The Study On The Cost Of The Inpatients Of Gynecology Oncology Based On Data Mining Method

ObjectiveIn recent years, the costs of medical has increased rapidly, how to control these rising costs of hospital is become one of the key of medical and health reform in our country. Gynecology tumor is a kind of severe disease which has high morbidity and mortality, can cause excessive consumption of medical resources. There is not a reasonable and feasible cost consultation system for inpatients of gynecology tumor, for current few studies about the cost of single-disease such as gynecology tumor, but the medical resources comsumption is medicare advance payment and health economy specialists wanted, so the study on the cost of inpatients of gynecology tumor is necessary.MethodsData mining is a new technology which has been arisen in recent years, it is a new field whether in our country or abroad in the medical and health economic fields. In the study, the selected cases based on ICD-10, which were collected from 2003 to 2008 year about one of hospitals in Henan Province, included about 7650 gynecology tumor. First, we eliminated the abnormal values and missing values for the data, then we consulted the methods of health economy and the key parameters in the National Bureau of Statistic of China to emend the correlative inpatient cost. Second, we used descriptive statistical analysis analyze the cost of hospitalization cases which were affected by the different factors. Then we did the normality transformation for variable and feature selection node method for the main factors which affecting the hospitalization costs. Finally, we seted up the inpatients of gynecological malignancies impact factors model with decision tree and artificial neural network model. All the steps were in the software of Microsoft Excel 2003 and SPSS Clementine.ResultsWe did descriptive statistical analysis for the pretreatment datas and analyzed the facts for the hospitalization costs of gynecology tumor, include the diagnosis, the age, the marital status, the marriage, the day before surgery, the day in hospital, the admissions, the discharge from hospital, the way of surgery and the type of anesthesia.In data mining, we divided the gynecological malignancies into surgery and non-surgical cases. Then we used feature selection node method and got the main factors for the hospitalization costs. The main factors for the hospitalization costs of the patients treated by surgery are the diagnosis, the way of surgery, the day in hospital, the number of the day before Surgery and the age; the main factors for the hospitalization costs of the patients treated without surgery are the type of anesthesia, the diagnosis, the day in hospital and the age.Finally we established the decision tree model and the artificial neural network model. We selected the decision tree model about 3-storey and 12 classification for the gynecological surgery malignancies and 2-storey and 7 classification for the non-surgical malignancies, and obtained the percentage of each category and the reference range of the hospitalization costs. The cases of surgical treatment of patients with non-surgical treatment by the artificial neural network model in the input layer, hidden layer and output layer were 25,3 and 4 neurons. After assessment, the correct rate of the decision tree and artificial neural network model predict the correct rates were all more than 80%.ConclusionsIn our study, we give the percentage of each category and the reference range of the hospitalization costs according the decision tree model. When the related departments design the cases of the gynecological malignancies hospitalization costs, they can use this reference range for the evaluation of medical resource consumption and the development of the standard range of medical expenses. In this study, while the output for the objective evaluation of medical care or health care, establishmenting the national standards for hospital reimbursement and to develop the reasonable standards of medical resource consumption is important.