A novel approach for breast cancer prediction using optimized ANN classifier based on big data environment

• Published in: Health care management science Vol. 23; no. 3; pp. 414 - 426
• Main Authors: Supriya, M., Deepa, A. J.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01-09-2020; Springer Nature B.V
• Subjects: Algorithms; Big Data; Breast cancer; Breast Neoplasms - diagnosis; Business and Management; Econometrics; Female; Health Administration; Health Informatics; Humans; Management; Neural Networks, Computer; Operations Research/Decision Theory; Modified dragonfly algorithm (MDF); Optimized artificial neural network (OANN); Gray wolf optimization (GWO); Improved weighted decision tree (IWDT); Big data
• ISSN: 1386-9620; 1572-9389; 1572-9389
• DOI: 10.1007/s10729-019-09498-w

Cancer is caused by the un-controlled division of abnormal cells in a body part. Various cancers exist in this world and one amongst them is breast cancer. Breast cancer (BC) threatens the lives of people and today, it is the secondary prime cause of death in women. Numerous research directions concentrated on the prediction of BC. The prevailing prediction model is time-consuming and have less accuracy. To trounce those drawbacks, this paper proposed a BC prediction system (BCPS) utilizing Optimized Artificial Neural Network (OANN). Primarily, the unprocessed BC data are regarded as the input. The big data (BD) storage comprises some repeated information. Secondarily, such repeated data are eliminated by utilizing Hadoop MapReduce. In the subsequent stage, the data are preprocessed utilizing replacing of missing attributes (RMA) and normalization techniques. Subsequently, the features are generally chosen by utilizing Modified Dragonfly algorithm (MDF). Then, the selected features are inputted for classification. Here, it classifies the features utilizing OANN. Optimization is done by employing the Gray Wolf Optimization (GWO) algorithm. Experiential outcomes are contrasted with prevailing IWDT (Improved Weighted-Decision Tree) in respect of precision, recall, accuracy, and ROC.