OP-DNN: Lung Cancer Survivability Prediction using Novel Optimized-Deep Neural Network Classification Method

Abstract

Lung cancer disease is the most widely recognized deadly disease in the world for the loss of life. Throughout this research, Electronic Health Records (EHRs) textual data are investigated, and survivability rates for lung cancer affected patients are predicted. If the Lunga cancer patients are survivable for more than one year, chemotherapy treatment can be started for those patients. This research paper examines an effective Batch Size-Optimizer based Deep Neural Network (Op-DNN) classifier framework model, which is developed to predict the patient’s survivability. Here the textual data set is classified and processed in batches for each iteration. The errors generated from the original classification of the initial batch size is fed back to the Op-DNN classification algorithm for further iterations with the reduced error loss that is free from underfitting and overfitting. The proposed method is compared with various parameters for Machine learning classifier algorithms demonstrating that the Op-DNN model has achieved a better accuracy rate of 91.353 % for prediction of Survivability.