Meteorological Drivers of Influenza A and B Positivity in a Subtropical Chinese City: A Six-Year Surveillance Study Integrating Distributed Lag Non-Linear Models and Deep Learning
Abstract
Influenza transmission in subtropical regions is shaped by complex, non-linear meteorological conditions, while surveillance-based forecasting has been further complicated by fluctuating testing intensity and non-pharmaceutical interventions during the COVID-19 period. Existing approaches often either characterize lagged environmental associations without forecasting capacity or apply deep learning without clear epidemiological interpretability. Using six years (2018–2023) of multi-site influenza surveillance data from Putian, a subtropical coastal city in southeastern China, we developed a two-stage framework to characterize meteorological associations and improve short-horizon forecasting. Influenza positivity rates were used as the primary outcome to mitigate testing-related surveillance bias. Distributed lag non-linear models (DLNMs) were constructed to estimate subtype-specific, lagged associations between meteorological variables and influenza positivity, and a Bayesian-optimized long short-term memory (LSTM) network integrating meteorological, autoregressive, and socio-behavioral covariates was developed to forecast influenza A and B activity. Influenza A positivity was associated with warmer conditions, peaking at 15°C (relative risk [RR] = 3.01, 95% CI: 1.03-8.78) and narrow diurnal temperature ranges, whereas influenza B positivity was associated with colder (8°C; RR = 26.12, 95% CI: 7.79-87.61), more humid (91%; RR = 1.52, 95% CI: 1.00-2.29), and lower-radiation conditions. The LSTM showed lower forecast error than covariate-matched autoregressive baselines for both subtypes (Mean Absolute Error [MAE]: 0.009 vs. 0.136 for influenza A, and 0.002 vs. 0.049 for influenza B; Symmetric Mean Absolute Percentage Error [SMAPE]: 0.521 vs. 1.212, and 0.484 vs. 0.810, respectively). Covariate-ablation and SHapley Additive exPlanations (SHAP) analyses suggested that historical positivity and testing-related variables contributed materially to forecast stability under surveillance variability. External evaluation in Sanming provided preliminary evidence of model portability. This integrated DLNM-LSTM framework offers an interpretable approach for characterizing subtype-specific meteorological associations and improving influenza forecasting in subtropical settings.
Importance
Influenza imposes substantial global morbidity and mortality each year, and forecasting is particularly difficult in subtropical regions, where transmission occurs year-round, meteorological drivers are non-linear and lagged, and the COVID-19 period introduced pronounced epidemiological non-stationarity. Forecasts based on raw case counts are further affected by surveillance-related bias, as counts partly reflect testing intensity and clinician vigilance rather than underlying transmission alone. Existing work has largely applied inferential and deep-learning approaches in isolation, limiting either interpretability or forecasting flexibility. We address these issues with a two-stage DLNM–LSTM framework that uses influenza positivity rates as the primary endpoint and incorporates weekly detection volumes, mask-wearing stringency indices, day-and-week and non-local population proportion as socio-behavioral covariates, in order to mitigate testing-related surveillance bias and accommodate pandemic-era non-stationarity. The framework reveals distinct, subtype-specific meteorological associations for influenza A and B, shows lower forecast error than a covariate-matched autoregressive baseline, and provides preliminary evidence of portability to a second subtropical city. It offers an interpretable, climate-informed methodological foundation for future operational surveillance developments in subtropical influenza forecasting.
Related articles
Related articles are currently not available for this article.