A Soft Measurement Technique for Searching Significant Subsets of Prostate Cancer Prognostic Markers
Hüseyin Şeker, Michael O. Odetayo, Dobrila Petrov́ić, Raouf N.G. Naguib, Freddie Charles Hamdy · 2000
We propose a soft measurement computed by means of the Fuzzy K-Nearest Neighbor (FK-NN) algorithm to determine the degree of importance of subsets of the prostate cancer prognostic markers for prognostic analysis. A class membership degree of each vector for each class was considered rather than its crisp assignment. We present results that indicate that some specific subsets are very capable of prediciting prostate cancer and therefore we do not need to use all prognostic markers. The results also show that the soft measurement can give an idea of the level of importance of each subset rather than ranking, and its outcomes are more flexible and interpretable.