Biomedical Engineering Reference
In-Depth Information
Shochat and colleagues (2) showed how
strong physiological perturbations
induced by medical therapy can generate data that can yield both quantitative
and qualitative insights, provided the data are compared with predictions of a
suitable quantitative model
. Another example is the comparison in (3) of viral
counts with model predictions following administration of a protease inhibitor to
HIV patients. This and related work established that, although progression to
AIDS may take many years, viral turnover is nonetheless rapid, mandating mul-
tiple-drug therapy to avoid fast viral evolution to drug resistance.
3.
EMPLOYING INFORMATION ON PROGRESS TOWARD
MULTIPLE GOALS TO REGULATE THE IMMUNE RESPONSE
3.1. Multiple Goals
Together with R. Bar-Or, I have argued that it is useful to regard the im-
mune system as simultaneously pursuing a variety of overlapping and conflict-
ing goals (4) (see also this volume, Part II, chapter 5, by Krakauer). Examples of
such goals are avoiding harm to self, killing dangerous pathogens, and acting
quickly against such pathogens, but acting economically. In the face of ever-
shifting challenges, the immune system monitors how it is doing with respect to
these goals by means of various sensors and uses the information thereby ob-
tained to shift its actions in order to improve performance with respect to the
variety of goals. In view of the information gleaned, performance can be im-
proved by selecting suitable effector classes for expansion and others for con-
traction (Th1 vs. Th2, choice of isotypes). Moreover, within a given effector
class, sensor-based information can be used to improve performance (when
should a given class be deployed, when expanded, when contracted).
To gain more insight concerning this role of sensors, see (5) for an exami-
nation of how partial information from sensors can improve performance in a
"team" version of the game Connect Four. A team that can gather and utilize
information even slightly more effectively tends to win. If one thinks of a team
as a model complex organism, then the results of many contests will lead to the
evolution of organisms with gradually improving information capabilities. By
monitoring what sensors give superior performance, one is led to the possible
characterization of performance in terms of goals, that is, to obtain the ultimate
goal of Four in a Row (analogous to survival of the organism) there are a variety
of intermediate goals, represented, for example, by board configurations that are
worth striving for.
3.2. A Model Immune system: Conflicting Roles of a Noxious Chemical
Here is a sketch of a simple model that illustrates some of the issues. Con-
sider a population of pathogens that in isolation grows exponentially at rate
r
.
Imagine an effector cell
E
of the immune system that kills pathogens
P
at a rate
kEP
(mass action assumption), where
k
depends on various factors. Suppose in
