An Expert System of Diagnosis and Prescription of Hypertension and Stroke in Chinese Medicine

• Main Authors: En-Haur Chang, 張恩豪
• Other Authors: Ching-Sung Weng
• Format: Others
• Language: zh-TW
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/58841079784173697368

碩士 === 中原大學 === 醫學工程學系 === 87 === The purpose of this research was to establish an expert system of diagnosis and prescription of hypertension and stroke in Chinese medicine by the theory of Perceptron neural network and rule base. In order to popularize this expert system to other doctors in WWW, we also used web base. First, the model of diagnosis system was built by Error Back-Propagation neural network. There were two layers in this neural network. The first layer was the layer for symptoms input, the second layer was the layer for syndromes output. There were four syndromes in the diagnosis of hypertension and six syndromes in the diagnosis of stroke. The next task of accomplishment of the diagnosis system was rule extraction. And the second, the model of prescription system was built by the combining of multi-layer Perceptron and rule base. There were three layers in this neural network. The first layer was for the symptoms input, the second layer was the handling layer for syndromes output and the third layer was the output layer for drugs output. In prescription of hypertension, eight prescriptions and ten herbs were chosen. In prescription of stroke, ten prescriptions and fifteen herbs were selected. The subjects of this research were provided by Taipei City Hospital of Traditional Chinese Medicine. There were 41 cases in the hypertension study and 36 cases in the stroke study. In the test of diagnosis system, the hypertension system reached an accuracy rate of 0.83 and the stroke one reached 0.75. In the test of prescription system, the hypertension system reached an accuracy rate of 0.86 and the stroke one reached 0.86. In order to improve the accuracy rate of diagnosis system the training data of every syndrome has to be increased uniformly. For raising the accuracy rate of prescription system, discussion with the doctors has to be more and the suitable weights of neural network have to be set. The expert system of Chinese medicine can offer an on-line assist for the doctors because of the function of diagnosis, prescription and web base. In the future, with adding of the device of tongue diagnosis and pulse diagnosis on the client, this expert system could be used in Telemedicine.