Biomedical Engineering Reference
In-Depth Information
L
AW NUMBER
1
When stating a value associated with a measurement or a
calculation, one
must
:
(a) provide an estimate of the uncertainty (e.g.,
through specifying a confidence interval); and
(b) specify the level of confidence associated with
that interval (e.g.,
±
1SD, 95% etc.).
There is simply no excuse for violating either part of Law number 1.
The uncertainty estimate may be generic, based on past experience
with similar problems; it may be a rough “back-of-the-envelope”
calculation; or it may be the result of a detailed analysis of the
particular measurement. Sometimes it will be sufficient to provide an
umbrella statement such as “all doses have an associated confidence
interval of
±
2% (SD) unless otherwise noted.” In any event, the un-
certainty estimate should never be implicit; it should be stated.
In graphical displays such as that of a dose distribution in a two-
dimensional plane, the display of uncertainty can be quite challenging.
This is for two reasons. First, it imposes an additional dimension of
information which must somehow be graphically presented. And
second because, in the case, for example, of the value of the dose at
a point, the uncertainty may be expressed as either a numerical un-
certainty in the dose value, or as a positional uncertainty in terms of
the distance of closest approach. One approach to the display of dose
uncertainty is shown in Figure 6.4 of Chapter 6.
H
OW TO
D
EAL WITH
U
NCERTAINTY
To act in the face of uncertainty is to accept risk
. Of course, deciding
not to act is also an action, and equally involves risk. One's decision
as to what action to take, or not to take, should be based on the
probability of a given consequence of the action and the importance
of that consequence. In medical practice, it is particularly important
that the importance assigned to a particular consequence is that of the
patient, and not his or her physician. I know a clinician who makes
major changes in his therapeutic strategy because of what I consider
to be a trivial cosmetic problem. Of course, some patients might not
find it trivial at all. So, since he assumes that all patients share his
