Summary: | 碩士 === 國立雲林科技大學 === 工業工程與管理系 === 106 === Because of the adjustment of the national health insurance system, and the medical industry is a high-tech and high labor costs work. All the managers in medical institutions have amount of pressure. It is difficult to change the personnel cost. In contrast, the control of the materials cost is easier to implement. It can also decrease the operating cost of the hospital effectively. Because of personnel habits and unexpected incidents, the demand could be difficult to predict. Too much material stock will waste the inventory cost and cause the problem of the inventory space. In the other hand, lack of the stock will affect the medical operation and reduce medical quality. Managers must transfer goods from other departments when run out of material stock. It will waste the time on administrative process. Therefore, it is important to maintain the hospital stock balance.
In this study, a regional teaching hospital in central of Taiwan is used for example. The ABC classification and critical value analysis are used to classify the inventory in this research. We use Holt exponential smoothing, ARIMA and CART with back-propagation neural networks as the prediction method. MAPE is the index of the model accuracy. The result shows that CART with back-propagation neural networks is better than other methods. The MAPE values are below 10%.The difference between Holt exponential smoothing and ARIMA is unapparent. The result of this study can provide hospital for a method of medical material management. Let patients receive a suitable environment for medical treatment. Moreover, it could improve the satisfaction and quality of hospital service.
Key words: medical materials demand、MAPE、ARIMA、Back-propagation Neural Network, CART
|