【作者】 卢文丽；

【导师】 王伟；

【作者基本信息】 天津医科大学 ， 流行病与卫生统计学， 2002， 硕士

【摘要】 随着国内医疗保险制度的实施，中国加入WTO等时事的发展，对于医院管理工作提出了更高要求，充分利用时间序列资料进行统计分析预测，对于医院管理的科学化，规范化和现代化具有重要意义。本文旨在对医院统计工作中的季节性时序资料的预测方法作一探讨。 在本次课题研究过程中，主要应用X-11季节调整方法和Box-Jenkins季节模型预测法对天津市中西医结合医院(南开医院)的季节性时序资料和《中国医院统计》1999年第6卷第4期的“ARIMA模型在医院卫生消耗材料需求量预测中的应用”一文中的X光片消耗量的数据资料进行预测，并将预测结果和预测误差与目前医院统计工作中常用的季节周期回归模型的预测结果和预测误差进行比较。 BOX-Jenkins模型在分析处理医院统计工作中的季节性时序资料时显示了明显的优势：通过ARIMA模型的拟合，可以综合考虑序列的长期趋势、季节变动、周期变化及随机干扰，并借助模型参数进行量化表达，最终使拟合的残差不再包含可供提取的非随机成分，成为白噪声(White noise)或近似白噪声序列，拟合效果的验证在理论上具有说服力，在应用上切实可行。 Box-Jenkins模型可以给出预测结果的可信区间，可以结合实际初步确定模型的有效预测期限，使对于预测期限的把握具有理论上的客观依据。Box-Jenkins模型发展的比较完善，对于模型的识别，拟合和诊断性检验均有较为成熟的方法，而且该方法有发展成熟的统计软件支持，如SAS，这对于该方法的推广应用非常有利。 医院季节性时间序列资料预测方法探讨 X叫1季节调整预测法先应用X刁 季节调整方法将原始数 据分解为趋势、季节和随机因子序列，然后将各序列的预测结 果相加。该方法对于各类因子的合理分解，使季节因素对于医 院出入院人次影响的规律性明朗化，使趋势性也跃然而出，达 到将复杂的问题分解为简单的子结构的目的，为进一步的分析 提供了更便捷的途径，对于指导医院管理工作中住院医师的安 排，卫生消耗材料的配备以及药品的采买等都具有重要的指导 意义，参考长期趋势变动还可以为医院制订长期规划提供依据。 季节周期回归模型采用回归的形式，利用原始数据资料建 立回归方程，数据的长期变动、季节变动和循环变动均包含在 方程中。该模型在X11季节调整有效的出入院人次预测中，效 果较好，而且在其他文献中也收到了较好的效果’“－“’，这说明 该方法是统计预测工作的经验总结，具有一定的实用性。 实践证明对于医院管理工作中常见的时间序列资料，根据 统计理论和数学方法，建立描述该序列适应的或最优的数学模 型，并进而进行预测和控制，是可能的、合理的、科学的。在 医院管理数据的统计预测实践中，影响预测方法选择的因素还 很多，还有很多问题有待于今后进一步的研究学习和总结c更多还原

【Abstract】 With the implement of the hospitalization insurance and the affiliation to WTO, the managers of hospital would face more challenge. It is helpful for realization the scientific management to analyze and to forecast the time series data of hospital. The objective is to investigate the most appropriate forecasting model for seasonal time series of hospital.In this paper X-ll seasonal adjustment method and Box-Jenkins seasonal models were selected for analyzing and forecasting the seasonal data. And in this paper one other paper was used to provide the data of X file wastage data. All the output and error were compared with the actual value and the error of the seasonal regression model.Box-Jenkins seasonal models show obvious advantage. The trend, the seasonal variation and the irregular were all taken into account in this model. The parameters were estimated and the residuals series was white noise. The modeling and the forecasting was feasible and had adequately theoretic basis. Box-Jenkins seasonal models also offered the forecasting confidence interval. The identification, the estimation and the forecast had mature method and SAS provide the procedure for identification, estimation and forecast.The two seasonal data were successfully adjusted by the X-ll seasonal adjustment method. The primitive seasonal data were decomposed into trend series, seasonal variation series and irregular series. Though the forecasting effect was not better thanthe effect of Box-Jenkins seasonal models, the infection of season appeared more clearly and the trend did also. It is significance to arrange physician, to prepare sanitary material and to stock drug. The managers could consult the trend time series to project the future also.The seasonal regression model was also used in this paper. The equation of the seasonal regression model were established, and the trend, the seasonal variation, the circular change were all estimated in each equation. In this paper, the method gained better forecasting effect. That is to say, the seasonal regression model is practical method in hospital statistics forecast.The practice of this paper proved it is feasible and reasonable to forecast the future data using the past time series with the appropriate method or model. Because there are many factors that affect the selection of forecasting method, in the future, there would be many question need further study.更多还原