2015. 5: p. 15924.
- IBM’s Watson is better at diagnosing cancer than human doctors. Available from: www.wired.co.uk/article/ibm-watson-medical-doctor.
I certainly agree there is a lot under the responsibility of a doctor that cannot be replaced by a machine (currently), but I would like to explore the potential impact of taking certain steps out of the value chain. Let’s say for example that IBM Watson is instituted as a permanent diagnostic step in all hospitals. MDs still have to undergo rigorous training and knowledge absorption for medical research (knowledge advancement), but practitioners are no longer performing Dx due to their inherent human bias. Therefore, after graduating from med school, a doctor may have fewer tracks to fall upon. Medical research and surgery remain open, but post-diagnosis implementation roles may swiftly be switched over to lower cost nurse practitioners. The healthcare space either needs fewer MDs or may start paying them less.
Awesome, and scary! I’m hoping that all those years of studying are not “obsolete,” but the diagnostic potential here is hard to debate. I do think that this point about patient resistance to non-human providers is an important one, especially given the suspicion that we already have about tech companies collecting other types of personal information. Of course, physicians already enter that data into databases, and computers can already integrate keywords to suggests diagnoses with decent accuracy. But by completely removing the physical person from that interaction, I think you lose an important element of trust. Trust between a doctor and patient is just as important to quality as good diagnostics, and I question whether or not people would be willing to reveal their anorexia, their depression, their history of abuse, to a computer screen that to date cannot reciprocate with any form of comfort or reassurance.
A really interesting post! I agree with Pete on the importance of trust in the doctor-patient relationship but I would love to see AI in healthcare continue to develop in two broad areas. Firstly, by playing an increasing support role for MDs through Learn > Analyze > Evaluate > Decide. Secondly, by building the capacity to execute in order to give patients the option of choosing a human vs. machine MD. That is, for W price you can have a human MD who has an X% success rate for this particular procedure or for Y price you can have a machine MD who has a Z% success rate for the procedure.
Hospitals are places where I do not find it pleasing to visit, but want to have a pleasant experience when I have to go. The basis of the pleasant experience is the psychological safety that I will be treated well by a trusted doctor. From this there is still a controversial question as to who is more trusted than the other – human doctor or machine doctor? With increasing capabilities of artificial intelligence with introduction of AlphaGo where the machines are now in a stage nearing intuition aspect of human beings, it will not be long before humans can develop a machine that acts like a human doctor. However, question remains as to whether humans will actually trust a machine doctor.
I would say the success of digitization in the healthcare industry specially in this case when machines are expected to replace doctors is contingent to the success rate of the job done by machines compared to the success rate of doctors for the same jobs. One would have to prove that errors done by machines are less frequent or less important compared to doctors at least to get regulatory approval (I hope so). Besides that, and as mentioned in previous comments, I wonder if humans will be willing to see such a change happen as one of the biggest added value of a doctor is empathy with the patients and their ability to ask the right questions and push people to open up and reveal important insights
Wow great post! Watson clearly has the potential to make a cataclysmic impact. The way I see it, once Watson becomes more sophisticated and is able to deliver medical care, physicians will be able to spend more of their time conducting research. I am confident that this shift will fuel innovation in the medical field. Also, I realize that dooctors may be concerned about their job function being taken over by computers, hoewever, I see Watson more as an opportunity than a threat. Phycisians can collaborate with Watson to come up with optimal treatments for patients, similar to what the “Best Doctors” program is accomplishing at Sloan Kettering. (The link below porvides details about this program)
http://www.imedicalapps.com/2016/10/ibm-watson-best-doctors-cancer/
Great Article. AI exhibits a striking paradox in it’s promise to drastically improve our lives, whilst raising troubling concerns for replacing humans in various knowledge-based roles. Curious to know if MSKCC has any prediction on the future of AI in healthcare?
Very informative article and I agree with you on man points. However, I don’t agree that “Watson’s ability to fulfill these responsibilities is limited”. I think that AI should help physicians to diagnose and evaluate. Physician should then spend more time on decision and execution. An operation doesn’t have standard procedures and it still takes experience for a physician to operate on a human body. I also think that trust and empathy are elements that can not be replaced by AI. Robots will be a good enchantment for physicians to be more efficient, but I won’t trust a robot to operate on surgery…
I think there will always be a need for doctors. A large part of a doctor’s job is get at the “real” information. We as patients sometimes can’t fully understand our pains or issues. I don’t think there is a series of questions and answers that could replace the human element of discovery. But I am in support of using technology to help reduce managerial tasks doctors are tasked with!