时间序列模型在预测平均住院日的应用

Application of Time Series Model in Meticulous Management of Average Hospital Stay

目的 运用时间序列模型对科室平均住院日进行预测,制定考核指标进行绩效管理,以提高指标精细化管理水平。方法 选取某三甲综合医院感染科作为试点科室,以2014年1月1日-2017年12月31日共5523例出院患者的月度平均住院日建立ARIMA模型,使用2018年平均住院日检验模型,以月为周期,动态调整ARIMA模型,对2019年科室指标进行预测。结果 建立ARIMA模型,残差均为白噪声序列,预测值与实际值平均相对误差为9.61%,且预测值均在95%可信区间内。利用该模型对感染科制定考核指标进行绩效考核,2019年平均住院日较2018年下降0.35天,P=0.0109,有统计学意义。结论 时间序列模型对平均住院日的拟合预测效果较好,能为科室指标精细化管理提供有效的数据支持,方便医院管理层合理调整管理方式,提升管理水平。

Objectives Time series model was used to predict the average length of stay in the department,and the assessment indicators were formulated for performance management,so as to improve the level of index meticulous management.Methods The infection Department of a Three A and Tertiary General Hospital was selected as a pilot department,and the ARIMA model was established based on the monthly average length of stay of 5523 discharged patients from January 1 st,2014 to December 31 st,2017.The average length of stay test model in 2018 was used as a monthly cycle to predict the department indicators in 2019.Results The residual errors of ARIMA model were all white noise sequences,and the average relative error between predicted value and actual value was 9.61%,and the predicted values were all within 95% confidence interval.The model was used to formulate assessment indicators for the department of Infectious Diseases,and the average hospital stay in 2019 decreased by 0.35 days compared with that in 2018,which was statistically significant.Conclusions The time series model had a good fitting and prediction effect on the average length of stay,which could provide effective theoretical data support for the fine management of department indicators,and facilitate the hospital management to adjust the management mode reasonably and improve the management level.