Information Technology Reference
In-Depth Information
More complex GAs can include various genetic operators, such as one-
point cross-over, blending operator, main-chain mutation and side-chain
mutation, insertion and deletion, and hop operator (Venkatasubramanian
et al., 1994).
A critical issue when running GAs is to try to preserve genetic diversity
of the population for as long as possible. GAs use a whole population of
individuals (potential solutions) and if that population starts to
concentrate in a very narrow region of the search space, all advantages of
handling many different individuals vanish, while the burden of
computing their fi tnesses remains. This phenomenon is known as
premature convergence (Eiben and Schoenauer, 2002).
5.4.3 Examples
GAs have been used in both qualitative and QSAR studies (Walters and
Hinds, 1994; Jones et al., 1995; Kubinyi, 1994). GAs have also been used
in computer-aided molecular design (Venkatasubramanian et al., 1994).
Several programs based on GAs have been developed to facilitate drug
design (Terfl oth and Gasteiger, 2001). Radial basis functions were based
on a GA and applied to the nondestructive determination of active
component in pharmaceutical powder by NIR spectroscopy (Qu et al.,
2009). GAs were used for variables selection in multiple linear regression
models for prediction of gastro-intestinal absorption of drugs (Deconinck
et al., 2007), as well as for prediction of oral bioavailability (Pintore
et al., 2003). Fitting of diffusion coeffi cients in a three-compartment
sustained release drug formulation was also performed using a GA
(Hirsch and Müller-Goymann, 1995). GAs were used to plan the path of
controlled drug delivery using micro robots (Tao and Zhang, 2005).
Example 1
Lack of process robustness in the pharmaceutical industry is
often related to site-specifi c or equipment-specifi c manufacturing
issues. Automated Intelligent Manufacturing System (AIMS) was
developed to demonstrate feasibility, viability, and value of applying
computational intelligence (machine learning) and evolutionary
algorithms, to, respectively, model and optimize real development
Search WWH ::
Custom Search