A step towards the functionality of hospital information managment system data using data mining techniqees : Selection of appropriate parameters for K- means

Administration of healthcare services require simultaneous handling many issues such as financial, medical, technological subsystems. A hospital management team has to gather many details and convert them into sufficient information to take rational administrative decisions. In this respect hospital information management systems (HIMS) combine information technology to support healthcare service in an easier and more efficient way. HIMS is whole of integrated information systems that help to store and access information of many services and products. Even though it stores huge data, HIMS in Turkey are used for basic fundamental reports. Thus, while there is a high demand for well-processed information; perception of metadata to point out evident relationships for healthcare data is lacking and there is need to benefit from HIMS in a more elaborate way. By use of data mining techniques one can process multiple databases, summarize big data from different perspectives and convert data into necessary information. K-means cluster method can be used to extract relationships from the big multivariate data in HIMS. The method mentioned require the user to provide the initial parameters (a number of clusters and initial seeds ) and this requirement becomes a disadvantage for the method. This study, attempts to solve this problem to propose a method for finding optimal number of cluster and initial seeds.The software which includes the algorithm named NAMGY (Neighborhood And Midpoint Gain Yield ) designed for this purpose will be able to report the information that the hospital administrators need using HIMS data more easily and large amount of cumulative data will be available.