1. Hi there, Guest

    Only registered users can really experience what DLP has to offer. Many forums are only accessible if you have an account. Why don't you register?
    Dismiss Notice
  2. Want some serious feedback about your writing? Enter the Q3 2019 Story Competition!

    Topic(s): EITHER
    - What the Professors get up to in the summers.
    - Ritual Magic!

    Word count: 17.5k max (no minimum)
    Deadline: September 9th
    Check out the Competition Page
    Dismiss Notice
  3. The Q3 2019 Story Competition has 18 DAYS LEFT! (September 9th)

    Check out the Competition Page and get writing folks!
    Dismiss Notice

Doctor rediscovers integrations; gets 75 citations

Discussion in 'Real Life Discussion' started by Lyndon Eye, Jan 5, 2011.

  1. Lyndon Eye

    Lyndon Eye Minister of Magic DLP Supporter

    Joined:
    Jul 4, 2007
    Messages:
    1,351
    Location:
    Germany
  2. Giovanni

    Giovanni God of Scotch

    Joined:
    Aug 4, 2005
    Messages:
    8,637
    Location:
    Gilligan's Island
    You mean doctors don't know math? No one who reads the Lancet or New England Journal of Medicine could have seen that co . . .

    Shit. Nevermind.
     
  3. Perspicacity

    Perspicacity Destroyer of Worlds ~ Prestige ~ DLP Supporter DLP Gold Supporter

    Joined:
    Nov 27, 2007
    Messages:
    985
    Location:
    Where idiots are not legally permitted to vote
    High Score:
    3,994
    If it's any consolation, Google Scholar lists it as having 136 cites. For a 1994 article, it's not obscenely many.

    Also, before extrapolating to a "doctors are stupid" conclusion, it's worth noting that several of Dr. Tai's colleagues have called attention to the error of letting such a trivial result be published in the literature. Indeed, a large number of the citations and follow-ups have been comments on exactly this (http://www.ncbi.nlm.nih.gov/pubmed/8137688?dopt=Abstract).
     
  4. Giovanni

    Giovanni God of Scotch

    Joined:
    Aug 4, 2005
    Messages:
    8,637
    Location:
    Gilligan's Island
    Pers you take all the fun out of life. With that said, my extrapolation was "doctors suck at quantitative analysis," and that conclusion has been gaining traction for a few years now ;)
     
  5. Perspicacity

    Perspicacity Destroyer of Worlds ~ Prestige ~ DLP Supporter DLP Gold Supporter

    Joined:
    Nov 27, 2007
    Messages:
    985
    Location:
    Where idiots are not legally permitted to vote
    High Score:
    3,994
    Let me revise my statement: Most doctors suck at quantitative analysis. ;)
     
  6. Giovanni

    Giovanni God of Scotch

    Joined:
    Aug 4, 2005
    Messages:
    8,637
    Location:
    Gilligan's Island
    They really are bad at it, at least the ones in the medical community ;)
     
  7. Link

    Link Order Member DLP Supporter

    Joined:
    Aug 28, 2006
    Messages:
    845
    Location:
    China
    It's funny because it's true. I've got something like 30 hours of mathematics in my first year of medical school - and we have covered briefly (in 2 hours IIRC) integrations, though half of those maths lessons were on probabilities and statistics.

    There were only a few questions in the exam relating to this particular matter, so failing it did not necessarily mean you failed the exam - which means that lots of students don't understand well integration, and some will probably forget all about it by the end of the following year.
     
  8. Xiph0

    Xiph0 Administrator Admin

    Joined:
    Dec 7, 2005
    Messages:
    9,087
    Gender:
    Male
    Location:
    People's Republic of California
    Ironically, this reminds me of Chem majors; they're more focused on formulas than they are on the theory or proof behind any of it. It's kind of disheartening; it's worse when they get better grades XD
     
  9. Tenages

    Tenages Order Member DLP Supporter

    Joined:
    Aug 19, 2010
    Messages:
    803
    Location:
    Philadelphia, USA
    Sadly true. Not all of us do, but a great many are this way. The ironic part of this is that Chem TA's, prof's and even us undergrads bitch amongst ourselves about how the biology students in chem classes just want to memorize the necessary information without understanding the theory.
     
  10. Coelacanth

    Coelacanth Denarii Host

    Joined:
    Sep 29, 2008
    Messages:
    333
    Location:
    New York, Ny
    This is new how? A lot of pre-med and med students have only a basic understanding of quantification. If at any level what-so-ever.
     
  11. Taure

    Taure Magical Core Enthusiast ~ Prestige ~ DLP Supporter

    Joined:
    Mar 5, 2006
    Messages:
    1,673
    Location:
    United Kingdom
    High Score:
    13,152
    Surely the necessary information is the theory?

    Anyway, I had the chance recently to ask a large number of medical students at my university about the amount of mathematics they do. In particular, I was interested in their understanding statistics and Bayes theorem, which I think most people would agree is something that all doctors should know, given that it's directly relevant to their subject and job.

    I was shocked to discover that, despite there being numerous studies showing medical students not understanding basic statistics*, medical schools still haven't responded by teaching it. Med students - at least in the UK - still don't get taught Bayes.

    It's very odd. I presented to them, informally, the following problem:

    1 in a million people have disease X. A population is tested for the disease. The test is 90% accurate. Given a positive test result, what is the likelihood that the person has the disease?

    Most answered 90%.

    But the answer is:

    0.0009%

    (Here's why).

    It was a bit worrying.

    Edit: It's really bad that I feel dirty every time I even say "Bayes". Thanks, Less Wrong.

    *E.g. Casscells, Schoenberger, and Grayboys 1978; Eddy 1982; Gigerenzer and Hoffrage 1995. Stolen straight from Less Wrong's site.
     
    Last edited: Jan 6, 2011
  12. Random Shinobi

    Random Shinobi Unspeakable DLP Supporter

    Joined:
    Aug 1, 2006
    Messages:
    716
    Um... You know, the reason why they answered that way is most likely the way you worded your question. Nobody would use a test with about 10% chance of a false positive (at least, not to test big masses as you described), and so they simply assumed you meant that 10% of the positive results are false.
     
  13. Taure

    Taure Magical Core Enthusiast ~ Prestige ~ DLP Supporter

    Joined:
    Mar 5, 2006
    Messages:
    1,673
    Location:
    United Kingdom
    High Score:
    13,152
    We have a saying here in England. It's "R.T.F.Q.". Not sure if you have it in the US.

    It means "Read The Fucking Question".

    I'm not sure if claiming that doctors lack basic reading comprehension is a great defence for lacking knowledge of statistics. "90% accurate" is pretty clear. 90% of the time the test will give the correct result, 10% of the time it will give the incorrect result.

    There isn't much room for ambiguity there.

    And anyway, the studies I mentioned have shown that the result can be consistently replicated, even when questions are phrased differently (though there is a slight increase in competence when dealing with proportions instead of percentages).
     
  14. World

    World Oberstgruppenführer Moderator DLP Supporter

    Joined:
    Apr 19, 2006
    Messages:
    3,212
    Location:
    Axis of Evil (Original)
    I can't believe I only just remembered this, but ...

    [​IMG]
     
  15. Random Shinobi

    Random Shinobi Unspeakable DLP Supporter

    Joined:
    Aug 1, 2006
    Messages:
    716
    There is. Your question overtly simplifies matters and that creates confusion. You are assuming that the likelihood of a sick person getting a negative result is same as a healthy person getting a positive result. In reality it is not so.
     
  16. Taure

    Taure Magical Core Enthusiast ~ Prestige ~ DLP Supporter

    Joined:
    Mar 5, 2006
    Messages:
    1,673
    Location:
    United Kingdom
    High Score:
    13,152
    No I'm not.

    Likelihood of "sick person" & "negative result": (1/1,000,000) * (1/10)

    Likelihood of "healthy person" & "positive result": (9,999,999/1,000,000) * (1/10)

    ...looks like you need to brush up on statistics too.

    Even if it had been the case, you're simply reiterating the idea that doctors (and apparently you) don't understand mathematics. When faced with a mathematics problem, you don't consider the likelihood of the statistics you have and then change the scenario to what you think is more likely, then perform the calcuations upon that. You just do the calcuations on the numbers you're given.

    Honestly, your explanations for the results make doctors look worse than if they simply hadn't been taught Bayes. In the latter case they simply lack an item of knowledge. In the former they're just retarded.

    And I find your explanation unlikely on top of that, from a psychological perspective. I think most doctors taking the test would understand that this is an abstract mathematics problem and that their job is not to judge whether or not it's likely that a 90% effective test would be used in real life. I find it likely that they would have given the same answer even if the test had been 99% effective, or 99.99% effective (adjusted, of course, for the change in accuracy). What their answers show is a lack of understanding of the role of base rates/priors and how new information acts as a modification upon that base rate.
     
    Last edited: Jan 6, 2011
  17. Perspicacity

    Perspicacity Destroyer of Worlds ~ Prestige ~ DLP Supporter DLP Gold Supporter

    Joined:
    Nov 27, 2007
    Messages:
    985
    Location:
    Where idiots are not legally permitted to vote
    High Score:
    3,994
    I don't believe that with restatement Taure's question would turn out all that much differently. (This is a result of practical importance to me--I just had to reinstate my Sigma 15 yesterday and make myself eligible for polygraph witch-hunting, which has a similar problem of false positives and which has never in the history of the country identified a spy).

    True story: A colleague and friend of mine at the Lab got prostate cancer. He did what many of us physicists do when we get something serious and are faced with questions about what treatment option to pursue, none of which are pleasant: We read the available literature and we figure out for ourselves whether the flow chart our physician memorized makes any sense.

    What he found appalled him and having heard his talk on this, I'm similarly disturbed by the shoddiness of the canonical studies in the medical field. My friend has since developed his own model for how the disease progresses under various therapy options and has run his model on himself to maximize his own situation. He wrote up his work and published it in the NEJM and he has given his talk to thousands of physicians. His model (which at its heart is simply a set of coupled first order ordinary differential equations, the kind of thing any of us can solve in our sleep); it's the application of the simple mathematics with transparent reasoning and clearly stated hypotheses that apparently has been missed by the medical community.

    What gets me about medicine is that bulk of the research is conducted by those who follow their names with "M.D." rather than "Ph.D." and who really have no substantial formal training in science or scientific methodology. They "seat of their pants" their statistics and draw conclusions, often erroneous and unsupported by the data, as Taure has illustrated with his survey.

    If you want to go into medical research (in the U.S., anyway), you're encouraged to go to medical school, where you are trained to treat disease, not study it. You're not asked, as you would be in a graduate program in science, to conduct your own research under the watchful eye of a research mentor. You're not trained in how not to fool yourself or publish garbage. You're not imbued in the culture of scientific ethics that prevents things like this fucker driving his Benz and sleeping at night knowing his malfeasance has killed children.
     
    Last edited: Jan 6, 2011
  18. Random Shinobi

    Random Shinobi Unspeakable DLP Supporter

    Joined:
    Aug 1, 2006
    Messages:
    716
    Huh... Apparently something was lost in translation, or perhaps you are being deliberately obstute. Anyway, what I meant to say is that if a sick person goes to that test, he has 10% chance of getting a wrong result and, similarly, if a healthy person goes to the test, he too has 10% chance of getting the wrong result. No real tests work like that.

    But whatever, I'm likely just having too much faith in humanity...

    EDIT: I really have no idea how that smiley got there...
     
  19. World

    World Oberstgruppenführer Moderator DLP Supporter

    Joined:
    Apr 19, 2006
    Messages:
    3,212
    Location:
    Axis of Evil (Original)
    Many a test has a chance of producing a wrong result - pregnancy tests for example.

    Once I thought about it (i.e. after reading the explanation), I find Taure is entirely correct, and the question reasonably defined, even if it does lead you to a wrong conclusion if you don't think about it or have no idea about statistics.

    Many more healthy people's test will result in a positive than sick people's, simply because there are way more healthy people than sick ones.
     
  20. Tenages

    Tenages Order Member DLP Supporter

    Joined:
    Aug 19, 2010
    Messages:
    803
    Location:
    Philadelphia, USA
    One would think so. But at least at the lower levels of undergrad chem (which is all that bio and pre-med students are required to take) it's very possible to just memorize formulas and lists of interaction results without gaining any understanding of why these formulas work or how these interactions take place. In fact the brute force memorization method is the most popular way to learn intro organic chemistry. The irony here is that with a competent teacher, for most people it would be much easier to learn the theory and apply it rather than memorize every reaction mechanism, etc. that you need to know for the class.

    So in the larger sense of being able to apply what you learned, and use in in real world scenarios, yes the theory is the necessary information. But in the sense of what's necessary in order to pass the chemistry class memorization of formulas can suffice.
     
    Last edited: Jan 6, 2011
Loading...