eLife · 20 Nov 2026
Integrating distributed lag non-linear models and deep learning reveals meteorological drivers of influenza A and B positivity in a 6-year subtropical surveillance study
Long Xieh-index 18, Meng-Jie Zhang, Jin-Lin Tan, Yi-Xin Ling, Zhe-Qiang Xue, Zhi-Shen Wu, Jun-Ju Huang, Jian-Ling Chen, Ze-Fan Ruan, Jing Qian, Hai-Yong Pan, Xiao Han, Sheng Xiong, Long-Mei Ling, Xi-Wen Jiangh-index 8
Citations by database
Each database indexes a different set of journals, preprints and books, so they are shown side by side rather than blended.
- Crossref1
- OpenAlex1
- OpenCitations0
- Europe PMC0
Citations per year
The current year is still counting.
- 2026 so far0
Abstract
Subtropical influenza forecasting is complicated by non-linear meteorological dynamics and surveillance bias stemming from pandemic-era testing fluctuations. Existing approaches either characterize lagged environmental associations without forecasting capacity or apply deep learning without epidemiological interpretability. We analyzed multisite influenza surveillance data from Putian, southeastern China (2018–2023); test positivity was the primary outcome to mitigate testing-related bias. Distributed lag nonlinear models estimated type-specific associations between eight meteorological variables and positivity over a 15-day lag window. A Bayesian-optimized long short-term memory (LSTM) network incorporating meteorological, autoregressive, and socio-behavioral covariates was benchmarked against covariate-matched multivariate autoregressive integrated moving average (ARIMA) and extreme gradient boosting (XGBoost) models. For influenza A, the temperature-associated relative risk (RR) peaked at 15°C (3.01; 95% confidence interval [CI]: 1.03–8.78), and positivity was associated with narrow diurnal temperature ranges. Influenza B positivity was associated with colder conditions (8°C; RR = 26.12, 95% CI: 7.79–87.61), higher humidity (91%; RR = 1.52, 95% CI: 1.00–2.29), and lower solar radiation. The LSTM had lower mean absolute error for influenza A (0.009 vs 0.136 for ARIMA and 0.138 for XGBoost) and B (0.002 vs 0.049 and 0.070); symmetric mean absolute percentage error was 0.521 vs 1.212 and 1.081 for A and 0.484 vs 0.810 and 0.737 for B. Covariate ablation and Shapley additive explanations analyses suggested that historical positivity and testing-related variables contributed to forecast stability. External validation in Sanming provided preliminary evidence of regional portability. This framework combines interpretable meteorological associations with short-horizon forecasting under variable surveillance conditions.
What the data shows
- Crossref counts 1 citations, the most of the databases. The others count 0 to 1.
- Long Xie has the highest h-index among the authors (18).
Graph
This paper in the middle, its authors and their most-cited papers below, the papers it cites to the left and the papers citing it to the right. The timeline lays every paper out by year.
Select a node for its details. Drag to pan, scroll to zoom, drag a node to move it.
Start with
Cited by 0 papers
Who cites this paper, from open citation data, with Crossref's record of each; the 100 newest when there are more.
References (56)
What it cites, from open citation data, with Crossref's record of each.
Sign in to see 36 more references
A free account opens the rest of the citing papers and references on every paper’s page.