Neural computing in pharmaceutical products and process development - Computer-Aided Applications in Pharmaceutical Technology - page 124

The neuro-fuzzy system was used to study tablet fi lm coating formulation
(Rowe and Woolgar, 1999). The data base used was for a fi lm formulation
consisting of hydroxypropyl methylcellulose of varying molecular weight,
containing either anatase or rutile titanium dioxide as an opacifi er of
varying mean particle size and particle size distribution at varying
concentrations applied to give fi lm thicknesses varying between 10 and
150 µm. The cracking propensity of each formulation was assessed by
calculating its crack velocity and opacity.
Some of the fuzzy rules generated are:

■ IF pigment size is small AND pigment concentration is high AND fi lm
thickness is thin THEN fi lm opacity is low (0.28) OR fi lm opacity is
high (0.72).
■ IF pigment size is small AND particle size distribution is small AND
pigment concentration is medium THEN crack velocity is slow (0.90)
OR crack velocity is fast (0.10).

As can be seen from these examples, fuzzy logic is ideal for
applications in process optimization and control, since it enables
generation of easily understandable and adaptable rules. This way,
process monitoring and control is improved, since it is possible to defi ne
corrective actions in the case of undesired changes in product/process
parameters.

5.3 Decision trees

5.3.1 Introduction

￿

￿ ￿

Decision trees are an important tool for decision making and risk
analysis, and are usually represented in the form of a graph or list
of rules. One of the most important features of decision trees is the
ease of their application. Being visual in nature, they are readily
comprehensible and applicable. Even if users are not familiar with the
way that a decision tree is constructed, they can still successfully
implement it. Most often decision trees are used to predict future
scenarios, based on previous experience, and to support rational decision
making.
Another important area of decision tree application is data mining,
where decision trees are used as a classifi cation and modeling technique,
which is explained in more detail below.