Can the internet replace family doctors?

Great Debates
1

Or, more to the point, can diagnosis be programmed to a greater degree of accuracy? Let’s assume that I could answer as many questions as the program needs to finalize a preliminary recommendation (urinalysis, blood tests, etc), and even offer psychiatric counseling (what to expect mentally, etc). Then when the tests were in, a final recommendation based on my file (including allergies) would be offered. I would then seek a prescription or surgery.

Note: Obviously this is not emergency treatment, but conceivably this could track disease progression more accurately and offer prevention in the most timely manner. Also, it would be educational in the extreme, perhaps assisting real doctors, and even offer lifestyle changes and monitor them while tracking one’s vital statistics and medical records over time. What are the dangers if any, besides lurking insurance companies? Does this service already exist?

2

The idea that an expert system diagnosis can be better than a doctor presumes that the system was well programmed and that it includes subtle, non-linear judgements that a good doctor will make. It also depends on the self-medicating patient can make all the same observations. It’s one thing to punch in some lab results but did you ever try to take a good look down your own throat?

3

I’m with Padeye: Garbage In, Garbage Out.

Besides, I just got back from taking Bonzo to the sports medicine guy for his Osgood Schlatter’s in his knees. How’s the Internet gonna poke a kid’s knee and say, “Yep, you got a bump there all right!” and then give him a souvenir football key chain? :smiley:

4

A nurse using a laptop and then prescribing most medications based on a database recommendation (a co-sign system)? That way we get someone to give out the soothing words and collect professional observations, and get a presciption to boot for half the time and cost.

5

Brian, even assuming you want the entire world (or at least a bunch of hackers in Kazakhstan) to know your medical situation, physical diagnosis involves a whole lot more than collecting a bunch of signs and symptoms, or even using a high-tech camera to do an “examination”.

When they come up with a magic fingers website to do rectal exams, that’s when I’m moving out to Ted Kaczynski’s old place. :wink:

Besides, don’t you think it would be more efficient to get examined and have your blood etc. drawn at the office for tests, rather than waiting your turn at home for Internet Health Professional X to get around to “see” you, then have to trudge off to Billybob’s FreeStanding Lab to donate your specimen?

6

I was assuming that a software program was doing all the diagnosis, not a person online, sorry for the misunderstanding. My goal is to spread health care around as a free/cheap prevention.

7

Medicine is still as much of an art as it is a science.

Unfortunately what we deal with many times during a diagnosis is 1) physical findings which are often ambigious or not present and 2) lab tests which are all normal. Diagnosis is then based on a patient’s complaints. A patient’s complaints are dependent in large part on what questions are asked. Good diagnosis is often based on hunches, fed (or at least not contradicted) by a patient’s appearance and perhaps by the ambigious physical findings. In order to verify the diagnosis, a doctor will ask questions in a manner which elicits certain responses. A good doctor will do this in a non-leading fashion. Herein lies the art of medicine. Any doctor knows signs and symptoms. A good doctor takes past experience into each diagnosis.

A good doctor can effectively triage patients quickly and knows when to provide urgent care.

I could give many examples of this, on both ends of the spectrum. A Hispanic woman in her mid-40s who enters the ER with pain in her abdomen which is causing her to writhe has biliary colic. Neck stiffness with photophobia and purpurae on the ankles means meningococcal meningitis. Etc.

On the other hand, some diagnoses are hard to elicit and require use of all of the senses except perhaps taste. Even one of the most severe diagnoses, that of heart attack, can often present looking exactly like esophagitis, costochondritis, or gas.

It is an art to make some of these diagnoses. Show me a computer that can smell ketoacidosis, feel a spleen tip, and most importantly take a good, thorough, caring history, and I may start to get worried about my future profession.

Lastly, we don’t trust computers to fly airplanes. This is a much more monotonous and easily automated task. In fact, IIRC, a typical airline pilot actually flies the plane for only a few minutes after takeoff and before landing. Can you provide an answer to why we don’t let computers do the whole thing? I can – Sioux City, Iowa, July 19, 1989, United Flight 232, Pilot Al Haynes.

8

Such a scheme wouldn’tbe prctical in the ner future. To begin with, the cost of the necessary tests to confirm inexpert diagnosis would be a lot more than the doctor’s fee. Physical examination is extremely important and cannot be done by a computer. Even the majority of nurses only know the very basics of physical examination which is far more subt;e and difficult than you are aware.

ANswering lots of questions is less important than answering a few key questions; often it is the way the question is answered or the expression on the person’s face as they answer it that is the key to diagnosis. Certain problems are extremely tricky to diagnose know, e.g. appendicitis; some 20% of operations performed for this have a different cause of abdominal pain. Using all the likelihood ratios you want will not lead to a more efficient algorithm. Few things in clinical medicine are studied well enough to come up with useful computer estimates.

Dr_Pap, MD, B.Eng

9

Many thanks, but what about the rest of the world that doesn’t have access to a western trained doctor, but might have one available “online” as an extensive software machine? (I estimate about 4 billion people). I am not attempting to replace an expensive visitation procedure that is affordable to many people, especially in Europe, but merely offer a safety net for everyone else. I know of two people who didn’t get to the doctor before it was too late. My wife knows of one person right now who has an inflamed skin lesion/mole that has grown large, with a previous cancerous one removed, and no insurance. She reports pain in her bones. We fear the worst. I was excluding emergency medicine anyway.

This was not intended as a replacement, but as an extension. Is it valid as an extension, perhaps? If not, how and why?

10

The idea in the OP can be summed up in two sentences:[list=1]
[li]Nice idea.[/li][li]There’s not a snowball’s chance in Hell of it ever happening[/li][/list=1]
Without wishing to insult any physicians (particularly the one who’s going to do my next rectal exam :slight_smile: ), we can say that, whilst the best diagnostic program may well be inferior to the best human diagnostician, it will likely be better than the average diagnostician, and certainly better than none at all; too, it will be much easier to clone than the human diagnostician. This is, after all, Brian’s contention. Moreover, we may say that, whilst the average person in Revoltistan may end up walking twenty klicks only to wait for an hour in a queue at the clinic to use one of three computers there, that is still better than doing so only to find nobody and nothing there.

However, there are points to be made in refutation. edwino states:

True, but basically irrelevant. A computer program (or perhaps an AI, using the term loosely, embodied in hardware) may be much better than appointing a human to perform the same task. However, the general feeling is “To err is human, but we expect perfection from machines”. We might ask what the reaction would be to someone saying, “Well, this program is 99% accurate as a diagnostician. That’s not good enough in the U.S. – we want 100% from a computer – but it’s good enough for them furriners; after all, they don’t feel pain the way we do”. Various products and services might be cited which aren’t allowed to be manufactured or sold in the U.S., but are allowed to be manufactured and sold overseas – as could be the pressure groups who indignantly demand to know how we dare flog off on the poor Third Worlders anything that we find unsuitable for our own use. How long before the demonstrators gather on Pennsylvania Avenue, and the websites go up urgently requesting that you write to your Congresscritter and demand that all funds be cut off for disseminating and supporting TIRESIAS V2.0? About half an hour, I’d guess.

(BTW, TIRESIAS was an actual program, written about twenty years or so ago, that did something along the lines of what Brian had in mind. Does anyone know what happened to it?)

11

Apparently you are under the misconception that putting something on the Internet is “free”. Have you checked out DrKoop.com? Last time I checked they were well on their way to being delisted from the NASDAQ.

12

to have an effective computer diagnostician, it would have to be a true AI in addition to having an understanding of human intelligence and behavior. When I talk to the patient, body language and tone of voice can tell me far more than the answers to my questions. And there’s no substitute for me getting to know my patients as people, and be able to note changes in mannerisms over time. At times, due to sheer hunches based on my knowing the patient, I’ve come up with real off-the-wall diagnoses that happened to be correct.

And physical exam is very important too, despite old sayings that a good history will give you the diagnosis 90% of the time. Listening to the lungs to hear how much they’re wheezing, poking the toes to see how much neuropathy they have, are really helpful for both acute and chronic care.

13

The problem with hooking up 4 billion Third World people to an Internet-wide medical diagnostic service, from just a logistical standpoint, is two-fold.

  1. Where are they supposed to get the computers, the telephone lines or satellite service to access them, the electricity to run the whole thing?

  2. What are they supposed to do with the diagnosis once they have it? They can’t run down to Walgreens to have a prescription filled. There’s no doctor in their village to implement the Internet MD’s suggestions (that’s why they’re looking it up on the Internet in the first place). So what are they supposed to do with this information? “Mr. Ahmad, you are suffering from biliary colic. You have gallstones.” Okay, so now what?