Short-Term Lag Effects of Climate-Pollution Interactions on Cardiopulmonary Hospitalizations: A Multi-City Predictive...

Area of Science:

• Environmental Health
• Epidemiology
• Artificial Intelligence

Background:

• Climate change and air pollution pose significant public health challenges.
• Understanding the short-term health impacts of environmental factors is crucial for timely interventions.

Purpose of the Study:

• To evaluate the lag effects of climate and air pollution on hospital admissions for cardiovascular diseases (CVDs) and respiratory diseases (RDs).
• To develop and assess deep learning models for predicting daily hospital admissions.

Main Methods:

• A multi-city study in Japan (Tokyo, Osaka, Nagoya).
• Random forest models to analyze lag effects (lag0, lag3, lag7) of environmental factors on hospitalizations.
• Hybrid Autoencoder-Long Short-Term Memory (AE+LSTM) deep learning models for prediction.

Main Results:

• Air pollutants, especially nitrogen oxides (NOx), showed immediate strong effects (lag0) on CVD and RD admissions.
• Temperature and precipitation became stronger predictors at longer lags (lag3, lag7).
• The AE+LSTM model significantly improved prediction accuracy for RD (32.4%) and CVD (20.94%) compared to standard LSTM.

Conclusions:

• Environmental exposures present dynamic, time-varying health risks.
• Deep learning models offer a powerful tool for predicting short-term hospital admissions.
• This predictive framework can enhance public health preparedness and resource allocation.