A Comparison of the Logistic Regression Model and Neural Networks to Study the Determinants of Stomach Cancer

Abstract

The logistic regression model and the neural network model are among the best models in the bilateral response data and the classification of different medical conditions. Therefore this study addressed the comparison between the two models using statistical classification criteria (model accuracy, model sensitivity, model specificity, the model's false alarm rate, the area under the Receiver Operating Characteristic (Roc) curve, Wrong classification rate). After applying these criteria to the study data, we concluded that the neural network model is better than the logistic regression model, as it was reached through the final reconciliation of both models that the factor of the method of diagnosing stomach cancer has the obvious effect on the classification of the patient's condition, and this was confirmed by the relative importance of the factors studied using the neural network model, which showed that this factor reached its relative importance 100%, which is a very large percentage compared to other variables.