Identifying Key Drivers of Foodborne Diseases in Zhejiang, China: A Machine Learning Approach

Area of Science:

• Public Health
• Epidemiology
• Data Science

Background:

• Foodborne diseases pose a significant global public health threat.
• Understanding pathogen-specific dynamics is crucial for effective control.

Purpose of the Study:

• To systematically analyze temporal dynamics, predictors, and seasonal patterns of pathogen-specific foodborne diseases.
• To leverage machine learning for risk prediction and public health interventions.

Main Methods:

• Analysis of 56,970 foodborne disease cases (2014-2023) from Zhejiang Province, China.
• Integration of epidemiological, environmental, socioeconomic, and agricultural data.
• Application of Lasso regression, Random Forest, and XGBoost for predictor identification and model training.
• Utilized Seasonal ARIMA and time-series decomposition for trend and seasonality analysis.

Main Results:

• Identified 41 important predictors, including patient age, food type, climate policy, and processing methods.
• Machine learning models achieved high accuracies (86-87%), demonstrating robust performance.
• Observed shifts in pathogen spectrum: declining Vibrio parahaemolyticus, increasing Salmonella post-2016, and seasonal peaks for Norovirus and Vibrio parahaemolyticus.
• Confirmed significant roles of seasonal and trend components in bacterial incidence.

Conclusions:

• Machine learning and time-series analysis are valuable tools for pathogen-specific foodborne disease surveillance.
• Findings support targeted public health interventions by highlighting key risk factors and temporal patterns.
• Understanding the interplay of host, exposure, and environmental factors is essential for mitigating foodborne disease risks.