A Fuzzy Soft Set and Evidential Reasoning Framework for Medical Diagnosis
Keywords:
Fuzzy soft set; Evidential reasoning; Basic probability assignments; Dempster-Shafer theory; Medical diagnosis; Decision-making under uncertainty; Deng entropy.Abstract
Clinical diagnosis increasingly requires accurate interpretation of uncertain, incomplete, or conflicting medical data. This paper presents a hybrid decision-making framework that integrates fuzzy soft set theory with the Evidential Reasoning (ER) rule for medical diagnosis. Fuzzy soft sets model symptom-disease relationships through graded membership, while modulated uncertainty and reliability weights are derived via Deng entropy and fuzzy preference relations. Basic Probability Assignments (BPAs) are constructed from these weights and aggregated using the ER rule to produce a consistent belief distribution over candidate diagnoses.
The framework is validated on a standard medical diagnosis benchmark adopted from Xiao [31], and it correctly identifies the optimal diagnosis in agreement with all prior methods [31–33]. Unlike the classical Dempster-Shafer (DS) rule, the ER rule avoids counterintuitive results under high-conflict evidence (the Zadeh paradox), and DS emerges as a special case of ER when all reliability weights equal unity. The proposed framework is therefore more general, robust, and theoretically sound for uncertain medical decision-making.

