Luzhao Feng

Human papillomavirus (HPV) infection is the primary cause of nearly all cervical cancers and a significant contributor to oropharyngeal, anal, and genital cancers. The global health burden of cervi...

Background: Influenza incidence trends are jointly influenced by multiple factors, making accurate forecasting challenging. Existing forecasting methods have difficulty simultaneously capturing multiscale disease dynamics, the complex effects of external environmental variables such as meteorological conditions, and dynamic interregional transmission relationships. We aimed to develop a unified spatiotemporal forecasting framework that addresses these limitations.Methods: We propose VMDP-Net (Variable-fused Multi-scale Dynamic Propagation Network), a spatiotemporal forecasting framework that integrates multiscale temporal features, meteorological factors, and interregional transmission relationships to jointly model short-term fluctuations, medium-term variations, and long-term trends in influenza activity. The framework comprises four modules: a Multi-scale Temporal Encoding (MTE) module that decomposes historical influenza sequences; a Multiple Variable Fusion (MVF) module that incorporates meteorological variables and enables adaptive selection of information relevant to the current epidemic stage; a Regional State Construction (RSC) module that builds unified regional node representations; and a Dynamic Mix-hop Propagation (DMP) module that adaptively learns dynamic interregional relations and captures both direct and indirect transmission effects. The model was evaluated on nine years of global influenza data, 23 years of influenza data from the United States, and nine years of influenza data from China.Findings: VMDP-Net achieved good forecasting accuracy and stability across different spatial regions, variable configurations, and forecasting horizons on all three datasets. The framework effectively captured the spatiotemporal heterogeneity of influenza transmission and the dynamic relationships among regions. In univariate forecasting, VMDP-Net achieved the lowest mean RMSE across the five horizons on all three datasets. In multivariate forecasting, the corresponding five-horizon mean RMSE was 7.8% lower on GlobalFlu and 4.5% lower on ChinaFlu.Interpretation: By jointly modelling multiscale temporal dynamics, meteorological drivers, and dynamic interregional transmission within a unified framework, VMDP-Net provides methodological support for real-time, coordinated public health early-warning systems.Funding: This study was supported by the National Natural Science Foundation of China (grant number 82473725).

Background: Severe fever with thrombocytopenia syndrome (SFTS) is a severe tick-borne viral disease characterized by dysregulated host immune responses. Although previous studies have suggested sex-related differences in immune responses in SFTS, the sex-specific magnitude of cytokine changes and their relationships with viral load and laboratory parameters have not been fully characterized. Methods: We investigated sex differences in cytokine profiles and their associations with viral load and disease severity in 100 confirmed SFTS patients and 60 healthy controls. Nineteen cytokines, viral load, and clinical and laboratory parameters were assessed in acute-phase sera. Correlation analyses were performed to assess the relationships of cytokine changes with viral load and clinical parameters. Multivariable regression analyses were used to evaluate the association between sex and cytokine changes after adjustment for potential confounders, and sex-stratified ROC analyses were conducted to assess the ability of cytokines to discriminate disease severity. Results: Cytokine levels were generally higher in SFTS patients than in controls and increased with disease severity. After normalization to sex-matched controls, females showed higher normalized cytokine ratios for CCL4, IL-17, IL-8, and FGF2, whereas males showed higher normalized ratios for IL-5 and PDGF-AB. Cytokine changes were associated with viral load in both sexes, but these associations did not differ significantly between males and females. After adjustment for potential confounders, FGF2 and IL-17 showed higher normalized cytokine ratios in females, whereas IL-5 and CXCL10 showed higher normalized ratios in males. Exploratory sex-stratified ROC analyses identified several cytokines with relatively high AUCs for discriminating concurrent mild and severe disease, including CXCL10 and CCL4. Conclusions: These findings provide preliminary indications of sex-related differences in cytokine responses during acute SFTS and suggest that the magnitude of cytokine changes may help better characterize sex-specific cytokine responses and their associations with disease severity.

Background: Traditional laboratory-confirmed influenza surveillance involves a 1- to 2-week reporting delay and captures only patients who have sought care and received a diagnosis, limiting early warning. Digital prescription data have been shown to signal influenza activity early, but their predictive value for forecasting remains unclear. Objective: We aimed to evaluate the predictive value of digital prescription data for influenza forecasting by assessing lead times, contributions across forecasting horizons, and added value when combined with other real-time sources. Methods: Using daily data for Jiangxi Province, China (2022-2024), we characterized temporal relationships for 13 multisource indicators using prewhitened cross-correlation and performed Granger-based predictor screening restricted to the 2022 to 2023 development period. Incidence was then forecast 1 to 14 days ahead through expanding-window rolling-origin validation trained on 2022 to 2023 data and tested on 2024 data. Six models (autoregressive integrated moving average [ARIMA], seasonal ARIMA [SARIMA], Prophet, least absolute shrinkage and selection operator [LASSO], random forest, and explainable boosting machine) plus a persistence baseline were compared across 3 exogenous-input scenarios (no external input, digital prescriptions alone, and selected sources). Each indicator was also forecast from its own history and compared with observed incidence. Results: After prewhitening, the digital prescription rate led influenza incidence by 14 days, whereas the online search index lagged incidence by 1 day, indicating near-synchronous tracking. In the development period Granger analysis, digital prescription rate, online search index, and temperature showed significant predictor-to-incidence associations after false discovery rate (FDR) correction (FDR-adjusted P <.001, P =.003, and P =.008, respectively), whereas nitrogen dioxide showed a marginal association (FDR-adjusted P =.11) and was additionally included as an environmental covariate. At 1 day, performance was similar across models and input scenarios ( R ² =0.925‐0.961; persistence R ² =0.948), but differences emerged at longer horizons. SARIMA performed best with the digital prescription rate as the sole exogenous input, and LASSO performed best with 4-source input, both reaching a mean R ² of 0.766 and root mean squared error (RMSE) of 0.240 across the 14 horizons. At 14 days, single-source SARIMA achieved an R ² of 0.510 (RMSE 0.318), and multisource LASSO achieved an R ² of 0.551 (RMSE 0.305), a gain of 0.064 in R ² and a 6.2% reduction in RMSE over LASSO without exogenous input. Forecasts of the digital prescription rate showed the closest temporal consistency with observed influenza incidence (mean r =0.897 vs 0.866 for the online search index). Conclusions: Digital prescriptions provide an early signal of provincial influenza activity, preceding routine reported incidence by approximately 2 weeks in this setting. Multisource input can further improve forecasting when paired with models capable of selecting informative signals, but indiscriminate integration does not necessarily help. This early signal may support earlier preparedness, including antiviral planning and health care resource allocation.

1. Influenza vaccination is a well-established, evidence-based intervention that has been shown to substantially reduce influenza-associated hospitalization (by approximately 36%) and all-cause mor...

Herpes zoster, a vaccine-preventable disease caused by the reactivation of varicella-zoster virus, has been linked to an increased risk of dementia in high-income countries. However, evidence from populations with low vaccination coverage remains limited. To address this gap, we conducted a nested case-control study using electronic health records from Yichang, China. A total of 51,843 incident dementia patients aged 50 y and older were matched with 338,877 controls by sex, age, and visit date. A documented clinical history of herpes zoster was significantly associated with an elevated risk of all-cause dementia (adjusted odds ratio [aOR] = 1.48; 95% confidence interval [CI]: 1.41-1.56), with the highest risk observed among individuals with herpes zoster involving the central nervous system (aOR = 1.59, 95%CI: 1.46-1.73). Among the dementia cases with prior HZ diagnosis, both Alzheimer's disease (aOR = 1.44, 95%CI: 1.14-1.79) and vascular dementia (aOR = 1.82, 95%CI: 1.49-2.21) showed increased risks following herpes zoster infection, with vascular dementia demonstrating a stronger association and more rapid progression (median time from first herpes zoster diagnosis to dementia onset: 1.5 y versus 2.2 y for Alzheimer's disease). The findings, derived from a population with minimal herpes zoster vaccination coverage, identify herpes zoster as a potentially modifiable risk factor for dementia. These also underscore the potential dual public health benefit of herpes zoster vaccination in preventing both acute infection and long-term cognitive decline.

Background: The post-pandemic era has witnessed potential shifts in the epidemiology of respiratory pathogens. Understanding the current age-specific patterns of pathogen detection and symptom reporting is crucial for refining public health surveillance and intervention strategies.Methods: Utilizing direct-to-consumer (DTC) testing data from four regions in China (October 2023-October 2025), we analyzed the epidemic patterns, temporal changes in age distribution of positive participants, and age-specific symptom reporting rates for seven respiratory pathogens in a community-based population of 975,107 participants. Generalized additive models (GAM) and spline regression were employed to analyze temporal and age-dependent trends.Findings: Temporal analysis revealed a significant shift of influenza A virus (IFV-A) and respiratory syncytial virus (RSV) toward younger age groups (average monthly decrease: -0.37%, -1.01%, respectively), whereas human rhinovirus (HRV) shifted toward older groups (average monthly increase: +0.27%). Symptom reporting was highly age-dependent. Fever reporting for IFV-A/influenza B virus (IFV-B)/human adenovirus (HAdV) showed an "L-shaped" decline with age, while for RSV/SARS-CoV-2, it exhibited a "V-shaped" pattern with a rebound in the elderly. Respiratory and pain symptoms predominantly followed an inverted "V-shaped" pattern, peaking in adulthood; however, respiratory symptom reporting in HAdV detection followed an inverted "L-shaped" pattern. Interpretation: The post-pandemic age distribution of respiratory pathogens shows divergent, pathogen-specific dynamics. The marked age dependency of symptom profiles, coupled with low classic symptom reporting in the elderly, suggests that current fever-centric surveillance case definitions may underestimate the disease burden in older adults. These findings highlight the necessity for age-stratified surveillance and optimized case definitions to guide targeted prevention and control measures.

Population movement significantly influences respiratory disease transmission; however, movement restrictions can impose substantial societal burdens. To understand spatiotemporal characteristic of a potential respiratory pandemic in Chinese mainland which could help in precise control, a spatiotemporal transmission model incorporating population movement dynamics was developed. The model was calibrated using Corona Virus Disease of 2019 data collected from an online survey of 3 million respondents conducted in December 2022. With the model, simulated hypothetical respiratory pandemics originating from each province and successfully predicted province-level transmission paths, peaking times and peaking infections. Beijing was identified as the most important location within the transmission network under various scenarios due to its strong mobility and socioeconomic connectivity. A global sensitivity analysis was conducted using the Partial Rank Correlation Coefficient (PRCC) method to evaluate the influence of key disease parameters on transmission dynamics. Transmission rate, progression rate and recovery rate were identified as key parameters influencing transmission characteristics. Additionally, the model provides a valuable predictive tool for understanding the transmission patterns of respiratory pandemics, which enables policymakers to gain insights into epidemic progression, facilitating the development of more targeted and effective control measures. Furthermore, our research offers a methodological framework for predicting epidemic transmission characteristics in advance, contributing to the field of public health.