The aim of this project is to improve the detection of coronary occlusions using an approach based on the recording and analysis of isolated diastolic heart sounds associated with turbulent blood flow in occluded coronary arteries. The nonlinear dynamic analysis method based on approximate entropy has been proposed for the analysis of diastolic heart sounds. A commercially available electronic stethoscope was used to record the diastolic heart sounds from patients diagnosed with or without coronary artery disease (CAD) based on their coronary angiography examination. The nonlinear dynamical analysis (approximate entropy) measures of the diastolic heart sounds recorded from 30 patients with coronary occlusions and ten normal subjects were estimated. Results suggest the presence of the high nonlinear (approximate entropy) values of diastolic heart sounds associated with CAD ( $p ≪ \hbox{0.05}$). This approach led to a sensitivity of 77%, a specificity of 80%, and an overall accuracy of 78%. As a summary, 23 out of 30 abnormal patients and eight out of ten normal patients were correctly detected.