A noninvasive method for coronary artery diseases diagnosis using a clinically-interpretable fuzzy rule-based system

摘要: Background: Coronary heart diseases/coronary artery diseases (CHDs/CAD), the most common form of cardiovascular disease (CVD), are a major cause for death and disability in developing/developed countries. CAD risk factors could be detected by physicians to prevent occurrence near future. Invasive coronary angiography, current diagnosis method, is costly associated with morbidity mortality patients. The aim this study was design computer-based noninvasive system clinically interpretable rules. Materials Methods: In study, Cleveland dataset from University California UCI (Irvine) used. interval-scale variables were discretized, cut points taken literature. A fuzzy rule-based then formulated based on neuro-fuzzy classifier (NFC) whose learning procedure speeded up scaled conjugate gradient algorithm. Two feature selection (FS) methods, multiple logistic regression (MLR) sequential FS, used reduce required attributes. performance NFC (without/with FS) assessed hold-out validation framework. Further cross-validation performed best classifier. Results: dataset, 16 complete attributes along binary CHD (gold standard) 272 subjects (68% male) analyzed. MLR + showed performance. Its overall sensitivity, specificity, accuracy, type I error (α) statistical power 79%, 89%, 84%, 0.1 respectively. selected features “age ST/heart rate slope categories,” “exercise-induced angina status,” fluoroscopy, thallium-201 stress scintigraphy results. Conclusion: proposed method “substantial agreement” gold standard. This algorithm thus, promising tool screening