| Summary: | 碩士 === 長庚大學 === 管理學院碩士學位學程在職專班資訊管理組 === 100 === As the progress of the information technology, lots of the medical information can be saved intact. Those information collected days by days are seldom used besides the random inspected by National Health Insurance and the law about holding those materials for more than seven years. Through the further research of the data mining theory combined with professional medicine is a probable way to solve the problem about how to reuse this enormous and disarranged database.
This study used decision tree and artificial neural network to discuss the medical examination data of hemodialysis from end-stage renal failure patients and empirically analyzed the correlation with the rate of hospitalization. End-stage renal failure patients need hospitalization for long-tern dialysis due to insufficient self-health management or demand for long-lasting hemodialysis. If the rate of hospitalization in the hemodialysis center is higher than the average, the quality of services in the center is defective. Therefore, to decrease the rate of hospitalization is a deserved consideration.
This thesis aims to process the risk assessment of the correlation between hemodialysis patient and its hospitalization which let medical staffs to monitor previously and clinically predict the hospitalization estimation, and suggests doing treatments immediately, which has an effect on the decline in hospitalization frequency and waste of medical treatment.
A medical center in the north Taiwan was analyzed in this issue. Total 3400 samples were collected from 2010 to 2011 and there were 254 conventional patients. Results of data mining: accuracy of artificial neural network is 75.15%; accuracy of decision tree is 68.09%
|
|---|
