|
Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
| ABSTRACT - Internet Electron. J. Mol. Des. August 2002, Volume 1, Number 8, 418-427 |
Support Vector Machines for Cancer Diagnosis from the Blood
Concentration of Zn, Ba, Mg, Ca, Cu, and Se
Ovidiu Ivanciuc
Internet Electron. J. Mol. Des. 2002, 1, 418-427
|
Abstract:
Machine learning techniques, mainly artificial neural networks,
clustering and classification algorithms, have recently received
considerable attention as successful methods for modeling medical
data. Using a wide variety of mathematical equations, machine
learning algorithms are able to generate predictive models for different
cancer types. Support vector machines (SVM) are a new machine
learning algorithm that found numerous applications in bioinformatics,
cheminformatics, computational biology, and structure-activity
relationships. In this study we have investigated the application of
SVM for cancer diagnosis from the blood concentration of Zn, Ba,
Mg, Ca, Cu, and Se. The SVM model with the best prediction power
was identified by a leave-10%-out cross-validation procedure, using
the dot, polynomial, radial basis function, neural, and anova kernels.
Extensive simulations demonstrate that the classification performances
of SVM depend strongly on the kernel type and various parameters
that control the kernel shape. The best prediction results were obtained
with a dot kernel with seven support vectors. The anova kernel offered
comparable predictions, but with 24 support vectors. Support vector
machines represent a powerful and flexible classification algorithm,
with many potential applications in modeling medical data. The results
reported in the present study demonstrate such an application in the
cancer diagnosis.
|