I was reading an article in New Scientist, the May 13 2006 issue, p 32-38, “The Incredibles”, about the enhancement of humans by biological and technical means. It’s all about how we are able to not only combat disease and reduce premature death, but how we are increasingly able to improve healthy human beings to a superhuman state and prolong life beyond the “normal” borders. Among the abilities we can and want to add are increases in learning abilities and hookups to electronic memories.
The very nature of learning has to change to enable us to take full advantage of these techniques, however. So far, the largest part of learning a profession has been memorizing facts. For instance, learning medicine has largely consisted of learning a number of diseases and their symptoms, evolution and treatment. Continued professional education involves unlearning some of the stuff that has in the meanwhile been discovered to be false and learning some of the new stuff. An “experienced” doctor is in general the one that has seen the most medical problems first-hand and has an easier time of remembering and recognizing them the second time he sees them.
Early on, it became obvious that one person cannot keep all the required knowledge about medical conditions in his mind, so the concept of subspecialities evolved. Each speciality is concerned with a subset of conditions, only. The generalist, primary care specialist or GP (like me), is also concerned with the subset of the most common conditions, but is also tasked with determining into which bin, which subspeciality, a condition could belong that he doesn’t know enough about, so he can send the patient to the right specialist. In other words, we’ve introduced a level of indirection into the system.
The GP, however, can’t simply send the patient to any number of specialists, just to see which one recognized the disease and holds on to the patient (polling is inefficient, we know that). The GP needs to investigate the disease on his own, to a sufficient degree to actually determine more or less what it is, before he can determine which specialist will know what to do with it, so the GP needs at least partial specialist knowledge, or access to it.
In some places, like Sweden, the specialists also require that some investigations belonging to their speciality are already performed before they accept the patient, simply to improve their efficiency. This requires the GP to know what the current care plan is for that disease and speciality, of course. Any half decent IT person already recognizes duplication of knowledge in this description, which is not how we would like to build knowledge systems.
I think it’s time for some reality based examples, here, so let’s drag me and my personal experiences into this.
I trained as a general and vascular surgeon long ago, then trained for general practice and worked as a GP for 12 years. After that I did nothing but IT for ten years and now I’m back to doing both at the same time since a couple of months. At the same time, I moved from one country (Belgium) with one kind of health-care system (private) to another country (Sweden) with another kind of health-care system (state run). On top of that, medicine advanced in the ten intervening years, of course.
So, the questions are: what changed? How do I handle the changes? How do I even know what changed? How do I know what didn’t change? How different is this experience from what a fresh out-of-university young doctor has to experience?
If this entire thing had happened 20 years earlier, my handling of it had been totally different. First of all, much less would have changed, since medicine evolved slower back then. Secondly, I would have had to rely on collegues and journals or books almost exclusively. The problem with collegues is that you have to have them. Most GP practices in Sweden have two to four doctors, most of whom are too busy to chat much. (That’s what’s nice about surgery; a lot of knowledge is exchanged across the operating table. A lot of nonsense is also exchanged, of course, especially if the operation is long and tedious, but that’s another story.) To bone up on knowledge, you’d have to consume a vast quantity of litterature, chosen more or less randomly, and hope for the best. Very tiring, inefficient, and unsatisfying.
Today, we’ve got the Internet, of course. What I do, when encountered with a disease I know little about, is hit Google. From the search results, I choose the sites I trust, so I definitely filter out the stuff from newsgroups and personal homepages. I usually do get interesting hits from Medline and some government and professional sites, so the results are as reliable as any I would have gotten from a textbook, except they’re much more up to date than the textbook. Still, I try to avoid the situation where I ask the patient a couple of questions, get a confounded look on my face, turn to the computer and hit Google, since it’s liable to convey the wrong impression. In my mind’s eye, I see the patient run screaming “Oh, My God! Google!” from my office. Bad kharma. It would be better if the patient didn’t see my screen, but since I’m changing offices all the time, I can’t move the furniture around like I’d wish.
Tip: it would be really useful to have a site that preselected the sources for a google search and call it something really medically professional sounding. Maybe I’ll do it myself.
Once I know what the actual diagnosis is, or am at least almost sure, I open up viss.nu, which is Stockholm’s collection of care plans, organised according to conditions. This site tells me, briefly but to the point, the characteristics of the disease, the work-up, the differential diagnoses, and finally the treatment in sufficient detail. It even includes where to send the patient for specialist care, except I can’t use that, since I’m not in Stockholm.
It’s about time I’ll be getting to the point, isn’t it?
To enhance the ability of doctors to do a good job, we don’t need more factual knowledge of diseases, we need better access to knowledge banks, such as publications and care plans. The time for memorizing symptoms and treatments is long gone. The training we get should be focused on clinical examination methods and use of technical tools, not least of which search-engines. We need to be trained to describe what we see with the right terms and keywords and how to quickly whittle down search results to the essentials. The only thing we need to know outside of that, is whatever cannot usefully be found out from text and images. Such as how to do a skin suture or a knot. Or how to give bad news to a patient. Or how to perform a Romberg test correctly.
When I see a patient, the first step is to get to know why the patient is there at all. Then I try to get the patient’s version of what may be wrong, including any guesses she may have about the cause and what she has heard from others. This is important, since I need to respond to her fears on the one hand. On the other hand, patients are increasingly well informed and know how to use Google, so they may actually be able to tip me off to important information I do not (yet) possess myself.
This illustrates what the medical profession is about. It’s is increasingly about being able to judge factual information and to translate patient’s symptoms into correct search terms. It’s less and less about the factual information itself, since that can be had through the Internet.
Let’s take an example: if you see a description of a disease that lists symptoms such as headaches, stomach cramps and weight loss, does that mean you suffer from the disease? I mean, this list fits most anyone, since everyone has some degree of these symptoms. This is where an experienced doctor can help, since he can judge what degree of these symptoms really mean anything and which don’t. In other words, you have to have seen a lot of patients with headaches to know which ones aren’t within normal bounds.
Information technology should adapt to this change. It should center around knowledge organisation and retrieval. I should be able to, at the snap of a finger, find out what the symptoms of a disease are, the recommended work-up plan as of today and who to send the patient to, if necessary. None of these things are built-in to medical record systems as of today.
Now, back to the New Scientist article, which discussed biotechnical enhancements. If we ever get bionic doctors, enhanced by implanted memory and/or communication abilities, these should be built around search in the first place. We all need to become real Google-heads, in other words.
2 thoughts on “Google in your head”
I think in order to justify this article, you should remark on the link below. It goes right against what you have stated, so probably a few words of wisdom would be nice.
Not hard to justify. The article says that non-MDs using Google are as good as MDs (not using Google) when making diagnoses. Ok, maybe not as good, a little worse. Now, isn’t that bloody amazing? They ought not to be one tenth as good, but they are.
What *I* am claiming is that an MD using Google is one heck of a lot better than either of the above. Sadly, that’s not something they studied.
In other words, training MDs should not be based on memorizing signs and symptoms (you forget anyway) or treatments (they change), but on finding symptoms, categorizing them, naming them, using them in searches and interpreting the results. Plus, of course, acting on the received advice in a sensible manner.
Compare this to aircraft maintenance. The engineer is not expected to know every part of the plane; he’s expected to read the diagrams and act on written instructions in a sensible manner.