急診室醫師排班問題之研究

本研究目的是為了探討南部某教學醫院急診部門醫師的排班問題，使得醫院能夠更有效率地處理急診室來診人數和滯留的病患人數。研究分成兩部份：(1)利用決策理論來評估彈性調整排班方案的優劣，和(2)建構適合時間序列模型來預測來診人數以及滯留人數。針對急診部門的不同科別，提出適當的調整方案，再利用統計方法決策理論中的平均絕對誤差(MAE)評估哪一方案的誤差最小，發現根據所設定的條件，三個科別皆選擇加班三小時決策。另外，建構來診量與滯留量共六組季節性ARIMA和計數資料時間序列模型來比較兩者預測的平均絕對誤差(MAE)，發現在預測來診人數與滯留人數方面，MAE差距不大；雖然季節性ARIMA的預測較優於計數型時間序列，但建構模型以及估計參數運算上比計數型時間序列來得慢。

關鍵字

計數資料時間序列模型；季節性ARIMA ；平均絕對誤差

並列摘要

Our purpose is to design a suitable schedule of doctors in emergency room and efficiently solve problem of manpower distribution for arrival patients and staying patients. This research is devided two parts, (1) Utilize decision theory to assess proposed flexible projects for schedule of doctors in emergency room, and (2) Establish suitable time series model to predict the number of arrival patients and the number of staying patients. For different departments of emergency room, we proposed some adjustment projects, and then assess which is better by mean absolute error by decision theory. The result showed that the best project is to extend three hours for three departments in emergency room under our setting conditions. Furthermore, we using the data of previous 12 months to establish time series models of the number of arrival patients and the number of staying patients, which have six models, and predict them on next month. We found the predicted error of SARIMA model is better than that of count time series model. But there are not too much different between two predicted errors, and the SARIMA model spent much time to fit model than using count time series model.

並列關鍵字

Count time series ； Seasonal ARIMA ； Mean Absolute Error

國際替代計量

急診室醫師排班問題之研究

全文下載

主題瀏覽