Could machines make doctors obsolete anytime soon or at all? This is most certainly one of the most debated topics of this decade and will become even more so as time proceeds. There is definitely nothing such as a lack of arguments on either of the both fronts of this debate. This short fact- and opinion-based article will give an insight in and explore both of these in as much depth as possible.
Camp One: Arguments for Doctors
Firstly, Vanessa Rampton argues that computers are not able to have emotions and hence will not be able to care for patients as persons or anything at all (Goldhahn, Rampton, & Spinas, 2018). This is probably one of the most significant arguments for many patients. They want to be able to build a connection with their doctor, they want to be heard by someone who they can trust and they want someone to try their best to cure their decease due to feeling for them. From a social point of view, this is most certainly understandable but is it actually rational? To analyse this view, we need to consider whether this kind of relationship between a doctor and their patient has any long-term medical use for the patient and whether doctors are really open for such a relationship. It may make patients feel better in the first instance but will this social connection cure their disease in the long run? It most likely will not, as the superior knowledge combined with a more objective treatment enabled by machines, will probably be better at curing diseases than an emotional connection with a person. Also, it might be nice that someone will try their best to cure the patient’s illness due to feeling for them as the robot will undeniably not do this. However, the machine likewise will not work less determined at curing the decease from someone who they care less about: they always work and try at their full potential. This may or may not be the case with a human doctor. Doctors might also not be open to such emotional relationships with their patients and hence not better in this aspect than robots would be. For them, a patient is a job that needs to be done in the fastest time possible without the time to build an emotional relationship. Doctors are just not “curious enough to cognitively and emotionally relate to a patient’s situation” (Weintraub, 2016). Nevertheless, there are definitely some exceptions, some doctors that solely care about the wellbeing of the patient and put all their energy into achieving this. They are just not the norm. Maybe patient will get lucky and find one of these doctors.
Secondly, another controversial theme of our age: data protection. What could or will happen if these machines or algorithms start to record, store and distribute the patient’s private data? We may face the risk of a human doctor telling their college about us and our illness but they most likely will not possess a photographic memory and remember us and thousands of others forever. A machine possesses this ability: it is able to store a patient’s information in detail. This machine’s owner or operator will then be able to distribute this information and use it for all sort of things. They may discard it, but they might sell it to health insurances or any other interested party instead. In the case of health insurances, they could use the patient’s data to determine whether they are willing to hand out insurances to that person, based on the possibility of that person becoming sick in the future. A great thing for the insurance’s profits but less so for society overall. This point will gain even more weight in the future, when prediction technologies advance. Moreover, the machine’s data bank could also be hacked, when the hospital does not possess a secure firewall. One of the numerous examples, where hacking of hospitals happened is the WannaCry ransomware attack on the NHS in 2017. “The attack led to disruption in over a third (34%) of trusts in England, with thousands of appointments and operations cancelled” (Navin, 2018). This might then obviously lead to disastrous issues in terms of privacy or patient treatment. Yet, most people tend to give this argument to much weight as they believe that their privacy will suffer tremendously in respect to their current situation. Even though the possibility of data mishandling is most definitely worth considering, we must also have a look at current, already existing data mishandling risks, that no one seems to be concerned about. An example would be credit or debit cards: nearly everybody possesses them, nearly everybody uses them and nearly nobody is worried about doing so. Still, whenever you insert your card into an ATM or pay with it, your bank will know. Your bank will know where you are, they will know what you buy and they will know about any consumption patterns you might have. Like the machine’s operator, they could sell this information. What about advertisement companies? They would love some more information about your consumption patterns, so that they can send you their new personalised news letter. Even more controversial, banks are able to block your card and hence your financial independence whenever they want to. Maybe someone decides that you should not be able to leave the country and hence blocks any payment related to international travel. What about personal rights now? Already existing data issues such as credit cards are most definitely equivalent to the ones created through the use of modern technologies in medicine.
Thirdly, doctors complain that they will lose their jobs when we start to introduce new machines into medicine and no one wants to lose their job. However, this will only happen, if we start developing machines that are able to outperform doctors and would this not make doctors obsolete anyways? Would it be fair for the patient to give them a worse treatment due to the fact that we want to keep the doctor’s job, even if we would be able to achieve better results using an already existing machine? Probably not. Using a machine, which is proven to outperform doctors would be cheaper for the hospital and hence for the tax payers and would enable a better treatment for the patient.
Camp Two: Arguments for Machines
Firstly, humans need sleep and become tired. Hence, their performance as a doctor will vary depending on how much sleep they get and other factors such as their emotional state. Statistics show for example, that there is an “unacceptably high rate of […] fatigue-related medical error and injuries among health care workers” (Goldman, 2015). This then also means that human doctors will not be able to perform at their full potential constantly, what a machine could, due to not being affected by a lack of sleep or its emotions. Hence, using machines for certain medical task rather than doctors, especially those needing an extreme amount of mental alertness, would eradicate mistakes made due to tiredness.
Secondly, Jörg Goldhahn argues that more and more patient data is collected through systems such as personal monitoring devices or electronic medical records, which enables machines to obtain a picture of the patient’s health and their disease. He further says, that due to the enormous volume of data, it “is an illusion” that “today’s physicians could approximate this knowledge” (Goldhahn, Rampton, & Spinas, 2018). Machines are able to store millions of data points and process all of them within seconds using machine learning. Humans would not be able to remember this much in the first place, humans would forget this data after a while and humans would be overwhelmed by managing that much data in their heads. Still, data combined with artificial intelligence and machine learning holds huge potential in revolutionising medicine and thus cannot be ignored.
Thirdly, statistics show that there are too few doctors. In the UK alone, there are about “108`000 jobs unfilled”, which means that “one role in 11 is vacant” (Matthews-King, 2018). This leads to enormous waiting periods for patients and rushed treatments. What would happen to this problem, if we were able to replace doctors with machines? We would not even have to replace doctors with machines completely but rather take some tasks away from them, so that they can focus on helping out in areas where we are not able to create machines for yet. This would then enable us to help more patients in need and cut down long medical waiting cues.
Fourthly, machines or algorithms do not care about profits. They will not base their decisions on whether this treatment will earn them more money than another one would. This would then decrease the amount of operations that are more done for their cost rather than their use for the patient. Maybe the patient would have been better of doing nothing, but the operation equalled more revenue for the doctor. This can be a huge problem, especially with private doctors, as statistics show. In the United States for example, “Tens of thousands of times each year, patients are wheeled into the nation’s operating rooms for surgery that isn’t necessary”. Lucian Leape, a professor at the Harvard School of Public Health, estimated that there were “2.4 million cases a year, killing nearly 12`000 patients” (Eisler & Hansen, 2013).
Lastly, there are many examples of machines that are already outperforming human doctors at certain task. One of the many examples reaching from detecting heart attacks by listening to making medical diagnoses via a Chatbot (van Hooijdonk, 2018) is an algorithm that detects skin cancer, developed by Stanford University (Kubota, 2017). It uses artificial intelligence and a database of almost 130`000 skin disease images, to “visually diagnose potential cancer”. Due to the ubiquitousness of camera-equipped smartphones, this algorithm could enable us to detect skin cancer easily at home without the need to see a doctor. This would be a significant medical advance, as melanoma has only a tiny 14 percent survival rate when detected in its latest stages but an astonishing large one (97 percent) when detected early. Skin cancer is often detected late as people tend to neglect visiting doctors when they are unsure whether they really have something worth visiting for or they merely do not have the time for it. In such situations, “diagnosis through your smartphone could be lifesaving”. Through the implementation of many more of such algorithms and machines, our healthcare systems could be improved tremendously.
Overall Picture and How to Move Forward
Perceivable, this article is clearly favouring the ‘Machine’ side of the argument. This is the case, as machines are already and will even more so, enter medicine in the future. In order to be prepared for this change, it is important that we start accepting robots and machines in medicine. Nowadays, humans tend to expect machines to be perfect at what they do: a single mistake is taken as proof, that they should not be used. A common example would be self-driving cards: the death of a single person due to these cars is taken as unacceptable by many and hence as proof that self-driving cars must be highly regulated. However, bringing this into context tells us that in the United States alone, in 2016 there were “40`200 vehicular fatalities”, with 90 percent of them being due to avoidable human errors (Demers, 2018). Many of these would most likely be avoidable through the usage of self-driving cars. Hence overall more lives would be saved, even if their use leads to some non-human caused casualties. Rather than being perfect, would it not be sufficient if they perform better than humans? We might not be able to achieve perfection instantly through using machines but improving our current standard by a huge margin seems pretty much excellent already. The real problem is, that if patients do not accept machines in medicine, they will not be willing to get treated by them and hence technology will not be able to establish itself to the extend, which is able to save lives in the future. Nevertheless, we still should not implement various kinds of machines in medicine without ensuring that important safety concerns, such as the patient’s privacy, are being taken care of.
Demers, J. (2018, June 2). Self-driving cars will kill people and we need to accept that. Retrieved February 25, 2018, from The Next Web: https://thenextweb.com/contributors/2018/06/02/self-driving-cars-will-kill-people-heres-why-you-need-to-get-over-it/
Eisler, P., & Hansen, B. (2013, June 20). Doctors perform thousands of unnecessary surgeries. Retrieved February 26, 2018, from USA Today: https://eu.usatoday.com/story/news/nation/2013/06/18/unnecessary-surgery-usa-today-investigation/2435009/
Goldhahn, J., Rampton, V., & Spinas, G. A. (2018). Could artificial intelligence make doctors obsolete? British Medical Journal.
Goldman, B. (2015, May 5). It’s time for doctors to admit that our lack of sleep is killing patients. Retrieved February 27, 2018, from Quartz: https://qz.com/389958/its-time-for-doctors-to-admit-that-our-lack-of-sleep-is-killing-patients/
Kubota, T. (2017, January 25). Deep learning algorithm does as well as dermatologists in identifying skin cancer. Retrieved February 21, 2019, from Stanford News: https://news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/
Matthews-King, A. (2018, September 11). NHS staff vacancies rise nearly 10% in three months amid unfolding ‘national emergency’, report shows. Retrieved February 27, 2018, from Independent: https://www.independent.co.uk/news/health/nhs-nurse-doctor-staff-shortages-vacancies-waiting-times-crisis-brexit-deficit-a8532701.html
Navin, D. (2018, April 20). Why are hospitals such a major target for hackers? Retrieved February 25, 2018, from IFSEC Global: https://www.ifsecglobal.com/global/hospitals-major-target-hackers/
van Hooijdonk, R. (2018, December 17). Artificial Intelligence in Medicine: Current Trends and Future Possibilities. Retrieved February 22, 2019, from Clinician Today: http://cliniciantoday.com/artificial-intelligence-in-medicine-current-trends-and-future-possibilities/
Weintraub, P. (2016, July 4). Doctors have become less empathetic, but is it their fault? Retrieved February 26, 2019, from aeon: https://aeon.co/ideas/doctors-have-become-less-empathetic-but-is-it-their-fault