A Deep Neural Network based Feature Selection and Classification for an Efficient Prediction of Lung Cancer

Abstract

The framework for the prediction of lung cancer is presented in this research and it incorporates a Deep Neural Network (DNN) for feature selection and classification. The f main intent of the research is to improve the performance of the model in predicting the outcomes based on the significant features obtained from the high-dimensional datasets. The proposed method enhances the DNN structure by incorporating a feature selection step that determines which features provide the most significant contribution in classifying lung cancer. This selection mechanism helps to minimize the data dimensionality and to prevent the problem of curse of dimensionality. Thus enhances the learning process and performance of the proposed models. Extensive testing also proves that the proposed method is more efficient compared to other methods in terms of accuracy, precision, and recall. This implies that the proposed method yielded the highest classification accuracy of 93.67%, which is significantly higher than basic ML and shallow neural network models. This superior performance confirms the effectiveness of the proposed feature selection method and its integration into the DNN architecture. The findings suggest that the proposed Feature Retrieval Mechanism Assisted with Deep Learning (FRM-DL) approach improves the predictive performance, and at the same time, presents a viable solution to handle big medical data. Therefore, this research is beneficial for the field of medical diagnostics as it presents a more accurate and efficient method of detecting lung cancer in its early stages, which can help in improving the prognosis and treatment of the disease.