Lung Cancer Accurate Detection Employing Machine Learning Stacking Approach and Relief Algorithm for Feature Selection Using Clinical Data

Lung cancer(LC) diagnosis is a critical health issue and an precise identification is expressively crucial for early cure. Different Artificial intelligence (AI) techniques particularly Machine learning (ML) techniques-based models have been proposed to detect Lung cancer. However, these models still have a lack of accuracy issue. To tackle the accuracy problem we proposed a stacking-based Machine Learning model for correct identification of LC. In designing the model ML algorithms are secondhand to diagnose LC using the Lung Cancer dataset. Feature selection(FS) algorithm Relief is incorporated to select appropriate features for effective training of the models. Further, the stacking approach is used to increase the predictive performance. The base classifiers NB, SVM, RF, AB, and KNN are trained selected features dataset and the meta classifier LR is selected for final prediction. Model performance is assessed using performance assessment metrics, and the data set is divided for training and validation using the held-out cross-validation (CV) approach. When accuracy was tested between the proposed model and baseline models, the suggested model SLC achieved a high accuracy. We suggest the suggested model (SLC) for the prediction of lung cancer in e- healthcare systems because of its excellent accuracy.