The Decision Tree

thedecsiontree.jpgLast May I had the opportunity to talk a while with Wired Magazine's Thomas Goetz about the idea of how people can take control their over own health care using the tools and data available on the Internet. His new book on the subject, entitled The Decision Tree, is a step above most health improvement books in terms of the scholarship/readability (i.e. it's based on good science and it's easy for me to understand.)

The big idea is this: A person's health doesn't happen all at once; it's a consequence of years of choices - some large and some small, some good and some bad. His book looks at the choices that advances in genomics, self-monitoring, new screening techniques, and collaborative health tools are giving the average patient. The trouble is, there's so much information available that it's really, really hard to interpret it all. What to do? According to Goetz, the answer is to make a decision tree.

decision tree example 2_Vga.jpg

Decision trees or flowcharts that make all of these decisions more visible and more obviously something we are actually choosing. Unfortunately, most current decision trees look like the one to the right: technical and hard to understand

But where we are apparently headed is in the direction of interactive ones like this one at Preventative It really makes the tradeoffs clear: If I do this (e.g. take a baby aspirin daily) I can expect to add X days to my life. For me, the daily aspirin adds a probable 293 day to life span - why wouldn't I do that?

The test and interface is simple. In fact, I wish there were a lot more factors to play with (e.g. how many days of life, if any, would a daily glass of red wine add, etc?) I eagerly await the day that an organization that I trust puts up a decision tree website like that with a lot more factors (daily alcohol intake, quantity of fruit eaten per day, basement radon test, etc.)



  1. Argh argh argh. Probabilities and risks for populations do not translate as hard numbers of days to your own personal individual life. The “preventive math” site tells me that getting screened for alcohol abuse will add 40 days to my life. No. It won’t.

    1. The “preventive math” site tells me that getting screened for alcohol abuse will add 40 days to my life. No. It won’t.

      Crap. There goes my plan to get screened for alcohol abuse every 39 days.

  2. I think this site gives one specific piece of bad advice. It advised me to take asprin, and (unless you’ve been diagnosed with specific sort of heart problem) recent studies seem to suggest that the risks outweigh the advantages, just.

  3. Just live your life and enjoy everything it has to offer in moderation.

    You will die eventually, no matter.

    Joggers get run down every day. Keith Richards is still going strong.

  4. “The trouble is, there’s so much information available that it’s really, really hard to interpret it all. What to do?”

    Go see a doctor, someone who is actually trained to interpret all this information for you.

    1. Exactly, bzkrt. Anyone can put anything they believe on the intertubes as well. There is plenty of misinformation about all kinds of illnesses and treatments, from holistic to Western. Generally you can tell someone’s “personal belief” blog from, say, WebMD, but a lot of people don’t bother to question whether the writer is a medical professional or simply someone posting a regimen or treatment that simply worked for them.

    2. You should see a doctor, but there is absolutely nothing wrong with being a little skeptical of the advice of doctors and their advice. Doctors fall into traps of anecdotes, bad statistics, and acceptance of common wisdom all the time.

      A famous example of how badly doctors can misunderstand their own profession comes in the accuracy of tests. Consider this famous example. You have a white heterosexual male who does not abuse drugs. He tests positive for HIV. The test gives false positives 1% of the time. What is the likelihood that the subject in question has HIV. If your doctor answers 99%, he doesn’t understand statistics.

      I’m not saying don’t listen to doctors. I am saying that a little skepticism is healthy. Doctors are people too,and like most people make mistakes, make bad assumptions, and do all of the normal stuff that people do.

  5. “…daily aspirin adds a probable 293 day to life span – why wouldn’t I do that?”
    1) Daily Aspirin may attack your stomach walls increasing the risk of ulcers/cancer
    2) The bood thinning effect of Aspirin, something that’s good if you have a heart condition, may increase the impact of a stroke by preventing the bleeding to stop or the risk of dying during an operation.

    The point being: just because something is really good for certain people and maybe moderately bad for other doesnt mean that it’s moderately good for everybody…

  6. @bzkrt: The problem is that doctors really aren’t particularly good at interpreting all that information, either. The field is too broad for a generalist, and specialists work from within their narrow range of experience. They’re prone to give too much weight to recent studies and not enough to a complete body of literature. They aren’t well trained in interpreting research. They over-treat. They suffer from confirmation bias. They give in to crazy patient demands.

    Worse, you as a consumer, have no way of knowing if your doctor is good or not. Tell me, how do your doctor’s patient outcomes differ from mine? No one tracks that.

    Then there’s the desire on the part of doctors to be considered a little magical and artistic. A holdover from before science-based medicine. It’s natural for a guy who has been practicing for a decade to build up a library of anecdotes that they use to justify treatments that aren’t backed up by good evidence. How often have you heard “many of my patients have had success with…”? That’s a doctor making a recommendation based on his experience with a handful of people. It’s a study with n=13 and a confidence interval of “I’m pretty sure”. That’s not science.

    A decision tree for medicine makes great sense in that at least you’ve got a baseline model of care established. Medical professionals can deviate from it, but they should have a real reason. You, as a patient, should have access to whatever the play book is and should be able to say to your friendly MD: “dude, why are you recommending surgery before physical therapy?”

  7. Using decision tree is a great way to search for root causes. However health, and the physical processes that define it, is a complex system and acts multicollinear.

    It’s our western preference to approach issues like this in a deterministic way. Unfortunately this leads to blindsight or suboptimalisation at best.

    See also how this can effect cancer treatment at:

Comments are closed.