| Summary: | 碩士 === 國立中正大學 === 醫療資訊管理研究所 === 102 === Chronic kidney disease (CKD) is considerable attention to public health issues in recent years, which is the World Health Organization effort to combat chronic diseases. Through regular follow-up during clinical management to monitor the patient's physical status, grasp the factors that might affect the course of CKD deteriorate to perform related treatment, slow the rate of progression to reduce the incidence of end-stage renal disease. However, the record data of CKD patients is a higher dimension of time series data. In the past, the assessment on CKD progression of clinically relevant research, there is no suitable method can be used to handle this type of data. Therefore, how to build reliable prediction models, and find the temporal patterns from the time series data which can be used to explain the evolution of the disease, has become an important issue. The purpose of this study is to construct reliable predictive models that integrate temporal abstract methods with data mining technology for analyzing the record data of CKD patients. In addition, through the results of data analysis, identify relevant factors that can slow the course of CKD progression.
The findings of this study showed that include temporal information in the prediction model for analysis, helps to raise the forecast efficiency. From the different stages of the selected forecast model is not the same，In the results of the analysis, AdaBoost + CART is the best effectiveness of constructing predictive models (Accuracy：64.29%，AUC：0.679). Furthermore, we find some factors that is associated with deterioration of renal function, which have confirmed in past studies(ex：Sex、Age、Chronic diseases…etc). Some influencing factors (ex：BUN、Blood pressure、Hct…etc) after the converted to the temporal patterns, which can go further to provide some information that include clinical knowledge to to explain the CKD progression，To assist medical personnel in clinical decision making can provide the appropriate diagnosis for patient, improved quality of care for patients with CKD.
|
|---|
