Xinyue Zhang, Xinyi Sang, Beibei Liu, Quanyu Wang,. Methods and Applications: Developing Machine Learning Prediction Model for Daily Influenza Reported Cases Using Multichannel Surveillance Data - A City, Hubei Province, China, 2023~2025. China CDC Weekly, 2025, 7(44): 1396-1402
Introduction
Public health surveillance is crucial for decision-making. Given the significant threat of influenza to public health, developing predictive models using multichannel surveillance systems is imperative.
Methods
Data were collected from multichannel surveillance systems, including hospitals, search engines, and climatological and air pollutant surveillance systems, in a southern Chinese city from January 2023 to January 2025. Spearman’s correlation analysis assessed the relationships between variables and reported influenza cases. Several machine learning models were used to predict trends in reported cases.
Results
Correlation analysis showed that all four surveillance systems were related to influenza, with 27 variables correlated with daily reported cases. The Long Short-Term Memory model, established based on variables with the highest lagged correlations (5-day to 7-day lag) through combined surveillance systems, outperformed other models for 5-day forecasts (R2=0.92; mean absolute error=156.92; mean absolute percentage error=79.95%; root Mean Squared Error=292.33).
Conclusions
Data from various surveillance systems effectively track influenza epidemics. The model shows potential for infectious disease surveillance and epidemic preparedness.
Public health surveillance is crucial for decision-making. Given the significant threat of influenza to public health, developing predictive models using multichannel surveillance systems is imperative.
Methods
Data were collected from multichannel surveillance systems, including hospitals, search engines, and climatological and air pollutant surveillance systems, in a southern Chinese city from January 2023 to January 2025. Spearman’s correlation analysis assessed the relationships between variables and reported influenza cases. Several machine learning models were used to predict trends in reported cases.
Results
Correlation analysis showed that all four surveillance systems were related to influenza, with 27 variables correlated with daily reported cases. The Long Short-Term Memory model, established based on variables with the highest lagged correlations (5-day to 7-day lag) through combined surveillance systems, outperformed other models for 5-day forecasts (R2=0.92; mean absolute error=156.92; mean absolute percentage error=79.95%; root Mean Squared Error=292.33).
Conclusions
Data from various surveillance systems effectively track influenza epidemics. The model shows potential for infectious disease surveillance and epidemic preparedness.
See Also:
Latest articles in those days:
- The evolution and transmission dynamics of H3N2 influenza virus identified in a respiratory disease outbreak in companion dogs in Alabama, USA 7 hours ago
- Structural and functional characterization of hemagglutinin E190D and Q226H minor variants in human H5N1 A/British Columbia/PHL-2032/2024 7 hours ago
- Spatial risk factors associated with outbreaks of highly pathogenic avian influenza in Pennsylvania’s commercial poultry farms 7 hours ago
- [preprint]H9N2 avian influenza virus in Bangladesh: three genetically distinct co-circulating subgroups retain a largely conserved antigenic phenotype 7 hours ago
- [preprint]Endogenous serine proteases in MDCK sublines support trypsin-independent influenza A virus replication 7 hours ago
[Go Top] [Close Window]


