Wow -
Ive been lurking here all this time, too busy writing grants to be
composing messages, but Ill get a good summary of SyM Bowl 97 on here
shortly. It was fantastic! Many models were significantly more mature
than last year; details to follow.
However, I do find it interesting to be eavesdropping on two fascinating
discussions relating to my personal uncertainties/questions.
The first is an attempt to sort out the interactions and
advantages/disadvantages of differential equations versus SD (?; if
"versus" is the right word). Naturally JWF summarized in about 4
sentences what I had been trying to ask last year in multiple messages and
hundreds of words . But then, hes been thinking about it a little longer
than I have . . .
And the second is the ongoing discussion about the relative merits of CLD
vs S&F diagrams.
I find these to be fascinating topics, with ongoing controversial opinions
expressed, but I think these are critical to the acceptance and broader
dissemination of SD into other disciplines, especially in view of the
recent comments about whether SD (or any other learning technique) is
allowed to be "fun" or not, and whether STELLA is "just a toy" for high
school students.
One thing that is becoming clear to me is that in many many situations, SD
is the natural tool for approaching a problem, much much easier than diff
equations, and for this reason, more powerful rather than less.
I have a recent book on PK/PD modeling.
Pharmacokinetics (PK) is the study of dosage regimens, absorption,
distribution, metablism, excretion, time course of drug action, etc. i.e.
"what the body does to the drug". PK is not concerned with the effect of
the drug per se.
Pharmacodynamics (PD) is the study of how the drug interacts with the
biological material (enzymes, receptors, etc.) to produce its biological
effects. (Binding equilibria, affinity, efficacy, antagonism, subsequent
mechanisms of actions, etc.), i.e. "what the drug does to the body".
As you might guess, PK/PD modeling combines these two topics to produce a
model that describes the drug effect as a function of time after a given
dose. This is a very important conceptual issue, and the mathematics is
absolutely hideous. In fact, one of the discussions in the book includes
the relative merits of "realistic" models, which are very difficult to
develop with diff eq; versus mathematical models which use such tricks as
spline curves to fit the data, even though it is recognized that this
technique does not provide any underlying structure for the behavior of the
system. Anyone involved with SD would immediately recognize this latter
type of curve fitting to be so intellectually barren as to be a joke, but
nevertheless, its in the book!
No one (yet) has ever published the suggestion that SD would be the ideal
tool here.
SD essentially removes the math requirements; simple algebra is quite
adequate for these problems.
SD DOES NOT REMOVE THE REQUIREMENT FOR RIGOROUS, DISCIPLINED, QUANTITATIVE
THINKING! Anyone who says this is a "toy for high school students" is
ignorant beyond words. The development of a basic set of PK/PD models
would be the *starting point* for some very innovative and progressive
medical research.
This leaves me in a quandary; PK/PD models are so difficult that they are
seen as very remote and arcane by the vast majority of the biomedical
community. And yet I am developing PK/PD models that can clearly be taken
into the high school biology/chemistry/health class. This is not the first
time I have experienced this "dissonance" between what we -expect- high
school students to be studying, and what SD -enables- them to study.
This relates directly to JWFs recent comment that Roadmaps is suitable for
high school on up. Isnt that interesting? The basics can be taught in
high school, but the graduate level professional -needs to learn the same
material-, and can use the same set of tutorials! This clearly says to me
that SD -enables- rigorous, disciplined thinking as soon as reasonable
algebra skills are available, and after that, the sky is the limit.
PK/PD models are far, far outside the standard biology curriculum, partly
because the layers and layers of math required to get there push this topic
into graduate school.
If I were to tell my colleagues that I was teaching PK/PD models to high
school students they would think I had taken leave of my senses! But it
can, and should, and will, be done.
When some of these students get to medical school I expect them to make a
naive comment to their instructor, "Why are you doing all that? Why
dont you just use a PK/PD model?"
Ed Gallaher
From: Ed Gallaher <gallaher@teleport.com>
Associate Professor
Physiology/Pharmacology and Behavioral Neuroscience
Oregon Health Sciences University