Fuzzy-Wiener Graph Framework for Heart Disease Classification

This study introduces a hybrid graph-theoretic framework for heart disease prediction that integrates the Fuzzy Connectivity Index (FCI) and Wiener Connectivity Index (WCI) to enhance classification accuracy and interpretability. The proposed model begins with data preprocessing using feature standardization and one-hot encoding to handle mixed data types. A Logistic Regression classifier generates baseline membership probabilities, representing each patient’s likelihood of having heart disease. To capture inter-patient relationships, a Gaussian-based fuzzy affinity graph is constructed, modeling the similarity between patients as weighted connections. The FCI evaluates the maximum connectivity strength between patients and seed nodes through a max–min path principle, while the WCI quantifies the global compactness and efficiency of the fuzzy network. These indices are fused with probabilistic membership scores to generate refined predictions. Experimental results demonstrate that the FCI–WCI hybrid approach effectively identifies key relational patterns among patients and improves diagnostic reliability compared to traditional classifiers.