A Modified Metaheuristic Algorithm Integrated ELM model for Cancer Classification

Background: In the rapidly defiled environment, cancer has emerged out as the most threatening disease to human species. Therefore, a robust classification model is required to diagnose cancer with high accuracy and less computational complexity. Method: Here, random parameters of Extreme Learning Machine (ELM) are optimized by Self Adaptive Multi-Population-based Elite strategy Jaya (SAMPEJ) algorithm. This strategy constructs a robust ELM classifier named as SAMPEJ-ELM model. This model is tested on Breast cancer, Cervical cancer and Lung cancer datasets. Here, a comparative analysis is presented between the proposed model and basic ELM, Jaya optimized ELM (Jaya-ELM), Teaching Learning Based Optimization (TLBO) optimized ELM (TLBO-ELM), SAMPEJ optimized Neural Network (SAMPEJ-NN), SAMPEJ optimized Functional Link Artificial Neural Network (SAMPEJ-FLANN) models. Numerous performance metrices viz. accuracy, specificity, Gmean, sensitivity, F-score with receiver operating characteristic (ROC) curve are used to estimate the proposed model. Moreover, this model is compared with eleven existing models. Results: SAMPEJ-ELM model resulted the highest degree of accuracy, sensitivity and specificity in Breast Cancer (.9895, 1, .9853), Cervical Cancer (.9822, .9948, .9828), Lung cancer (.9787, 1, 1) datasets. Conclusion: The experimental results reveal that SAMPEJ-ELM model classifies both the positive and negative samples of cancer datasets significantly better than others.