High-resolution mapping of allergenic pollen risk across China using ensemble machine learning

Jie Yin ¹, Yuan Zhang ², Yifei Du ¹

⁰School of Public Health, Capital Medical University, Beijing 100069, China; Beijing Key Laboratory of Environment and Aging, Capital Medical University, Beijing 100069, China; Beijing Laboratory of Allergic Diseases, Beijing Municipal Education Commission, Beijing 100069, China; Laboratory for Clinical Medicine, Capital Medical University, Beijing 100069, China.

Ecotoxicology and Environmental Safety +

|

January 8, 2026

View abstract on PubMed

Summary

A new machine learning model provides the first nationwide, long-term daily pollen dataset for China, crucial for understanding airborne allergens and public health risks amid climate change.

Area Of Science

Environmental Science
Allergen Monitoring
Machine Learning Applications

Background

Airborne pollen is a significant allergen in China, with increasing prevalence linked to urbanization and climate change.
Accurate, long-term pollen data is vital for public health and ecological risk assessment.

Purpose Of The Study

To develop a novel ensemble machine learning framework for estimating daily tree and herbaceous pollen concentrations across China.
To reconstruct a nationwide, long-term daily pollen dataset from 2011-2023.

Main Methods

An ensemble model combining random forest and gradient boosting was developed.
Models were trained using monitoring site data and predictors like meteorological, vegetation, and land use variables.
Historical environmental data was used to reconstruct pollen concentrations from 2011-2023.

Main Results

The models demonstrated high accuracy (R² of 0.90 for tree, 0.89 for herbaceous pollen).
Nationwide daily pollen concentrations were reconstructed for 2011-2023.
Seasonal pollen peaks and key drivers like temperature and vegetation indices were identified.

Conclusions

This study presents the first nationwide, long-term daily pollen dataset for China.
The dataset is a valuable resource for ecological research, public health, and allergy forecasting.
The modeling framework supports climate-responsive risk management for airborne allergens.

Keywords:

Allergenic airborne pollen China Machine learning Spatiotemporal distribution

Related Concept Videos

Allergic Reactions

Overview

We speak of an allergy when the immune system triggers a response against a benign foreign structure, like food, pollen or pet dander. These elicitors are called allergens. If the immune system of a hypersensitive individual was primed against a specific allergen, it will trigger allergic symptoms during every subsequent encounter of the allergen. Symptoms can be mild, such as hay fever, to severe, such as potentially fatal anaphylactic shock.

Sensitization Is the First Step of an...