A probabilistic neural network approach for BCTTCM classification

In view of the limitations of classifying body constitutional type in traditional Chinese medicine (BCTTCM) with the traditional method, the probability neural networks (PNN) method was put forward. The characteristic parameters of BCTTCM were obtained with the median, and a PNN model was designed for BCTTCM classification. The PNN model was trained and tested using 30 samples, and it yields good classification accuracies for BCTTCM. The overall accuracy of the neural network was 95% in the training set. It was 80% in the validation set. It indicated that the model was effective and the method based on the PNN for BCTTCM classification is feasible.