Affiliations: Decision Systems Group, Brigham and Women's Hospital,
Boston, MA, USA | Division of Health Sciences and Technology, Harvard
and MIT, Boston, MA, USA
Note: [] Corresponding author: Lucila Ohno-Machado, Decision Systems
Group, Thorn 310, Brigham and Women's Hospital, 75 Francis Street, Boston, MA
02115, USA. Tel.: +1 617 732 8543; Fax: +1 617 732 9260; E-mail:
[email protected]
Abstract: Microarray technologies have allowed the measurement of expression
of multiple genes simultaneously. Gene expression levels can be used to
classify tissues into diagnostic or prognostic categories. As measurements from
different microarray technologies are made in different scales (which are not
guaranteed to be linear and not easily re-scalable), it is helpful to develop
an easy-to-interpret technology-independent classification scheme. To capture
the essentials of the problem of classification using gene expression data, we
show how fuzzy logic can be applied using two examples. Using information from
genes previously shown to be important, the classification performance of the
fuzzy inference is similar to that of other classifiers, but simpler and easier
to interpret. The fuzzy inference system has the theoretical advantage that it
does not need to be retrained when using measurements obtained from a different
type of microarray. Although the data sets for gene expression analysis
utilized in this paper are relatively small, they are among the largest
available in this domain.
