A Framework For Cancer Gene Classification And Prediction Using Soft Computing Techniques

Abstract

Microarray data analysis plays a significant role in accurately classifying cancer, newlineenabling early detection and tailored treatment strategies. However, challenges such as newlinehigh dimensionality and imbalanced sample distributions need to be addressed. In this newlineresearch, we propose a comprehensive framework for cancer gene classification and newlineprediction using soft computing techniques. In the first contribution, an ensemble newlineclassifier combining Support Vector Machine (SVM) and Naive Bayes with Principal newlineComponent Analysis (PCA) for dimensionality reduction is presented. The proposed newlineSVM-NB ensemble consistently outperforms existing techniques, achieving higher newlineaccuracy, precision, recall, and F-measure. Experimentation validates the superior newlineclassification efficiency of the proposed ensemble classifier by 5-7% for considered newlineparameters. The second contribution addresses the selection of accurate and robust newlinegene subsets using a hybrid framework, Non-Dominated Sorting Genetic Algorithm newlinewith Learning Automata (NDSGA-LA). NDSGA-LA optimizes both classification newlineaccuracy and feature stability, surpassing other algorithms such as SVM and Decision newlineTree in terms of accuracy, precision, recall, and F-measure by 10-15%. The third newlinecontribution focuses on prediction, proposing a hybrid CNN+BiLSTM approach for newlinecancer type prediction based on microarray gene expressions. Experimental results newlinedemonstrate superior performance with high classification accuracy (98.3%), precision newline(98.1%), recall (97.8%), and F-measure (97.94%) as compared to existing CNN and newlineLSTM classifiers. Overall, the proposed framework enhances cancer gene analysis by newlineaddressing challenges, achieving superior classification accuracy, and providing newlineinsights into disease progression, recurrence, and treatment response. This work newlinecontributes to improving cancer diagnosis, treatment, and patient outcomes newline