Thrombus is one of the major problems in ventricular assist devices (VADs). However, method for detecting thrombus in early stage has not been established yet. In this study, we propose an intelligent function that the VAD itself can detect thrombus automatically and alert it to medical staffs. In the proposed method, thrombus formation inside a blood pump is detected by monitoring blood viscosity. This viscosity measurement is performed by using magnetic levitation system for the impeller. Hence, it can be implemented without any additional sensors or mechanisms in principle. For verification of the method, at first, we visualized inside of the pump during thrombus formation with measuring blood viscosity by using erythrocytes removed porcine blood. The result showed that the viscosity of the blood increased as blood coagulation progressed. Then, we conducted in vitro principle verification experiments with three different whole porcine blood. In all experiments, the measured blood viscosity increased and small thrombus was observed inside the pump. From these results, we confirmed that the proposed method has a possibility to detect and predict the thrombus in early stage.