Prominent label identification and multi-label classification for cancer prognosis prediction

Cancer prognosis prediction improves the quality of treatment and increases the survivability of the patients. Conventional methods of cancer prediction deal with single class by limiting the prognosis prediction to one response variable. The SEER Public Use cancer database has more prominent variables that support better prediction approach. The objective of this paper is to find the prominent labels from cancer databases and use them in a multi-class environment. The implementation consist of three phases namely, pre-processing, prominent label identification and multi-label classification. Breast, Colorectal and Respiratory Cancer Data sets have been used for the experimentation. Also random samples from all three data sets are generated to form a mixed cancer data. Patient survival, number of primaries and age at diagnosis are the prominent labels identified from others using the Decision tree, Naïve Bayes and KNN algorithms. The three prominent labels have been tested using multi-label RAkEL algorithm to find the relations between them. The results of the empirical study are comparatively better than the traditional way of cancer prediction.