SARIMA在某三级精神专科医院医院感染发病预测中的应用

Application of SARIMA in predicting outbreaks of hospital-acquired infection in a tertiary psychiatric hospital

    摘要
  • 摘要:
    目的 通过收集某三级精神专科医院医院感染发病情况, 构建季节性差分自回归移动平均模型(SARIMA), 为预防与控制院内感染提供参考。
    方法 采用卫宁医院感染信息管理软件和纸质医院感染报卡收集某三级精神专科医院2016年1月-2024年8月各月医院感染发病率, 对2016-2023年发病率进行分析并建立SARIMA模型, 对2024年1-8月发病率进行预测, 根据实测值评估SARIMA模型预测准确性。
    结果 2016-2023年共收治患者98 075例, 其中936例患者发生医院感染, 医院感染发生率为0.95%, 发病率为0.79%~1.23%。2016-2023年时间序列图不满足序列平稳性要求, 对原始数据行差分、自相关图(ACF)和偏自相关图(PACF)分析及多次评估验证, 最终确定SARIMA(1, 1, 1)(1, 1, 1)12、SARIMA(1, 1, 1)(1, 1, 0)12、SARIMA(1, 1, 1)(0, 1, 0)12和SARIMA(1, 1, 1)(0, 1, 1)12 为备选模型, 采用经Ljung-Box Q检验保留P>0.05符合白噪声序列的模型和统计获得最小贝叶斯信息准则(BIC)值, 确定SARIMA(1, 1, 1)(0, 1, 1)12为最佳模型；经2024年1-8月医院感染发病率验证SARIMA(1, 1, 1)(0, 1, 1)12模型预测感染率始终落在预测值95%CI范围内, 预测准确性较高。
    结论 SARIMA模型可较好地预测某三级精神专科医院住院患者医院感染月发病率, 为预防和控制精神疾病住院患者医院感染的院感决策起到辅助作用。

     

    Abstract:
    OBJECTIVE To construct a Seasonal Autoregressive Integrated Moving Average (SARIMA) model based on the incidences of hospital-acquired infections (HAIs), and provide a reference for the prevention and control of HAI in such hospitals.
    METHODS The incidences of HAIs in a tertiary psychiatric hospital from Jan. 2016 to Aug. 2024 were collected by the Weining Hospital Infection Information Management software and paper-based HAI reporting cards. The incidence rates from 2016 to 2023 were analyzed and a SARIMA model was established. The incidence rates from Jan. to Aug. 2024 were predicted, and the accuracy of the SARIMA model was evaluated based on the actual measured values.
    RESULTS From 2016 to 2023, a total of 98, 075 patients were admitted, including 936 patients who developed HAIs, with an incidence rate of 0.95% ranged from 0.79% to 1.23%. The time series plot from 2016 to 2023 did not meet the requirements for sequence stability. After differentiating the original data, analyzing the correlation plot (ACF) and partial autocorrelation plot (PACF), and conducting multiple assessments and verifications, it was finally determined that SARIMA (1, 1, 1) (1, 1, 1)12, SARIMA (1, 1, 1) (1, 1, 0)12, ARIMA (1, 1, 1) (0, 1, 0)12, and SARIMA (1, 1, 1) (0, 1, 1)12 were the alternative models. The Ljung-Box Q test was used to retain the models with P> 0.05 that met the sequence with white noise and the minimum Bayesian Information Criterion (BIC) value was obtained, it was determined that SARIMA (1, 1, 1) (0, 1, 1)12 was the optimal model. When validated with Jan. to Aug. 2024 HAI incidence data, the infection rates predicted by SARIMA (1, 1, 1) (0, 1, 1)12 model remained within the 95% confidence interval, indicating high prediction accuracy.
    CONCLUSIONS ARIMA model can effectively predict the monthly HAI incidences in a tertiary psychiatric hospital, and it plays an role in the decision-making of HAI prevention and control in psychiatric inpatients.

     

    • HTML全文
    • 参考文献(22)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回
下载中心