气象因素对四川省手足口病发病率的影响及预测模型构建

杨雅斯; 卢亚陵; 方莅媛; 孙照华; 郑琳; 谢丽娇; 苟于婷; 赵云; 司翾; 丁显平

doi:10.12182/20200960106

气象因素对四川省手足口病发病率的影响及预测模型构建

Influence of Meteorological Factors on HFMD and Construction of Prediction Model in Sichuan Province

摘要

摘要:
目的分析2013−2019年四川省手足口病（hand foot and mouth disease, HFMD）的流行特征，研究气象学因素与HFMD发病率的相关性，并构建预测模型。
方法通过中国疾病预防控制中心和中国气象数据网收集2013−2019年四川省HMFD监测数据和气象数据。采用Spearman相关分析探究HFMD月发病率与气象学因素的关系，同时采用多元回归模型和支持向量回归（support vector regression, SVR）模型分别构建HFMD月发病率预测模型。
结果 2013−2019年四川省累计报道HFMD 615 840例，死亡病例81例，年均发病率107.31/10⁵，死亡率0.16/10⁶。Spearman相关分析显示月平均相对湿度（r=0.342）、月平均气温（r=0.284）、月平均水汽压（r=0.304）、月平均降水天数（r=0.259）与HFMD月发病率呈正向弱相关。多元回归模型预测结果（R²=0.375）准确性不如SVR模型（R²=0.836），SVR模型对2013−2018年HFMD月发病率有较好的拟合，能较好地预测2019年发病高峰。
结论月平均相对湿度对HFMD月发病率影响最大；SVR模型的拟合值与实际值吻合度较高，对四川省HFMD发病情况预测具有一定价值。

Abstract:
Objective To analyze the epidemiological characteristics of hand, foot and mouth disease (HFMD) in Sichuan Province from 2013 to 2019. To study the correlation between meteorological factors and the incidence of HFMD and construct a prediction model.
Methods The HMFD surveillance data and meteorological data from 2013 to 2019 in Sichuan Province were collected through the Chinese Center for Disease Control and Prevention and the China Meteorological data Network. Spearman correlation was used to analyze the relationship between HFMD incidence and meteorological factors. Multiple regression model and support vector regression (SVR) model were used to construct HFMD incidence prediction models respectively.
Results A total of 615 840 cases of HFMD and 81 deaths were reported from 2013 to 2019. The average annual incidence rate was 107.31/10⁵, and the mortality rate was 0.16/10⁶. Spearman correlation analysis showed that the monthly incidence rate of HFMD was correlated with monthly average relative humidity (r=0.342), monthly average temperature (r=0.284), monthly average water vapor pressure (r=0.304) and monthly average days of precipitation (r=0.259). The prediction effect of the SVR model (R²=0.836) was better than the multiple regression model (R²=0.375). The SVR model provided a good fit to the monthly incidence of HFMD from 2013 to 2018, and can predict the peak incidence of HFMD in 2019.
Conclusion Relative humidity has the greatest influence on the incidence of HFMD. The fitting value of SVR model is in good agreement with the actual value, which is valuable in predicting the incidence of HFMD in Sichuan Province.

HTML全文

参考文献(24)

施引文献

资源附件(0)