Jing Du

Background: Respiratory infectious disease surveillance increasingly requires timely integration of routine case reporting, laboratory pathogen surveillance, and digital behavioral signals. However, the incremental predictive value of spatially stratified regional pathogen activity and internet search data for short-term urban respiratory disease prediction remains unclear. Objective: This study aimed to develop and initially validate an exploratory multisource prediction framework for short-term prediction and early warning of respiratory infectious disease activity in Beijing by integrating reported case data, spatially stratified regional pathogen surveillance indicators, and Baidu index search data. Methods: We conducted a retrospective time-series modeling study using weekly reported respiratory infectious disease cases in Beijing, regional pathogen surveillance data from 31 provincial-level administrative regions in China, and Baidu index search data from January 2023 to December 2025. Pathogen indicators were aggregated by spatial strata and lagged by 0 to 4 weeks. Ten candidate LASSO (least absolute shrinkage and selection operator)–Poisson models were constructed to compare historical case terms, seasonal terms, local and regional pathogen indicators, and Baidu index indicators. One-week-ahead prediction was the primary task. Models were evaluated using a chronological train-test split and rolling-origin cross-validation for penalty parameter selection. Performance was assessed using root mean squared error (RMSE), mean absolute error (MAE), symmetric mean absolute percentage error (SMAPE), and RMSE reduction relative to the baseline historical case model. Results: The M10 core multisource model achieved the lowest overall RMSE among all candidate models, with an RMSE of 16,658.24; an MAE of 10,303.07; and an SMAPE of 0.766, corresponding to a 73.5% RMSE reduction relative to the M1 baseline model. The M3 regional pathogen model ranked second by RMSE, whereas the M7 Baidu index model achieved the lowest SMAPE, indicating better relative performance during low-burden periods. Selected predictors included regional overall pathogen positivity, influenza A positivity, coinfection-related indicators, historical case terms, and Baidu index disease name search indicators. Conclusions: The core multisource model improved the 1-week-ahead prediction of respiratory infectious disease case counts in Beijing, although relative prediction performance differed across models. Spatially stratified regional pathogen activity indicators, particularly overall pathogen positivity, influenza A positivity, and coinfection-related indicators, served as important supplementary signals, while Baidu index data provided complementary digital information. These findings support the feasibility of an exploratory multisource prediction framework in Beijing. Further prospective temporal validation and spatial validation in other megacities are needed before broader application. The core multisource model improved 1-week-ahead absolute case count prediction of respiratory infectious diseases in Beijing according to RMSE but did not achieve the lowest relative error according to SMAPE. These findings support the feasibility of an exploratory multisource prediction framework in Beijing. Further prospective temporal validation and spatial validation in other megacities or urban settings are needed before broader application.

Introduction: The hierarchical healthcare delivery system is an important measure to improve the allocation of medical resources and promote equitable distribution of basic medical and health services. It is one of the key factors in the success or failure of China’s medical reform. This study aims to analyze the factors influencing patients’ healthcare-seeking behaviors, including socioeconomic and clinical outcomes, under China’s hierarchical healthcare delivery system, and to provide potential solutions.Methods: Patients receiving outpatient treatment in the past 14 days and inpatient care in the past 1 year were investigated. The multivariate logistic regression was used to analyze the influencing factors of patient’s medical treatment behavior selection, and to compare whether the clinical outcomes of primary medical institutions and grade A hospitals are the same.Results: Nine thousand and ninety-eight person-times were included in the study. Of these, 4,538 patients were outpatients, 68.27% of patients were treated in primary medical institutions; 4,560 patients were hospitalized, 58.53% chose to be hospitalized in grade A hospitals. Provinces and cities, urban and rural areas, occupation, education level, medical insurance type, income, whether there are comorbid diseases, and doctors’ medical behavior are the factors affecting the choice of medical treatment behavior. Patients who choose primary medical institutions and grade A hospitals have different control levels and control rate for the blood pressure, blood lipids, blood glucose.Conclusion: Under the hierarchical diagnosis and treatment system, the patients’ choice of hospital is mainly affected by their level of education, medical insurance types, and the inpatients are also affected by whether there are comorbid conditions. Clinical outcomes of choosing different levels of hospitals were different.

Background: The impact of coronavirus diseases 2019 (COVID-19) related non-pharmaceutical interventions (NPIs) on influenza activity in the presence of other known seasonal driving factors is unclear, especially at the municipal scale. This study aimed to assess the impact of NPIs on outpatient influenza-like illness (ILI) consultations in Beijing and the Hong Kong Special Administrative Region (SAR) of China. Methods: We descriptively analyzed the temporal characteristics of the weekly ILI counts, nine NPI indicators, mean temperature, relative humidity, and absolute humidity from 2011 to 2021. Generalized additive models (GAM) using data in 2011–2019 were established to predict the weekly ILI counts under a counterfactual scenario of no COVID-19 interventions in Beijing and the Hong Kong SAR in 2020–2021, respectively. GAM models were further built to evaluate the potential impact of each individual or combined NPIs on weekly ILI counts in the presence of other seasonal driving factors in the above settings in 2020–2021. Results: The weekly ILI counts in Beijing and the Hong Kong SAR fluctuated across years and months in 2011–2019, with an obvious winter-spring seasonality in Beijing. During the 2020–2021 season, the observed weekly ILI counts in both Beijing and the Hong Kong SAR were much lower than those of the past 9 flu seasons, with a 47.5% [95% confidence interval (CI): 42.3%, 52.2%) and 60.0% (95% CI: 58.6%, 61.1%) reduction, respectively. The observed numbers for these two cities also accounted for only 40.2% (95% CI: 35.4%, 45.3%) and 58.0% (95% CI: 54.1%, 61.5%) of the GAM model estimates in the absence of COVID-19 NPIs, respectively. Our study revealed that, “Cancelling public events” and “Restrictions on internal travel” measures played an important role in the reduction of ILI in Beijing, while the “restrictions on international travel” was statistically most associated with ILI reductions in the Hong Kong SAR. Conclusions: Our study suggests that COVID-19 NPIs had been reducing outpatient ILI consultations in the presence of other seasonal driving factors in Beijing and the Hong Kong SAR from 2020 to 2021. In cities with varying local circumstances, some NPIs with appropriate stringency may be tailored to reduce the burden of ILI caused by severe influenza strains or other respiratory infections in future. Graphical

Background: In megacities, there is an urgent need to establish more sensitive forecasting and early warning methods for acute respiratory infectious diseases. Existing prediction and early warning models for influenza and other acute respiratory infectious diseases have limitations and therefore there is room for improvement. Objective: The aim of this study was to explore a new and better-performing deep-learning model to predict influenza trends from multisource heterogeneous data in a megacity. Methods: We collected multisource heterogeneous data from the 26th week of 2012 to the 25th week of 2019, including influenza-like illness (ILI) cases and virological surveillance, data of climate and demography, and search engines data. To avoid collinearity, we selected the best predictor according to the weight and correlation of each factor. We established a new multiattention-long short-term memory (LSTM) deep-learning model (MAL model), which was used to predict the percentage of ILI (ILI%) cases and the product of ILI% and the influenza-positive rate (ILI%×positive%), respectively. We also combined the data in different forms and added several machine-learning and deep-learning models commonly used in the past to predict influenza trends for comparison. The R2 value, explained variance scores, mean absolute error, and mean square error were used to evaluate the quality of the models. Results: The highest correlation coefficients were found for the Baidu search data for ILI% and for air quality for ILI%×positive%. We first used the MAL model to calculate the ILI%, and then combined ILI% with climate, demographic, and Baidu data in different forms. The ILI%+climate+demography+Baidu model had the best prediction effect, with the explained variance score reaching 0.78, R2 reaching 0.76, mean absolute error of 0.08, and mean squared error of 0.01. Similarly, we used the MAL model to calculate the ILI%×positive% and combined this prediction with different data forms. The ILI%×positive%+climate+demography+Baidu model had the best prediction effect, with an explained variance score reaching 0.74, R2 reaching 0.70, mean absolute error of 0.02, and mean squared error of 0.02. Comparisons with random forest, extreme gradient boosting, LSTM, and gated current unit models showed that the MAL model had the best prediction effect. Conclusions: The newly established MAL model outperformed existing models. Natural factors and search engine query data were more helpful in forecasting ILI patterns in megacities. With more timely and effective prediction of influenza and other respiratory infectious diseases and the epidemic intensity, early and better preparedness can be achieved to reduce the health damage to the population.

Background: Health care quality and insurance coverage have improved with economic development in China, but the burden of cardiovascular diseases (CVDs) continues to increase with ongoing gaps in prevention. We aimed to compare the uptake of secondary CVD prevention between stroke and coronary heart disease (CHD) patients in China. Methods: In a cross-sectional community-based survey of 47,841 adults (age ≥45 years) in 7 regions of China between 2014 and 2016, we identified those with a history of stroke or CHD to quantify disparities in conventional secondary CVD prevention strategies in multivariable logistic regression models. Results: There were 4,105 and 1,022 participants with a history of stroke and CHD, respectively. Compared to participants with CHD, those with a history of stroke were significantly less likely to be taking blood-pressure-lowering (39.7% vs. 53%), lipid-lowering (13.7% vs. 36.8%), and antiplatelet (20.8% vs. 50.6%) agents, at least one (48.9% vs. 70.8%) or all 3 recommended medicines (6.1% vs. 24.0%), and were less likely to achieve a lipid-cholesterol target (30.3% vs. 44.0%). Participants with a history of stroke achieved less optimal secondary prevention goals for medication use, either from any (adjusted odds ratio [aOR] 0.54, 95% confidence interval [CI] 0.44–0.66) or all 3 medications (aOR 0.27, 95% CI 0.20–0.36), as well as better blood pressure (aOR 0.81, 95% CI 0.66–0.98) and low-density lipoprotein cholesterol (aOR 0.34, 95% CI 0.27–0.43) levels of control. There were no significant differences in weight, smoking, or physical activity between the groups. Conclusion: Stroke patients had lower use of secondary CVD-preventive medication and achieved lower levels of risk factor control than those of CHD patients in China. Nationwide disease-specific strategies, and better education of participants and health care providers, may narrow these gaps.