Spring and summer-autumn pollen grading and forecasting model based on daily visits of allergic rhinitis patients
10.3760/cma.j.cn115330-20240823-00491
- VernacularTitle:基于变应性鼻炎就诊人次的春季和夏秋季花粉分级和预报模型
- Author:
Yuhui OUYANG
1
;
Zhaoyin YIN
;
Yun YAN
;
Jingguo CHEN
;
Wenxuan FEI
;
Lili GONG
;
Weiwei LIU
;
Xiaojia LIU
;
Daoliang SONG
;
Zhendong XU
;
Ying ZHANG
;
Yuan ZHANG
;
Luo ZHANG
Author Information
1. 首都医科大学附属北京同仁医院过敏科和耳鼻咽喉头颈外科，北京 100730
- Publication Type:Journal Article
- Keywords:
Rhinitis, allergic;
Pollen;
Patient visits;
Grading threshold;
XGBoost algorithm;
Prediction model
- From:
Chinese Journal of Otorhinolaryngology Head and Neck Surgery
2025;60(3):313-320
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish graded forecast models of pollen concentration in spring and summer-autumn in northern China, based on long-term data of pollen and allergic rhinitis (AR) medical visits in 8 cities of northern China.Methods:Pollen concentration and the characteristics of AR patients from 8 cities of northern China, including Beijing, Baotou, Hohhot, Xi′an, Xining, Cangzhou, Liaocheng and Zibo, were analyzed. Spearman′s correlation was used to examine the relationship between pollen concentration and daily AR patient visits. A pollen concentration grading was establish, and a pollen forecast model was created using the eXtreme gradient boosting (XGBoost) algorithm. The model incorporated meteorological factors and the 3-day moving average of pollen concentrations.Results:The spring pollen period started early and lasted long in Beijing and Xi ′an, while the summer-autumn pollen period started earlier and persisted longer in Xining, Baotou and Hohhot. During summer-autumn pollen period, and the spring period in most cities (except Baotou and Cangzhou), average daily patient visits were significantly higher than those in non-pollen periods. A strong correlation was observed between daily AR patient visits and the 3-day moving average of pollen concentrations in both the spring and summer-autumn periods across all cities. Based on the correlation, a pollen concentration grading standard of northern China was established. The accuracy evaluation of pollen concentration prediction model showed that the percentage of forecasts with either completely accurate or within one level difference exceeded 91% in spring and 95% in summer-autumn. The most important predictive variable in the model was the pollen level from previous day, followed by the temperature and humidity.Conclusion:The grading prediction model for pollen concentration provides guidance for AR patients in term of travel, early defense and treatment, as well as the determining medication schedules for clinical drug research and specific immunotherapy.
