Explainable AI integrated Fuzzy Rule-Based Machine Learning and Nonlinear Variation Inequalities for Oral Cancer Disease Detection and Treatment Methodology in Healthcare

Abstract

Diagnosis of oral cancer at an early stage is a decisive factor in the treatment and survival rates. The machine learning-based diagnostic methods outsmart the traditional methods of detection. However, data representations fall short of accuracy and precision. To resolve such data handling problems, this study proposes a hybrid decision model combining fuzzy logic with decision trees and nonlinear variation inequalities. The inference of fuzzy logic facilitates the accuracy and reliability of making decisions on the detection of cancer and suggesting treatment. The biological and clinical features are considered for study. The fuzzy logic integrated decision trees framework is developed in this research work to evolve a more comprehensive decision-making system for detecting the stages of oral cancer. The fuzzy rule-based nonlinear variation inequalities are applied in developing the treatment recommendation system. The results of the hybrid model are compared with the conventional prediction approaches and the performance measures favour the proposed models. Explainable AI (XAI) is also applied especially SHAP and LIME are used to determine the credibility of the results. These integrated models are more robust and competent in oral cancer management comprising both detection of the cancer stages and treatment suggestions.