The diagnosis coronary heart disease’s system mostly uses techniques of data mining. This system requires an examination carried out starting from risk factors until supporters’ inspection. The examination of the model generates a lot of attributes, which may be contradictory when it is being processed in data mining. This study aims to propose tiered models for the diagnosis of coronary artery disease based on fuzzy inference system. The proposed system is divided into three stages: First, an initial screening using the approach of Framingham risk score. Second, the examination of symptoms and electrocardiograph. Third, an examination using scintigraphy and fluoroscopy supporters. In the second and third stages, the examination begins with rule based on C4.5 decision tree algorithm model, and then uses a fuzzy rule base. Performance parameters used are sensitivity, specificity, and area under the curve (AUC). The results show the system performance at the initial screening at 70.21% sensitivity. Diagnosis in the second stage produces 77.02% sensitivity and 73.53% specificity. The third stage with additional support inspection amounts 82.87% AUC. Model-tiered system based on fuzzy inference system is capable of delivering the performance in good categories.

fuzzy inference system, data mining, coronary heart disease, decision tree, C4.5.