Estimating the Tumor Size Using Ordered Logistic Regression with Combining Fuzzy Techniques.

Introduction: Oral cancer is a disease resulting from an abnormal growth cell found in the mouth region such as in lips, tongue or throat. Furthermore, oral cancer commonly occurs at the buccal mucosa (cheek), tongue, the floor of the mouth and lip. The previous study shows that many studies focus especially on detecting the factors with the highest probability of cancer but the study which emphasizes the prediction of the tumor size category is still less from the point of computational biostatistical. Objective: The aim of this research paper is to build the methodology for computing ordinal regression model through SAS language by considering the clinical pathological prognostic factors in patients who attended Hospital Universiti Sains Malaysia (HUSM) from 2005 to 2015 (based on secondary data). Method: In this paper, there are three main methodologies proposed in this study. The first and second methodology is on the ordinal regression methodology and followed by bootstrap method, and the second is a nonlinear programming (NLP) methodology, which aims to obtain fuzzy regression modeling for the prediction purposed. Results: The result from ordinal regression had shown that smoking and nerve invasion factors contributing significantly to the growth of a tumor. The significant result from this finding (output based) can be used to educate people or stakeholder of how important this factor toward patients management. Conclusion: From a statistical point of view, the integration of computing methodology can expose and provide the researcher with an integrated method of analysis. [ABSTRACT FROM AUTHOR]