Maternal And Child Health Inspection Data Mining And Visualization Research

With the continuous improvement of technology, the means of hospital to collect data continuously improved, the hospital to the rising number of original data. The basic information of the large number of patients and various cases such as the original data is stored, the data is a valuable asset to the doctor, hide that a lot of useful information. How to put behind these abstract information, become one of research hot topic in recent years. Data mining, as a kind of effective method of information extraction method and found that, from a hospital database to extract useful data, and then analyze these data and evaluation, find out the hidden rule, these data and provide the scientific medical methods and judgment.In recent years, maternal and child health care hospital in a lot of women during pregnancy test data sets, the knowledge and information hidden in the data is very important for doctors to understand and study the disease. This topic with a maternal and child health hospital of pregnancy test data as an example, the data mining technology in the application of women during pregnancy test data analysis research, through analyzing the influence factors of pregnancy women, for eugenics and adult's health provides quantitative help.First, this paper introduces a method of testing data pretreatment, and presents a new method of data cleaning on the basis of the original one. By contrast, this method has a great improvement in efficiency and quality than the Bayesian networks. Then by the way of data processing such as data conversion, dimension reduction and etc, the missing data, noise data, and inconsistent data are effectively treated. On this basis, the data preprocessing on test data during the detailed analysis and integration has made from the cleaning and choosing of data to the integration. It will make the data more orderly and is of great help to the analysis latter. To a great extent, reduce the error of the experiment, and experiment, improve the efficiency of the experiment.Next, the processed test data will be clustered. First analyzing the clustering algorithm, it has been improved on the basis of the original k-means algorithm, the selection of k value will be more accurate and concise. Compared with the original k-means algorithm, the improved k-means algorithm can improve the clustering effect greatly. Then the improved algorithm clustering analysis will be carried out on the test data, then using the density clustering algorithm(DBSCAN) and expectation maximization(EM) clustering algorithm to do clustering analysis of the data. The several different clustering algorithm of inspection data have a comprehensive cluster analysis. The results of the analysis can be illustrated with data visualization method. The visualization can show the law of the data and information more intuitively and, are of great use to the data analysis and understanding. It can not only allow doctors to have more clear decisions about patient' illness, but also convenient for the patient to the understanding of the data. So that patients can have a clear understanding of their own conditions, treat them selves and treat their illness.Finally, using the weka software to analyze several clustering algorithms and combined evaluation, and then integrating the algorithm with the pregnancy test data comparison so we can choose the optimal clustering algorithm, which is the most suitable test data of the algorithm.