IJMLC 2015 Vol. 5(1): 62-67 ISSN: 2010-3700
DOI: 10.7763/IJMLC.2015.V5.484

High Dimensionality and Scaling-up Performance of RBF Models with Application to Healthcare Informatics

Julio de Alejandro Montalvo, George Panoutsos, Mahdi Mahfoufand, and James W. Catto

Abstract—In this paper, the scaling-up performance of Radial Basis Function (RBF) Neural-Fuzzy models is investigated. The work presented is based on the challenge of analyzing microarray data for the prediction of the patients’ cancer survival. The aim is to find the limit for the maximum number of inputs to use in the model while maintaining low computational complexity and high accuracy. The combination of Fuzzy C-means and RBF-Neural-Fuzzy models presents the challenge of scaling-up when more than a thousand inputs are used. To overcome this challenge we introduce a Weighted Fuzzy C-means (WFCM) algorithm. A second contribution is a cluster optimization algorithm based on the Xie-Beni cluster validity index to improve the quality of the clusters calculated by the WFCM. The best performance (0.87 AUC) for the prediction of the patients’ bladder cancer survival was achieved by a 500-gene signature model with a modeling structure having only five rules.

Index Terms—Bladder cancer, feature-selection, fuzzy logic, health-care informatics, high dimensionality low sample size, microarray, neural-fuzzy, radial-basis-function (RBF).

Julio de Alejandro, George Panoutsos, and Mahdi Mahfouf are with The University of Sheffield, Department of Automatic Control and Systems Engineering, Mappin Street, Sheffield, S1 3JD, UK. Julio de Alejandro and George Panoutsos are also with the Institute for In-Silico Medicine (INSIGNEO), University of Sheffield, UK (e-mail: j.montalvo@sheffield.ac.uk, g.panoutsos@sheffield.ac.uk, m.mahfouf@sheffield.ac.uk).
James W. Catto is with The University of Sheffield, Academic Urology Unit and Institute for Cancer Studies GU22, G Floor, The Medical School, Beech Hill Road, Sheffield, S10 2RX, UK (e-mail: j.catto@sheffield.ac.uk).

[PDF]

Cite: Julio de Alejandro Montalvo, George Panoutsos, Mahdi Mahfoufand, and James W. Catto, "High Dimensionality and Scaling-up Performance of RBF Models with Application to Healthcare Informatics," International Journal of Machine Learning and Computing vol. 5, no. 1, pp. 62-67, 2015.

General Information

  • ISSN: 2010-3700 (Online)
  • Abbreviated Title: Int. J. Mach. Learn. Comput.
  • Frequency: Bimonthly
  • DOI: 10.18178/IJMLC
  • Editor-in-Chief: Dr. Lin Huang
  • Executive Editor:  Ms. Cherry L. Chen
  • Abstracing/Indexing: Scopus (since 2017), EI (INSPEC, IET), Google Scholar, Crossref, ProQuest, Electronic Journals Library.
  • E-mail: ijmlc@ejournal.net