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.

1. Introduction

As early as 50s in the 20th century, health assessment had been already developed. However, in quite a long time, medical care was still stress related, and only when people felt uncomfortable or even the circumstance was very serious they would go to hospital. With vigorous development of data mining technology, consistent improvement of information science, advances in analytical methods and widely used digital medical equipment, establishment and continuous improvement of HIS (hospital information system), the capacity of information database keeps expanding. In these massive data, people haven't yet discovered hidden essence relationships beneath superficial phenomenon; how to take full advantage of these valuable medical information resource to find relationships and rules, and the development trends of future, finally to provide a scientific basis of decision-making for early warning of disease, diagnosis and treatment, to promote development of medical research, has become focus of today's study. To do the research, we not only want to dig medical data for health essence, but also take external action to explore human health, disease development rules and conquer the nature. As time goes on, the research range is developing gradually from disease risk and process evaluation to other aspects of health assessment, and it has a broad application in prevention, medical care, insurance and decision-making of population health.