No, they really don't.<p>I'm married to a physician and I can tell you that Medicine is an incredibly nuanced profession, and though doctors and nurses can benefit greatly from computer assistance, they cannot be reliably and safely replaced by computer algorithms. The single data point of Watson diagnosing lung cancer is not enough to generalize over the entire profession and range of ailments.<p>Medicine is a great deal more than differential diagnoses, doctors are a great deal more than "professional diagnosers", and nurses are a great deal more than morphine dispensing diaper-changing automatons.<p>Most medical schools (in my country, at least) have moved to an entirely evidence-based and patient-centred-care model years ago. The problem of course is that evidence is not always easily gathered because of ethical issues (e.g. most hypothermia knowledge comes from Nazi human experimentation on POWs and Jews. Also it's hard to conduct a study on OB/GYN topics because you could potentially be putting infants, fetuses, and their mothers at undue risk). Being treated by a computer is not going to change this problem of poor evidence.<p>That forum post was written by someone who has only a shallow appreciation for/understanding of the field of medicine.
Of course, they won't be replaced anytime soon, but will be augmented via technology, just like the rest of us. Hopefully sooner rather than later. Every field is disrupted by software in the same fashion: massive resistance where practitioners claim the human element is crucial, then of gradual swapping out of manual for automated where it helps, to in the long run largely automated.<p>I see a big parallel here with education: I am related to teachers and of course they will swear up and down that hand-crafted, personalized lesson plans are key, and teacher-student interaction invalidates the need entirely for automation. I'd imagine in the long run 80% of these "only humans can do it" tasks will be shown to be performed better by expert systems and data-driven algorithms.<p>Of course the catch is that nobody wants to be told that a large part of the skills they've developed over a lifetime have been made obsolete or relatively worthless through some code and electrons. Software engineers, and many engineering fields in general are used to the idea that every 5 or 10 years most practical knowledge is obsolete, that computers are relentlessly automating things done manually just a few years previous, and to not fight it but embrace it and try to leverage it. I don't think this is a common mentality for most professions. Even physicians for the most part, unless they work in an academic institution, let their reading fall by the wayside and develop habits and ritual around what they were doing when they finished residency.
I hear people saying that medicine can't have doctors be replaced by computers and it's more of an art than a science. I'd like to get more information and see why they think that and if their reasons really hold up. It intuitively seems to me computers can do a better job. Here are a few reasons:<p>1) When you think about what a doctor does, they use a combination of their memory, experience, knowledge and judgment.<p>2) Computers can remember a lot more relevant facts and wonmt forget them.<p>3) A doctor's experience is just with their own patients and what they hear at conferences and the literature they read. A computer network can literally aggregate experiences of many different doctors and cases around the world. A new superbug coming out, for instance, may have a treatment that few doctors know, because it's more prevalent in another country -- but a person using the network will be made aware of this treatment. The network will have much more experience than an individual. It will aggregate the outcomes of many therapies and studies.<p>4) The knowledge of the network can be extracted from the experiene and be much more extensive. Both beause the experence is extensive and because various methods of statistical analysis can be employed. A doctor isn't going to sit there and cross reference worldwide statistics on the success of using remedy X given factors of race, gender, age, etc. The computers can.<p>5) Finally, the judgment. Here is why the computers can beat a doctor's "gut" overall. They can cross correlate the knowledge from all around the world and find the most major factors. Then given the symptoms, race, age, location, etc. they can suggest the possible diagnoses given realtime epidemiological statistics and the additional tests that need to be done to establish which diagnosis is most probable. In many cases this establish a path to just one diagnosis with statistical accuracy exceeding what one doctor can produce. It can also list the next several possible diagnoses and the tests to establish them. After the diagnosis is established, it can find the therapies along with risk factors gleaned from all the relevant outcomes from around the world.<p>And this is just for conventional medicine. When we get into genetic therapies, computers and information science will become a huge part of medicine.
I'd like to make two points.<p>Firstly, before anyone builds an AI that can look after people, maybe they can fix the software for looking at Xrays WHICH ALWAYS STOPS WORKING IN THE MIDDLE OF A BUSY CLINIC.<p>Secondly, I challenge someone here to put a number on how much it would cost to produce an AI that can operate robustly in a clinical environment. This includes the costs of developing such a thing, deploying it as well as testing it. Testing it would arguably be one of the most difficult and complex software testing problems ever. Now whatever that number may be, I hazard to guess that it's a lot. So what you are proposing is to spend a lot of money, so that we can practice medicine... slightly better... MAYBE slightly better.<p>The greatest improvements in healthcare come from the development of novel therapeutic technologies (antibiotics, vaccines, anaesthesia) or public health measures that change behaviour (sanitation, seat belts, quit smoking campaigns etc).<p>All of this frankly makes me sceptical of the wisdom of spending zillions on building medical AI, even if such a thing were possible.
I feel like a sucker for responding to such obvious self-serving link bait, but full of gratitude to the very human staff at the hospital where I had a minor procedure last week I will anyway. Here's the simplistic refutation that is all the simplistic premise deserves:<p>Until you pass a Turing AND a Voight-Kampff test, stay the fuck out of my health system.
absolutely.<p>case in point: automated call centers. so much better than a human being. instant recognition of what i need, how to help me - simply awesome.<p>i can see it now - press 1 if you're stomach hurts, 2 if your leg hurts ... OPERATOR!<p>machines work great when steering and diagnosing other machnes? the fuzzy and subjective human condition? not so much.
I work on a supposedly state-of-the-art GE scanner. If I let it select parameters it wants there are real, measurable negative consequences. It is a buggy, bad piece of hardware. Sure, there are better machines out there, but codifying the processes the human operator has is not going to happen anytime soon I don't think - hell, GE don't know how to get a user selection of a bool parameter right (it has a text field for the user to type in 1 for on, 0 for off. No toggles, buttons etc). Still, the craptastic system spurred me to learn a bit of coding to make stuff happen that the scanner can't/won't do. Don't rush in to replacing the humans too soon.
At the very least we can get started by having doctors be thoroughly assisted by computers. Transition from humans to full automation is always difficult, so it's good to start with the computer being a very accessible and very powerful resource that helps inform the doctor's decisions through data.
Ha! Nurses need to be replaced by computers and robots? The only mention of 'nurse' in the rant is 'the uniforms carry bacteria'... because a robot wouldn't? Every surface in a hospital carries this stuff. I'd like to see the pricetag on a robot (or range of robots) that can do all the physical skills of a nurse, let alone the social management.<p>I remember having to hold onto a patient's leg for a nurse to put in a needle in a tiny vein in the foot, as all the others had collapsed. The patient was in an altered conscious state and was moving around - I was keeping the leg <i>relatively</i> still. I'd like to see a robot succeed that task without being explicitly designed for that exact use case. It would be genuinely impressive.
Medicine is no longer an intuitive game. The best results come from following rigorously tested algorithms, not intuition. Research, on the other hand, is where the creativity is and that should continue to be done by humans.
I'm a physician and software developer.<p>Diagnosing is a actually a rather small part (but very important) of my daily job as a physician. Doctors can't be replaced by computers and robots at the moment because much of what I do is listening, comforting, reassuring, educating, managing, administrating etc etc.<p>We urgently need better computer support tools.
Will doctors be replaced my robots? When you feel comfortable being comforted by a robot they will.
Other posts by this author:<p>"Programmers need to be replaced by computers...no more buggy code!"<p>"Lawyers need to be replaced by computers--no more sleazy ambulance chaser ads!"<p>"Money needs to grow on trees! No more financial problems!"