Dual Clustering for Clinical Care Construction

A hospital information system (HIS) was introduced about twenty years ago and all the clinical environment has been dramatically changed [1]-[3]. HIS stores all the histories of clinical activities in a hospital, such as electronic patient records, laboratory data, x-ray photos, and so on. The advantage of HIS is that all the data are input through the network service and they can retrieve from the terminals inside the hospital [3], [4]. Data stored in HIS can be viewed as histories of clinical actions, described as the results of clinical actions with time stamp. Thus, data mining techniques, explored in web mining or network analysis can be applied to the HIS data. Data mining in HIS may become an important tool for hospital management in which spatiotemporal data mining, social network analysis and other new data mining methods may play central roles [2], [5]. 1 The method consists of the following five steps (Fig. 2): first, histories of nursing orders are extracted from hospital information system. Second, orders are classified into several groups by using clustering on the pricipal components (sample clustering, Fig 1). Third, attributes clustering is applied to the data. Fourth, the method compares between generated functions for sample and attribute clustering which relate the number of clusters and calculated similarities. Fifth, if attribute clustering gives better performance with respect to the function, the dataset is decomposed into subtables by using the grouping of attribute clustering. Then, the first step will be repeated in a recursive way. After the grouping results are stable, a new pathway will be constructed from all the induced results. The results show that the proposed method is useful for construction of a clinical pathway Clinical environment is very complex, and flexible and adaptive service improvement is crucial in maintaining quality of medical care. Thus, incremental software development in hospital information system and its evaluation is important. This paper introduces a statistical estimation method of an embedded software in which service logs are used to measure the differences between responsive time before and after a new interface has been introduced. The empirical results show that statistical methods are useful to evaluate the system performance in a real clinical environment.