The purpose of this study is to demonstrate the best strategy for the variable selection, using the developed Ordered Logistic Regression (OLR) and Multilayer Perceptron Neural Network (MLP). At the first stage, all the selected variables will be a screen for their important relationship point of view through ordered logistics regression and bootstrap methodology. After considering for 1500 of the bootstrapping methods, it was found that smoking factor, total cholesterol factor, and triglycerides come to a significant relationship to the level of hypertension. By considering the level of significance of 0.25 for ordered logistic regression, these three variables are being selected and used for the input of the MLP model. The performance of MLP was evaluated through the Predicted Mean Square Error (PMSE) of the neural network for the (MSE-forecasts the Network). PMSE is used as a measurement of how far away from our predictions are from the real data. The smallest MSE from MLP, indicate the best combination of variables selection in the model. In this research paper, we also provide the R syntax for OLR and MLP better illustration.

Oral cancer is an important global health concern, representing the sixth most frequent malignant tumor. The oral squamous cell carcinoma (OSCC) is the most common malignant neoplasm of the oral cavity with up to 50% of mortality rate (highest prevalence being identified in Asia). In 2012, it has been reported that 14.1 million new cancer cases and 8.2 million cancer deaths. Numbers of studies have been performed to investigate the factors that have direct and indirect or both associated with the OSCC, including their survival time. In this paper, the potential clinicopathological prognostic factors will be determined in patients who attended Hospital Universiti Sains Malaysia from 2005 to 2015. For such prediction, the use of hazard regression is used previously, but here an attempt is made to propose a covariate-dependent prognostic model to identify the factors and the predictor importance according to the statistical significant point of view. The proposed model is very useful for the prediction and for the inferences of the patient’s management time with the high-risk clinicopathological factors.