This paper presents a novel approach to diabetes prediction using a Fuzzy Enhanced Apriori Rule-Based Classifier, using the Pima Indian Diabetes Dataset. The dataset, containing 768 records, was divided into training and testing subsets with a 65–35% split. Missing values were replaced by the mean of the respective attributes, and a triangular fuzzy membership function was employed to derive crisp attribute values, enhancing the predictive capabilities of the Apriori algorithm. Experimental evaluations were conducted using J48, Random Forest, Naive Bayes, and Logistic Regression classifiers for performance comparison. The proposed classifier achieved superior results, with an accuracy of 85.6%, demonstrating a significant improvement in prediction accuracy and classification metrics. Notably, the Fuzzy Enhanced Apriori Classifier outperformed the benchmark algorithms, achieving higher accuracy and robustness across various performance indicators. The Random Forest classifier correctly classified 79.18% of instances, Logistic Regression 80.30%, and Naive Bayes 78.07%. In comparison, the proposed classifier excelled in correctly identifying both diabetic and non-diabetic instances, underscoring its efficacy in handling complex datasets with missing or ambiguous values. These results highlight the potential of integrating fuzzy logic with rule-based classification to improve predictive accuracy in medical diagnostics, offering a promising tool for early diabetes detection and personalized treatment strategies.