Similarity-Based Uncertainty Modeling in Liver Disease Diagnosis

A Comparative Analysis of Fuzzy, Intuitionistic Fuzzy, and Neutrosophic Frameworks.

Abstract

This study presents a systematic comparison of three frameworks for modeling uncertainty: fuzzy sets, intuitionistic fuzzy sets, and neutrosophic sets, and their ability to represent ambiguity and uncertainty. It employs an analytical review of the literature, focusing on the characteristics of each framework and its key algebraic operations.

Similarity measures of the three frameworks are then examined and applied to real clinical data of the most important liver enzymes (AST, ALT, bilirubin) for seven patients. The results indicate that classical sets are unable to deal with ambiguity, while fuzzy sets allow for partial assessment of membership. Intuitionistic fuzzy sets distinguish between membership and non-membership, whereas neutrosophic sets represent three independent components (truth, indeterminacy and falsity), enhancing their ability to assess incomplete and contradictory uncertainty. Applying similarity measures demonstrates how each framework presents the degree of convergence of patient data with each other and with reference measures. The weighted average allows for comprehensive assessment of patient health status. Therefore, similarity measures serve as important elements in clarifying the scientific differences between the three theories, explaining how each framework measures information proximity differently and its ability to represent components. Results indicate that neutrosophic similarity provides the most precise distinction in borderline or ambiguous cases, enhancing medical decision-making accuracy.