Integrating the Data of Attitude and Behavior in Seeking Medical Care to Predict Loyal Patients

• Main Authors: Chien-Yi Tsai, 蔡建誼
• Other Authors: Fu-Ming Lee
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/96129123312382483954

碩士 === 朝陽科技大學 === 資訊管理系碩士班 === 96 === After implementing national health insurance, the competition between various hospitals in Taiwan is intense day by day, which cause the operation of the hospitals becomes more challenging. The hospitals focus gradually on customer relationship management to retain loyal customers, i.e., loyal patients. In this research, the data of attitude and behavior in seeking medical care are integrated to predict loyal patients. The inputs of forecasting models consist of several hospital choice factors and patient’s attributes, which are produced by factor analysis and data transformation. The patients surveyed are partitioned into two groups, namely, loyal patients and non-loyal patients, by their total attitude score, visiting frequencies, and monetary paid. The clustering results described above serve as the output of forecasting models. With the input and output variables, several forecasting decision tree models are built. To improve the forecast performance, the predictability of all input variables are calculated, and over-sampling or multi-classifier committee approaches are used to balance the loyal and non-loyal training examples. Finally, the best forecast model is selected from the models created in this research. The results show that the best model is built by using multi-classifier committee approach with four patient’s attributes: age, discount, chronic disease, and location. The recall rate of the best model can be as high as 93.75%. With the best model, the cost is very low to predict the loyalty of the patient, since the model does not need surveyed data of patients’ attitude in seeking medical care.