An Unsupervised Multi-valued Stochastic Neural Network Algorithm to Cluster in Coronary Heart Disease Data

Clustering of attributes in unsupervised medical data presents a major challenge for many researchers. In this paper We carry out a clinical epidemiology survey of Coronary Heart Disease and obtain 1069 cases. Each case is certainly a CHD case based on the evidence from Coronary Artery Angiography. It includes 78 symptoms and is diagnosed by TCM mentors as syndrome or syndrome combinations. We proposed an unsupervised stochastic neural network algorithm to partition 78 symptoms into several clusters. Each cluster is diagnosed by TCM mentor as syndrome and is clinically verified. The unsupervised stochastic neural network with multi-valued neurons implements clustering of attributes in short duration and obtains seven clusters in the data. The seven clusters correspond to seven syndromes in TCM verified by TCM mentors, which indicates that the cluster is successful and the data surveyed is of high quality, The investigation of stochastic neural network to CHD data to retrieve syndromes in CHD successfully bridges gap between western medicine and TCM. The work here presents a better insight into healing CHD.