差分—指数平滑预测模型在医院管理统计中应用的探讨  

Exploration on Application of Difference-Index Smoothing Predictive Model in the Managing Statistics of Hospital

作　　者：杨建南[1] 张常红[1] 

机构地区：[1]成都铁路中心医院,成都市610081

出　　处：《中国医院统计》1999年第4期198-200,共3页Chinese Journal of Hospital Statistics

摘　　要：目的 探讨差分—指数平滑预测模型在医院管理统计中的应用。方法 根据某院1977年～1996年门诊总诊疗人次历史资料,建立差分——指数平滑预测模型,对该院1997年总诊疗人次进行预测。结果 本例预测模型评价:①平均百分误差MPE=0.001％,接近于0;②平均绝对百分误差MAPE=7.97％<10％,说明本预测模型是无偏的,且为高精度预测模型。本例1997年总诊次点估计值为60.78万人次,区间预测值为53.30～68.26万人次,实际值为64.79万人次,确在预测区间内,与点估计值相对误差为6.19％<10％,实际预测效果满意。结论 差分—指数平滑预测模型经医院统计工作者在统计实践中应用,效果满意,是医院管理统计中值得推广应用的一种定量预测方法。To investigate the application of difference-index smoothing predictive model in themanaging statistics of hospital. Methods According to some hospital' s historical data of total person-times of diagnosed and treated outpatients from 1977 to 1996, we set up a difference-index smoothing predictive model and predicted the total person-times of diagnosis and treatments in 1997. Results The evaluation of the prediction model is as follows: (1)Mean percentage error(MPE) is 0. 001%, approaching 0. (2)Mean absolute percentage error(MAPE) is 7. 97%<10%, showing that the evaluation with high accuracy is unbiased. In 1997, the point evaluation of total person-times of diagnosis and treatment is 0. 607 8 million person-times, interval predictive value is 0, 533 0 to 0. 682 6 million person-times, practical value is 0. 647 9 million person-times. The relative error between the practical value and point evaluation in predictive interval is 6. 19%<10%, and predictive effects are satisfactory. Conclusion The method of prediction made by using the difference-index smoothing predictive model is a ration predicting method being worth to be extensively applied in hospital managing statistics.

关 键 词：差分-指数平滑 预测模型 门诊诊次 

分 类 号：R195.1[医药卫生—卫生统计学] R197.323[医药卫生—卫生事业管理]