Disclaimer - by "Doctor" I mean Family Practitioners/General Practice doctors. I don't mean Surgeons, ER doctors, researchers, etc.

I'm wondering how long until we replace Doctors with a good Diagnostic AI program. Most GPs gather a bunch of symptoms, use their training and memory to determine a likely cause, and then they determine the best treatment for that likely cause.

We can do all of that a lot more effectively with a computer program and a much less trained person whose job is just to get symptoms and input them into the computer. Why are we still training doctors for like 9 years when a good chunk of that time is just spent memorizing things that they don't need to memorize anymore?

In the past 5 years I've gone to the doctor for 3 separate issues. The first was a simple issue where I knew what was wrong and just needed a prescription. The second issue they ran a bunch of tests but never figured out what was wrong and after a month or two it just went away. The third issue I've been to the Doctor twice and both times I was told exactly what the problem was and both times the doctor was wrong (they didn't agree).

I'd say we're still decades (30+) away from having a machine we stand in or lay down in and it makes a decision as to what medicines we need to take to alleviate a pain or if we need to go for surgery etc. Plus this machine would have to be globally connected with your EMR so it can track your body over a long period of time so it can recognize when you're gaining weight, the rate of change of heart rate, blood pressure etc. Instantaneous readings would be meaningless (unless its something obvious) without a built up baseline. So even when these things come out, only the young will truly benefit.

And to take it a BBV4L route - would you really want a machine to stick its finger up your butt to check your prostate? :P

I'm not asking for when we don't need people, I'm asking for when we don't need heavily trained and expensive Doctors. Taking your temperature, asking you questions, measuring your blood pressure, etc. are all things that a minimally trained and paid person could do. They then enter the data into the computer and we get a result.

We sort of have them already. Nurse Practitioners and Physician Assistants. They still school for a longer period of time, but not an MD length of time.

They are more highly paid then regular nurses but less then Doctors in their fields. Its all about being qualified to write prescriptions and perform "simple" yet invasive "operations"

I keep editing this.

The computers input has to be accurate - the only way to remove the human component of the error is to have the computer do the entire reading. Could an MD make a mistake? Sure, but it would hurt him a lot more to make that mistake then it would someone making 1/3rd the pay.

I feel like this is where we should be right now.

Obv a big problem with this is just that a lot of people want to see a doctor, not a computer. That's probably irrational, but it's definitely true.

Agreed. But I don't and there have to be a lot of people like me. I have way more faith in a tested and validated classification system then in a person who needs to know and remember thousands and tens of thousands of pieces of information (plus constantly learning new pieces of information).

Also the fact that I'm not typically presented with a range of possible diagnosis with an approximate probability behind each one is kind of ridiculous to me. I guess lots of people prefer being told something (even if its wrong) then hearing that no one knows for sure what the problem is.

Me too, but why aren't we then?

I don't buy that. If anything you could take one doctor and replace him/her with two people making 1/3rd the pay and then have double redundancy on measuring symptoms while still saving money.

Besides, my understanding is that the current difficulty isn't in determining symptoms - its combining them into a reasonable diagnosis.

When will we stop needing doctors? after 2012 with the collective consciousness shift.

You're obviously well versed in the topic at hand.

They are called nurses, and then they tell the doctor.

I don't understand what the problem is? Can't you measure these things yourself, and look up a chart online?

http://en.wikipedia.org/wiki/The_Age...itual_Machines

The real question is, "How long before we can copy someone's brain into a computer?"

The problem is people do this with WebMD already, and the diagnosis is almost always ridiculous. Medicine at its core is a humanities profession shrouded in science (not discounting the science aspect; it's huge), and you need to be able to talk to someone who can objectively observe your circumstances and diagnose. You will undoubtedly have biases in your diagnosis of yourself (often they are ultimately overblown and err on the side of "I'm going to die"), so why leave it to yourself when you can go to someone who has trained intensely for years to do what you're attempting to do?

I think we're a long time from not needing GPs. The current shortage in them is due more to reductions in Medicare reimbursements, and the growing cost of student loans than the "death" of the field. You can put a list of symptoms into a computer, and a diagnosis might pop out, but odds are you'll get a multitude of diagnoses ranging from the mundane to the life-threatening (see: your local ER where people come in having convinced themselves that they are dying via WebMD). You need some way to pick through these, and a lot of it comes down to putting the symptoms in the context of the patient which a computer can't do.

If I had gone into Family Medicine, this was going to be my plan. Primary care has been absolutely hammered by managed care so that some FM doctors have gone to crazy lengths to provide additional services in order to try to make more money. Others pack their schedule to the gills and work 12 hour days. Between the decreased pay and the increasing specialization of medicine, it only made sense to me that this area of medicine could be better served by less extensively trained, cheaper care providers.

My dream was to assemble a team (funny, but I never bothered to check how many NPs could work under the supervision of 1 doctor, but I assume it's at least 3) of nurse practitioners and basically supervise their care, "consult" on difficult cases and care for a handful of patients myself. There's probably a problem I hadn't considered but it looked like a win-win-win situation to me.

who says they haven't done it already? then again time is a confusing subject in itself.

I'm not sure if House would be better or worse with a robot main character

Long time.

Lawyers gotta eat, you know.

MM MD

While it's true that the same symptoms could be caused by many different illnesses, that doesn't mean a computer can't fairly accurately depict a percentage range of likelihood breaking down what is causing the problem. Most people who do have a serious condition don't go to a doctor and tell them their symptoms and are correctly diagnosed. They are usually misdiagnosed with something less severe to begin with (because this is more likely to be the correct diagnosis), and only when the problem persists do they return to a doctor and is the problem looked into in more depth.

The same would be the case with a computer. They would diagnose the most likely problem, print out a prescription and the patient would be on their way. If they don't get better then they would schedule an appointment to see a
"real" doctor.

The computer could also be programmed to refer the patient to a real doctor straight away (possibly in another room) if there is any reasonable level of doubt as to what the problem is. Now what a "reasonable level" is obviously needs to be decided, but I do think this idea has a lot of merit to it and should be implemented in some way already.

Even if they had a computer in the waiting room where the patients would enter their symptoms into so that when they went to see the doctor they already had a fairly accurate idea of what was wrong - they could double check to make sure that the computer's diagnosis is in line with what he is seeing and if it is, print prescription, and on your way you go.

Another way this would be very effective is in the case of immunisations. As it stands (at least in Australia), if you are going to travel overseas and need to be immunised, you have to go to your doctor and ask him what you need. He then prints a prescription, you go to the chemist, buy the injection, go back to the doctor, and he administers it. This seems like a colossal waste of time for all involved. A patient should be able to fill out a questionnaire on a computer regarding where they are traveling to, what they intend to do there, any other medications that they are currently on, and the computer should be able to look up the patients history and print out the prescription(s) for whatever immunisations they might need.

While I agree that we're a long way away from getting rid of GPs entirely (if that ever happens), there do seem to be a multitude of ways their practices could be more efficiently run with the use of computers.

when robots can peel a membrane off the hymen of an ant

This is a terrible oversimplification of what doctors do. I'm sorry that your past GPs haven't been as competent as they should be.

I imagine this is a scam of some sort, nobody is interested in saving your time here.

I would guess (knowing very little about it) 75% of a GP's work could be replaced by a computer and a less-trained NP but hardly anyone has a vested interest in making that happen.

But when you have to fire two people and retrain two of them, you've lost a lot of man hours. The idea is to push volume through on these two people. So if you lose them because of errors you lose a lot of volume that has to be picked up by someone else - and you aren't going to hire two bums off the street - you'll need educated people that additionally have to go through a week or so of training, at least a few more adapting to your work place etc. All the while the volume has gone down, reducing the margins.

Sometimes, one highly trained person making very limited mistakes and does a volume of X is just better to have then 10 limited trained people making more mistakes and having a volume of X*3. Many companies are discovering this now after they outsourced IT jobs to India. What are they doing? Pulling it back in house purely because of the loss of customer service.

What makes you think the computer result would be anywhere close to the human result, much less close to optimal? Go enter "fever" and "aches" into WebMD and tell me what you have. Or try even describing "aches" to a computer. There are just too many human nuances and subtleties in both symptoms and human descriptions of those symptoms to make me think we're anywhere close to getting rid of GPs.

Also, did you really not know that nurse practitioners existed, and only MDs could write scripts?

I'm actually really interested in trying Web MD but it isn't working on my computer. Are there any websites similar to it where you enter in symptoms and it attempts to diagnose you?