Explored research on data preprocessing and mining technology for clinical data applications

On the basis of introducing data preprocessing and mining technology, research is developed on clustering and modeling of mining about clinical data (biochemical indicators), to find potential information related with health assessment and disease prediction, and to indicate further research direction. Based on characteristics of clinical data, Sigmoid function is used to preprocess the original data, then the self-organizing neural network is selected to model, at last modeling results are analyzed and compared with clinical diagnosis. In data processing, the Sigmoid function can maintain the same geometry of raw data, and the output of the modeling is almost the same with the clinical diagnosis. Through preprocessing clinical data, the quality of network input data is improved, and obstacles for further clinical data mining modeling are removing. At the same time, it is found that when carrying out health and disease state clustering, the application of SOM neural network is feasible, consequently making foundation for improving ability of assistant diagnosis and developing health risk assessment.