General Discussion and Future Directions - DNA Microarray Technology and Data Analysis in Cancer Research - page 103

that people have adopted, but until now (with the exception of scanning)
they all depend either on a starting design, some form of local optimiza-
tion, or some random variation. Each method will usually give rise to dif-
ferent solutions. For gene subsets using a large number of genes, these are
still the only methods possible. In contrast, the SDL optimization
described here is a methodical global method.
The proposed pyramidal hierarchy of the predictor for classification
can effectively improve the signal-to-noise ratio in mining the high-
dimensional microarray datasets. While research in cancer classifica-
tion with microarray expression data is the first to benefit from this
method, the mathematical procedure used in this study — SDL global
optimization — is also applicable to a variety of other unsolved prob-
lems related to linked multi-variable problems. The application of this
technique will undoubtedly have implications well beyond cancer clas-
sification application.
It is still too early to predict what the ultimate impact of microarray
will be on our understanding of cancer, although the possibility of an
accurate diagnosis of cancers based on microarray expressions has
emerged. This innovative research truly brings to light one of the hardest
problems yet: the ability to accurately classify medical neoplasm. The
SDL method provides a precise diagnostic tool that can find the true
global optima with questions relating to gene malignancy. Furthermore,
genetic screening for diseases is playing an increasingly important role
in preventative medicine — if we can detect the presence of disease or
predict the malignancy through microarray expression data with a desktop
computer before clinical diagnosis, a more efficient and clear-cut treat-
ment plan can be formulated, eliminating the possibility of clinician bias.
More importantly, an unbiased and digital data-based approach can be eas-
ily applied to distinctions relating to future clinical outcome, such as drug
response or survival. In cancer research, fundamental mechanisms that cut
across distinct types of cancers could also be discovered through mining
microarray data using SDL global strategies. In the future, the use of our
SDL global optimization method in DNA microarray data analysis has
great potential to help simultaneously gather disease information related to
the expression of all genes from a patient — i.e. personalized medicine.