Research On Urology Clinical Decision Support System Based On Data Mining

In recent years, the research on medical engineering makes a rapid development. A lot of medical data is recorded accurately. Data mining methods and techniques are used to find and summarize various knowledge from such a large dataset, such as the clinical features of various diseases, the relationship among various diseases, the developing regulation of various diseases, and the efficacy of various treatment options, which has great value and prospect to diagnosis, treatment and the research of medical science.This paper reviews the main features, basic process, key technology and development of medical data mining. Based on traditional data mining classification algorithms, we study urological diagnosis and present a new algorithm-associative classification based on closed frequent itemsets (ACCF). Because closed frequent itemsets can get all the frequent itemsets, and the number of closed frequent itemsets is much smaller than that of frequent itemsets, the association rules generated by closed frequent itemsets can get all the rules. Combined urologic diseases data characteristics, ACCF can improve its efficiency on pruning and matching rules.Our experiments on 18 datasets of UCI machine learning datasets repository and urology diagnosis dataset indicate that ACCF can generate a smaller set of classificative association rules with higher quality and no redundancy, and has better average classification accuracy than the representational algorithms, i.e. CBA. Our extensive experiments on urology dataset also indicate that ACCF has the highest accuracy in comparison with other traditional data mining classification algorithms. The main contents of this paper and the results achieved are as follows.â‘ We have done a lot of research and comparative experiments on the traditional urology diagnosis data mining classification algorithms,such as decision tree, naive bayesian classifier, BP neural network and CBA.â‘¡Through theoretical analysis of traditional association rules classification, we propose a new algorithm ACCF. Extensive experiments show that ACCF has the highest accuracy in comparison with other data mining classification algorithms.â‘¢Based on the ACCF algorithm, a urology clinical decision support prototype system is designed and implemented.