Robust Meta-Classification Stategy for Cancer Detection from Mass Spectroscopy Data

Gyan Bhanot (a), Gabriela Alexe (b), Babu Venkataraghavan (b), Arnold Levine (b)
(a) IBM Research, Thomas J Watson Research Center, Yorktown Hts., NY 10598
(b) Institute for Advanced Study, Princeton, NJ 08540

Abstract:

In this talk I will describe how to robustly identify peptide peaks in mass spectroscopy SELDI-TOF data to separate Cancer from Non-Cancer using a data base for Prostate Cancer. The talk will give a brief overview of the disease and discuss the potential of mass spectroscopy in cancer detection and therapy. Next I will describe pattern identification, data preparation and machine learning and noise analysis techniques that must be applied to the data to obtain robust classifiers. Finally, I will describe the metaclassifier which is based on a number of machine learning approaches and show how it improves both specificity and sensitivity of predictions. I will conclude with some suggestions for future work.

back to the list of talks