差分自回归移动平均模型在区县级公立医院门急诊量预测中的应用实践

Application of autoregressive integrated moving average model in prediction of outpatient and emergency visits in a district-level public hospital

目的利用差分自回归移动平均(autoregressive integrated moving average,ARIMA)模型精准预测区县级公立医院的门急诊量,为医院预算及运营决策提供重要依据。方法采集成都市双流区某公立医院2012年1月—2023年11月逐月的门急诊量,使用R 4.3.1软件,将2012年1月—2022年12月的逐月数据用于构建ARIMA模型,预测及验证2023年1月—11月的门急诊量。结果除2023年1、3月外,其他月份的预测门急诊量与实际门急诊量吻合较好,2023年1月—11月的平均绝对百分比误差为8.504%。2023年1月—11月的实际和预测门急诊总量分别为144.196万、141.713万人次,相对误差为–1.722%。结论ARIMA模型能较好地预测区县级医院门急诊量,但新型冠状病毒感染疫情高发等因素会影响短期预测的精准性。

Objective To accurately predict the outpatient and emergency visits of a district-level public hospital based on autoregressive integrated moving average(ARIMA)model,providing important basis for hospital budget planning and operational decisions.Methods The monthly outpatient and emergency visits of a public hospital in Shuangliu District,Chengdu City from January 2012 to November 2023 were collected,and R 4.3.1 software was used to establish an ARIMA model based on the data from January 2012 to December 2022.The outpatient and emergency visits from January to November 2023 were predicted and validated.Results Except for January and March 2023,every monthly number of predicted outpatient and emergency visits for 2023 matched the actual one relatively well.The average absolute percentage error for January to November 2023 was 8.504%.The actual total number of outpatient and emergency visits from January to November 2023 was 1441960,and the predicted value was 1417130 with a relative error of–1.722%.Conclusions ARIMA model can predict the outpatient and emergency visits of district-level hospitals relatively well.However,factors such as the high incidence of COVID-19 may affect the accuracy of short-term prediction.