9 ms·
AI in medicine will help doctors, not replace them
- __s 7y agoIf an AI assisted doctor is 125% more efficient than a non assisted doctor, you only only need 80 doctors where you'd otherwise need 100. That's 20 doctors that have effectively been replaced Then consider the amount of not-doctor roles in medicine which can be automated/assisted
- roberte3 7y agoYour looking at it wrong. A doctor's appointment schedule is crazy busy, with "LONG" appointments being under 20min per patient. Imagine a scenario where a doctor is able to spend time with the patient/their medical history etc. I spend a lot of time going to doctors appointments these days, and a large number of the doctors that I see, now have an assistant/transcriptionist in the room to manage the EMR/charting app, because that enables them to focus on the patient.
- x220 7y agoWould the doctors take fewer appointments per day and make less money? Or do doctors get paid by the hour?
- godelski 7y agoThat's definitely one way to look at it. And I hope that's what happens. But is it really wrong to speculate that that isn't going to happen? I think it would be naive to not speculate how we would handle such a situation. Contingency plans are important.
- iudqnolq 7y agoI've never understood this argument. Right now we have decided implicitly as a society we're fine with the current level of care. If we can provide that same level of care cheaper, what makes you think health care providers will decide to pass those savings onto patients in the form of more care for the same price?
- Engineering-MD 7y agoI would disagree with this statement. Every year, expectations rise, and it is becoming increasingly difficult to meet these expectations. Furthermore, I think expectations are tempered by the knowledge that staff are over worked and time is limited. It is infeasible to practice medicine the way it us taught in medical school in practice due to time constraints. And like everything else, the work tends to fit the time available.
- iudqnolq 7y agoI hope very much you're right, and that as medicine gets cheaper we get more of it.
- Swizec 7y agoOr do you increase throughput by 20% and now more people can get healthcare who otherwise wouldn’t?
- roenxi 7y agoThe elephant in the room with doctors is that there are many countries where the supply is constrained by some sort of guild system and only very clever people do well. If AI can lower the bar to the point where Ned the Nitwit can type in the symptoms, read the screen and get a reasonable diagnosis then 'doctors' are going to be a completely different class of people even if they share the same title.
- JamesBarney 7y agoThere are a lot of current doctor work which involves. "Any issues with the medication your on, ok, can I see your bloodwork, no red values, here is your prescription."
- dragontamer 7y agoWhich really should be the job of a nurse or physician's assistant. Doctors should be reserved for the cases that requires the ~10 years of training they receive. Most people probably only need a yearly-visit to the physician's assistant or a nurse for bloodworks. There shouldn't be a need to see an actual doctor on a regular basis (unless you have some chronic ailment that nurses / physician assistants can't handle). ------------ The prescription is probably the only part of that routine that requires a doctor.
- toomuchtodo 7y agoEntirely accurate. It's the AMA/doctor lobby (purposely constraining supply to keep wages high) that is preventing nurses and nurse practitioners from filling this role. My prescriptions and bloodwork are already in an EHR/EMR (EPIC) and shared with Apple Health, machine intelligence should be doing as much heavy lifting as possible.
- x220 7y agoThis question might be out of place, but why should I trust ML algorithms for my health? To be a doctor you need at least a decade of education and training, as well as licensing and adherence to a code of ethics. To be a tech bro you just need to get hired. You don't even need a degree. You could come to work stoned and as long as your boss doesn't notice, you're fine and can continue to write the code that my health could rely on.
- bschne 7y agoI feel like I've read somewhat similar pieces about every single field affected by automation and AI at this point. In every other article, there's a dichotomy between "everyone will keep their jobs and be more efficient and have more time to focus on what matters" and "everyone will lose their jobs and be replace by machines". This then gets resolved to "we will never be replaced because 'human factors'", therefore option a). I am sure the author knows their field well, but this just doesn't seem to provide any interesting / new viewpoint on the issue beyond the arguments that usually come up in superficial discussion of the topic.
- mr_toad 7y agoThe “everyone will lose their job” articles are usually conflating the current state of the art in AI with artificial general intelligence, which is still just science fiction.
- godelski 7y agoThe bigger question, imo, is how we handle those job losses. Retraining doesn't really work. So what do we do with those people? Just say "whoops, your job is automated away. I'm sure you'll find some new equally paying job."? Because that seems like a good way to start a very bloody revolution.
- tedivm 7y agoSomebody has to annotate the data. Joking aside, I think for a lot of medical stuff the shift is already happening- there are less people going into certain fields (such as radiology) due in part to fear that it isn't a long term career. This is driving up radiology pay as the demand is outpacing the incoming supply. While I ultimately believe that most things are going to get automated, I do agree that in the short to medium term were going to see AI augmenting medical professionals rather than replacing them. This is just the natural progression of things- the technology can start off with the low hanging fruit and gradually take on more and more of the work. This provides immediate benefit while building funding and knowledge that can be used to take the next big step.
- rscho 7y agoIf people and especially doctors were serious about improving healthcare, they should strive to build reliable data collection means. That's the prerequisite to efficient 'AI', and would be a far better use of funding than all the present half-baked 'improvements' and star-system research. We could really do science, then. A man can dream...
- ivalm 7y agoThere is a lot of data collection. With everyone using EMRs massive datasets are available. They just aren't available to people outside the walled gardens for PHI/privacy reasons. At Kaiser Permanente, I regularly train models on 10s/100s M patient/dr encounters. Our transformer language models are fine tuned on multi-billion word corpuses. Some of our models do real time inference on millions of patient notes per month. The thing is, from data governance standpoint our org, and most other health orgs, just aren't comfortable sharing this data with outside businesses or even each other. And of course orgs like KP strongly don't believe in licensing internally developed products to other health orgs.
- rscho 7y agoI personally generate several Gbs of healthcare data per day. I know we generate a lot of data. I also know it's data that's so unreliable that its business value does not lie in its real-world use for improving healthcare pathways. It's very valuable politically and from a managerial standpoint, though. Unfortunately.
- ivalm 7y agoI mean it depends. There is a lot of fairly reliable discrete data in medicine: medications/labs/imaging studies/procedures/flowsheet/etc. ICD10 diagnoses are discrete and fairly reliable. The progress notes have lots of copy/paste/smart-phrase/macro-generated trash, but at least Epic saves a lot of rtf markup about the source of the text data. I am pretty optimistic about data quality, it is availability to 3rd parties that I think is limiting the AI boom.
- throwaway122378 7y agoThe fact that I could see a doctor today, see a different doctor tomorrow, give them different scenarios and both will think of it as the truth is a bigger problem that needs to be solved
- Herodotus38 7y agoOne boring truth is that so far the most notable universal and true advance AI has made in medicine (at least in the US) is in NLP with dragon dictate.
- rscho 7y agoSpoken by someone who truly knows the job :-)
- dr_dshiv 7y agoWhat do you think of m*modal? (If you've heard of them)
- ska 7y agoThe nice thing about medical reports is that a lot of factors conspire to make it an easier problem that the general case.
- Herodotus38 7y agoI have not heard of them but from reading their website looks like using AI to help charting efficiency and to squeeze more blood out of the turnip so to speak by making sure you get all those HCCs. This is the direction I’ve thought AI in medicine will go for a while: speech to text, suggested diagnosis and then suggested treatment plans. When you can get a HIPAA compliant way to record what the pt is saying and write the note (an AI scribe, so to speak), that is where I think it will start. How is m*modal, have you used it?
- ska 7y agoThis is probably true. And the big gains were made years before deep learning was a thing :)
- godelski 7y agoI'm always confused with these types of articles. "Will AI cause job loss" or "Will AI create jobs" aren't really the important questions. Because those questions are about net gain and loss, because yes it will cause job losses and yes it will create jobs. But that's not what's important to the average person. If your job is automated away, what are you going to do? Hypothetically let's say that AI can accomplish everything that an X-ray tech can (or that with the help of automation a single tech can do what 10 techs do today). What happens to those people? They spent a lot of time and money getting that training. Retraining programs suck and are shown to be really ineffective. So do these people just go underemployed the rest of their lives? What about people who are 10-15 years from retirement? Even if retrained you go from peak earnings to starting wages. That's a huge disrupt in life. Job loss is extremely important to consider (assuming you care about people), even if the total number of jobs are increased. We've seen this in the past and we're seeing it today. Lots of automation came into farming and many of those jobs disappeared. Family legacies were lost. But at the same time it is inhumane to not allow the progression of technology (like we could even stop it...). My concern is that people aren't discussing the transitions of economies. We all want to live in a post scarcity world like Star Trek. Where food, housing, and basic essentials are effectively trivially obtained (for the most part in the show). But transitioning to that kind of society is extremely disruptive and has a lot of pitfalls on the way. It isn't unimaginable to ask "What if 10% of your population is unemployable?" (several scenarios: jobs just automated away; only existing jobs require high skill and training; transition period where those people are trained for jobs that recently got automated away; etc) What do we do? How do we handle that? We're a society where people still believe your worth correlates with your wealth (diction intended). I see very few people talking about this, and these kinds of scenarios are plausible within the next 50-100 years. Net gains or losses of jobs is a distracting question and the wrong one to ask.
- DoreenMichele 7y agoI basically went through this by getting divorced after two decades as a homemaker. I spent several years homeless. I previously had a class in Homelessness and Public Policy while studying to become an urban planner. Some things that would help a whole lot: 1. We need to address our housing issues. This isn't necessarily disastrous if you can move to some little hole in the wall in a walkable area with good transit so you can still have a life while living on meager earnings and retraining. The problem is that we've torn down about a million SROs in the US and the average size of new housing has more than doubled since the 1950s, so there just aren't enough places like that. 2. We need to resolve our healthcare issues in the US. Healthcare costs something like 20% of GDP and disproportionately negatively impacts poor people, unemployed people and people with chronic health issues (who are often pushed into poverty by that fact). If you can go to a doctor no matter how poor you are, these problems are vastly less likely to snowball out of control. 3. We need to embrace gig work and figure out how to make it a positive instead of decrying it and vilifying it. When I was deathly ill and homeless, gig work was a godsend. My earning capacity gradually went up and I eventually got back into housing. I still struggle, but it's better than it used to be. Gig work allowed me to develop an earned income and marketable skills at a time and under circumstances where no regular job would have worked, yet all I seem to ever hear is how evil it is. It's not inherently evil, though certainly some gig work is handled in a problematic fashion that keeps workers trapped in a dead end.
- oarabbus_ 7y agoIt'd be a pretty good thing if AI replaced (most) doctors, actually. Not for the doctors or the insurance companies, but for most everyone else.
- oneepic 7y ago...yet. Maybe it will be 100 years in the future, but I think there's a good chance of it.
- hgjbhujxgjbv 7y agoAI is not here for doctors but for insurance companies, to charge you more if risky group is detected. There is the money and far lower risks than "helping doctors". I am glad I have forseen this step into distopia and never shared any of my data with any online bussiness (lineage, xprivacy and netguard help here).
- imvetri 7y agoMachines are inevitable. The concept of sensing and learning from surroundings should survive. Human's feed AI, AI should go interstellar. But what we will have to make sure is are we feeding the AI the right things? Nope, Tech giants, once built by nerds are now ruled by political criminal minds. Without living in harmony there is no way AI will serve only good purpose. If you are a developer, think thorough whom are you helping, why they need you. The code you write is a part of your brain. Have self righteous that code should be used for good and good only. How will you guarantee that? you cant. Then don't innovate!. Stop being the old school nerd boy and level up. Do not show off and gain attention unless you fully solve the problem. Circle back to start. Question again and again how this will be used flawlessly. If it doesn't its not worth sharing knowledge. Share morals.