Neural Network Ensemble-Based Prediction System for Chemotherapy Pathological Response: A Case Study

Abstract

During the diagnosis of breast cancer, neoadjuvant chemotherapy is supplied intravenously. Physicians recommend chemotherapy before surgery to reduce the invasive tumor’s large size. This research work suggests a model for Neural Network Ensemble Machine Learning, Implementation of a series of machine learning algorithms to create an enhanced and efficient predictable solution patients ‘maximum pathological response after Neoadjuvant Chemotherapy. The quality of the neural network ensemble framework for machine learning is measured using multicriteria technique of decision making known as simple weighted additive (WSAW). The performance score for WSAW is calculated by taking into account ten measurements, namely accuracy, mean absolute error, root mean square error, TP rate, FP rate, accuracy, recall, F-measure, MCC, and ROC. The results are verified using the technique of cross-validation K-fold to achieve 97.20% accuracy. The findings are quite positive when the execution of the proposed system is coupled with the output of state-of-the-art classificators, for example, Bayes Net, Naïve Bayes, logistic, multilayer perceptron, SMO, voted perceptron, etc. With the increasing trend of artificial intelligence applications in cancer research, machine learning has a great future in forecasting and decision making.