高级检索
刘杨, 周浩月, 陆锦琪, 亓云鹏, 王远航, 贾磊, 查亦微, 周晚玲, 富小飞. 基于Prophet模型对流感样病例发病趋势的预测[J]. 疾病监测. DOI: 10.3784/jbjc.202306300314
引用本文: 刘杨, 周浩月, 陆锦琪, 亓云鹏, 王远航, 贾磊, 查亦微, 周晚玲, 富小飞. 基于Prophet模型对流感样病例发病趋势的预测[J]. 疾病监测. DOI: 10.3784/jbjc.202306300314
shu
Liu Yang, Zhou Haoyue, Lu Jinqi, Qi Yunpeng, Wang Yuanhang, Jia Lei, Zha Yiwei, Zhou Wanling, Fu Xiaofei. Prediction of incidence of influenza-like illness based on Prophet model[J]. Disease Surveillance. DOI: 10.3784/jbjc.202306300314
Citation: Liu Yang, Zhou Haoyue, Lu Jinqi, Qi Yunpeng, Wang Yuanhang, Jia Lei, Zha Yiwei, Zhou Wanling, Fu Xiaofei. Prediction of incidence of influenza-like illness based on Prophet model[J]. Disease Surveillance. DOI: 10.3784/jbjc.202306300314
shu

基于Prophet模型对流感样病例发病趋势的预测

Prediction of incidence of influenza-like illness based on Prophet model

    摘要
  • 摘要:
    目的  通过Prophet模型对流感样病例（ILI）发病趋势进行预测和评价，为流感性感冒（流感）防控工作及哨点医院监测工作提供新的思路和方法。
    方法  通过R 4.2.2软件的“forecast”“prophet”包构建Prophet模型，对2013—2022年浙江省嘉兴市流感哨点医院的ILI就诊人数占门急诊就诊总人数比例（ILI%）监测数据进行拟合和预测。
    结果  2013—2022年嘉兴市流感哨点医院年平均ILI%为2.97%，ILI%整体呈现波动变化，每年的冬春季会有明显的高峰，此后逐渐回落。 Prophet模型拟合和预测结果显示，2013—2021年嘉兴市流感哨点医院ILI%监测数据拟合趋势呈上升趋势，对2022年ILI%预测结果的平均预测误差率为−16.73%，国家级哨点监测医院ILI%监测数据拟合趋势呈现先上升后下降的趋势，对2022年ILI%预测结果的平均预测误差率为−14.85%，预测结果均在95%置信区间内。
    结论  Prophet模型拟合和预测的误差较低，拟合预测效果良好，可用于ILI发病趋势的预测。

     

    Abstract:
    Objective To predict and evaluate the incidence trend of influenza-like illness (ILI) by using Prophet model and provide new ideas and methods for influenza prevention and control and sentinel hospital surveillance.
    Methods A Prophet model was constructed using the “forecast” and “prophet” packages in software R 4.2.2 to fit and predict the annual average proportion of ILI outpatient visits in total outpatient and emergency visits (ILI%) by using the surveillance data from sentinel hospitals for influenza in Jiaxing, Zhejiang province, from 2013 to 2022.
    Results From 2013 to 2022, the average annual ILI% was 2.97% in sentinel hospitals in Jiaxing. The overall ILI% showed fluctuating patterns, with annual peak occurring during winte spring and a gradual decline thereafter. The fitting and prediction results of the Prophet model indicated that the fitted ILI% showed an upward trend in sentinel hospitals in Jiaxing from 2013 to 2021, and the average error rate of ILI% prediction was −16.73% for 2022. The ILI% fitted with surveillance data from national level sentinel hospitals first increased then decreased, with the average error rate of ILI% prediction as −14.85% for 2022, and all the prediction results were within the 95% confidence interval.
    Conclusion The Prophet model fitting and prediction has relatively low fitting and prediction error and show good effects, which can be used for the prediction of ILI incidence.

     

    • HTML全文
    • 参考文献(19)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回

在线交流