Bayesian Multivariate Model Research Articles

An examination of the effect of area-level characteristics on juvenile justice and child welfare referrals using multivariate Bayesian spatial modeling.

Neighborhood disadvantage is linked to a higher risk of referrals to child welfare and juvenile justice systems. While past research has explored these associations independently, no study has concurrently examined the spatial overlap of child maltreatment and juvenile justice involvement. We examine the spatial overlap of involvement in juvenile justice and child welfare systems to identify areas of shared risk. Youth who received either a juvenile justice or child welfare referral in New Mexico between 2008 and 2015 aggregated to census tracts. We examined the spatial overlap of child welfare and juvenile justice involvement using multivariate Bayesian spatial modeling. Results show a significant positive association between juvenile justice and child maltreatment referrals across neighborhoods. After adjusting for residential instability, immigrant concentration, and residential racial segregation, children in the least deprived 20% of neighborhoods were 95.2% and 55.5% less likely to be referred to child welfare or justice systems, respectively, compared to those in the most deprived 20%. Our findings highlight the value of geospatial analyses to guide public health interventions by targeting the shared overlapping risk factors associated with neighborhoods with high risk for both child welfare and juvenile justice system involvement.

Delaying bud-break on pecan trees: a Bayesian longitudinal multinomial regression approach

A multivariate Bayesian Probit model is adapted to analyze a longitudinal multiclass-ordinal response, with a linear plateau as the longitudinal model. Measurements on pecan bud growth were collected on irregular time intervals, about a week apart from late March to mid April, using a six-level ordinal scale. The data are from two randomized complete block designs with four blocks each. The experiments were setup and initiated in 2018 in a pecan orchard, at two different locations, to evaluate the effect of two sets of four treatments on delaying growth of recently broken pecan buds to minimize bud loss due to low temperatures. A simulation study was successfully carried out to validate the model implementation. Treatment 3 of Experiment 1 was associated with the greatest reduction in bud growth rate. In Experiment 2, Treatments 2 and 3 had some effect on delaying bud growth. Although treatment effects were not statistically different in either experiment, this paper presents a practical and efficient modeling technique for longitudinal multinomial ordinal data, a common data type in applied agricultural research studies.

Issues Associated With the Formulation of a Small Area Model for Estimation of State-Level Crime Victimization Rates

Subnational estimation is an important and challenging problem in the context of the National Crime Victimization Survey (NCVS). Direct estimates for subnational domains are often unreliable due to small sample sizes. Model-based procedures have potential to improve upon direct estimators. However, model-based estimation presents several new challenges. One must identify suitable covariates, and an appropriate model form must be specified. This article discusses issues associated with the formulation of a small area model for production of state-level estimates of crime victimization rates. An analysis of direct estimates and covariates motivates the development of a Bayesian multivariate model. A model in the original scale is compared to a model in the log scale. Efficiency gains from the model relative to the direct estimators are examined. One challenge is that direct estimates can equal zero. Two ways of handling zero direct estimates are discussed.

Evaluating MCMC Convergence in a Bayesian Model of ST-Elevation Myocardial Infarction Female Patients in Malaysia

The majority of research that looked at the Bayesian Markov Chain Monte Carlo (MCMC) approach for prognostic modelling of cardiovascular disease only focused on the use of the Bayesian approach in variable selection, model selection, and prior distribution selection. But very few of this research has looked at the Markov chains' convergence in the model. In this study, convergence diagnostics were carried out to evaluate the convergence of Markov chains using both visual inspection and additional diagnostics. The National Cardiovascular Disease Database-Acute Coronary Syndrome (NCVD-ACS) registry, which included 1248 female patients with ST-Elevation Myocardial Infarction (STEMI) between 2006 and 2013, was used for this study's analysis. The multivariate Bayesian model identified six significant variables: dyslipidaemia, myocardial infarction, smoking, renal disease, Killip class, and age group. The trace plots did not reveal any distinctive patterns based on these significant variables, and the model's MCMC mixing is typically good. While for the autocorrelation plots, mild autocorrelations for age group, Killip IV, as well as the intercept term in the model. Since there were only mild autocorrelations, no thinning is needed. Also, the Geweke diagnostic showed that the chain is divided into two windows containing a set fraction of the first and last iterations which produced standard Z-scores. The Geweke diagnostic did not provide evidence of non-convergence, as none of the Z-scores fell in the extreme tails of the N (0,1). In this study, a number of plots and additional diagnostic tools showed that the Markov chains have reached convergence, which is relevant to the general use of the MCMC approach.

A Bayesian multivariate hierarchical model for developing a treatment benefit index using mixed types of outcomes

BackgroundPrecision medicine has led to the development of targeted treatment strategies tailored to individual patients based on their characteristics and disease manifestations. Although precision medicine often focuses on a single health outcome for individualized treatment decision rules (ITRs), relying only on a single outcome rather than all available outcomes information leads to suboptimal data usage when developing optimal ITRs.MethodsTo address this limitation, we propose a Bayesian multivariate hierarchical model that leverages the wealth of correlated health outcomes collected in clinical trials. The approach jointly models mixed types of correlated outcomes, facilitating the “borrowing of information” across the multivariate outcomes, and results in a more accurate estimation of heterogeneous treatment effects compared to using single regression models for each outcome. We develop a treatment benefit index, which quantifies the relative benefit of the experimental treatment over the control treatment, based on the proposed multivariate outcome model.ResultsWe demonstrate the strengths of the proposed approach through extensive simulations and an application to an international Coronavirus Disease 2019 (COVID-19) treatment trial. Simulation results indicate that the proposed method reduces the occurrence of erroneous treatment decisions compared to a single regression model for a single health outcome. Additionally, the sensitivity analyses demonstrate the robustness of the model across various study scenarios. Application of the method to the COVID-19 trial exhibits improvements in estimating the individual-level treatment efficacy (indicated by narrower credible intervals for odds ratios) and optimal ITRs.ConclusionThe study jointly models mixed types of outcomes in the context of developing ITRs. By considering multiple health outcomes, the proposed approach can advance the development of more effective and reliable personalized treatment.

Applying a Bayesian Spatiotemporal Model to Examine Excess County-Level Cardiovascular Disease Death Rates During the COVID-19 Pandemic.

Amid the COVID-19 pandemic, national cardiovascular disease (CVD) death rates increased, especially among younger adults. County-level variation has not been documented. Using county-level CVD deaths (ICD-10 codes: I00-I99) from the US National Vital Statistics System, we developed a Bayesian multivariate spatiotemporal model to estimate excess CVD death rates in 2020 based on trends from 2010-2019 for adults aged 35-64 and ≥65 years. Among adults aged 35-64 years, 64.7% of counties experienced significant excess CVD death rates. The median county-level CVD death rate in 2020 was 150 per 100,000 persons, which exceeded the predicted rate for 2020 (median excess death rate: 11 per 100,000; median excess rate ratio: 1.08). Among adults aged ≥65 years, 15.2% of counties experienced significant excess CVD death rates. The median county-level CVD death rate was 1,546 per 100,000 in 2020, which exceeded the predicted rate in 2020 (median excess death rate: 48 per 100,000, median excess rate ratio: 1.03). Counties with significant excess death rates in 2020 were geographically dispersed. In 2020, disruptions of county-level CVD death rates were widespread, especially among younger adults, suggesting the continued importance of CVD prevention and treatment in younger adults in communities across the country.

Bayesian multivariate receptor model and convolutional neural network to identify quantitative sources and spatial distributions of potentially toxic elements in soils: A case study in Qingzhou City, China

Determining sources and spatial distributions of potentially toxic elements (PTEs) is a crucial issue of soil pollution survey. However, uncertainty estimation for source contributions remains lack, and accurate spatial prediction is still challenging. Robust Bayesian multivariate receptor model (RBMRM) was applied to the soil dataset of Qingzhou City (8 PTEs in 429 samples), to calculate source contributions with uncertainties. Multi-task convolutional neural network (MTCNN) was proposed to predict spatial distributions of soil PTEs. RBMRM afforded three sources, consistent with US-EPA positive matrix factorization. Natural source dominated As, Cr, Cu, and Ni contents (78.5 %∼86.1 %), and contributed 37.1 %, 61.0 %, and 65.9 % of Cd, Pb, and Zn, exhibiting low uncertainties with uncertainty index (UI) < 26.7 %. Industrial, traffic, and agricultural sources had significant influences on Cd, Pb, and Zn (30.2 %∼61.9 %), with UI < 39.3 %. Hg originated dominantly from atmosphere deposition (99.1 %), with relatively high uncertainties (UI=87.7 %). MTCNN acquired satisfactory accuracies, with R2 of 0.357–0.896 and nRMSE of 0.092–0.366. Spatial distributions of As, Cd, Cr, Cu, Ni, Pb, and Zn were influenced by parent materials. Cd, Hg, Pb, and Zn showed significant hotspot in urban area. This work conducted a new approach exploration, and practical implications for soil pollution regulation were proposed.

Peptides and primate personality: Central and peripheral oxytocin and vasopressin levels and social behavior in two baboon species (Papio hamadryas and Papio anubis)

The neurohormones oxytocin (OT) and arginine vasopressin (AVP) are involved in social behaviors and psychiatric conditions. However, more research on nonhuman primates with complex social behaviors is needed. We studied two closely-related primate species with divergent social and mating systems; hamadryas baboons (Papio hamadryas, n=38 individuals) and anubis baboons (Papio anubis, n=46). We measured OT in cerebrospinal fluid (CSF, n=75), plasma (n=81) and urine (n=77), and AVP in CSF (n=45), and we collected over 250 hours of focal behavioral observations. Using Bayesian multivariate models, we found no clear species difference in hormone levels; the strongest support was for hamadryas having higher CSF OT levels than anubis (posterior probability [PP] for females = 0.75, males = 0.84). Looking at nine specific behaviors, OT was associated with affiliative behaviors (approach, proximity, grooming, PP ∼ 0.85 – 1.00), albeit inconsistently across sources of measurement (CSF, plasma, and urine, which were uncorrelated with each other). Most behaviors had low repeatability (R ∼ 0 – 0.2), i.e. they did not exhibit stable between-individual differences (or “personality”), and different behaviors did not neatly coalesce into higher-order factors (or “behavioral syndromes”), which cautions against the use of aggregate behavioral measures and highlights the need to establish stable behavioral profiles when testing associations with baseline hormone levels. In sum, we found some associations between peptides and social behavior, but also many null results, OT levels from different sources were uncorrelated, and our behavioral measures did not indicate clear individual differences in sociability.

Adolescent Cultural Identity Development in Context: The Dynamic Interplay of the Identity Project With Classroom Cultural Diversity Climate in Italy and Germany

While both the classroom cultural diversity climate and curriculum-based interventions can promote cultural identity development, they have not been studied together. Drawing on theories of ethnic-racial identity development, the current study aimed to understand the dynamic interplay of a curriculum-based intervention (the Identity Project) with the classroom cultural diversity climate (heritage culture and intercultural learning, critical consciousness socialization and equal treatment) on cultural identity exploration and resolution. Our sample included 906 mid-adolescents in Italy (32.36% immigrant descent, Mage (SD) = 15.12 (0.68) years, 51.73% female), and 504 early adolescents in Germany (53.86% immigrant descent, Mage (SD) = 12.82 (0.89) years, 42.37% female). Bayesian multivariate linear models show that the Identity Project and a stronger critical consciousness climate in the classroom before the intervention promoted cultural identity exploration at post-test in both countries. However, effects of the intervention and facets of the diversity climate on subsequent resolution were only observed in Italy. There was some evidence that the intervention could alter the classroom cultural diversity climate in Germany, while the intervention could compensate for a less positive diversity climate in the slightly older sample in Italy. Thus, it seems promising to systematically build in opportunities to engage with students’ diverse heritage cultures and identities when developing new curricula, as well as to train teachers to implement such curricula.

Multivariate Bayesian Time-Series Model with Multi-temporal Convolution Network for Forecasting Stock Market During COVID-19 Pandemic

The paper proposes a hybrid algorithm for forecasting multiple correlated time-series data, which consists of two main steps. First, it employs a multivariate Bayesian structural time series (MBSTS) approach as a base step. This method allows for the incorporation of potentially high-dimensional regression components, and it utilizes spike and slab priors to identify a parsimonious model. Second, the algorithm includes a post-model fitting diagnostic step where the residuals from the MBSTS step are processed through a multi-input/output temporal convolutional network (M-TCN) with multiple time scale feature learning. This step serves as an alternative to traditional subjective residual-based diagnostic procedures in time-series analysis, with the aim of improving forecasting accuracy. The key advantage of the M-TCN is its ability to capture sequential information efficiently. The M-TCN expands the field of convolution kernel without increasing the number of parameters, thus enhancing the capacity of model to capture complex sequential patterns. The paper presents two applications showcasing the effectiveness of the proposed hybrid algorithm. First, it utilizes pre-lockdown data from eleven Nifty stock sectoral indices to predict stock price movements, including the initial post-lockdown upturn. In the second application, it focuses on stock market data from pharmaceutical companies involved in manufacturing COVID-19 vaccines. In both cases, sentiment data sourced from newspapers and social media serve as the regression component. Through rigorous analysis, the paper demonstrates that the hybrid model outperforms various benchmark models, including LSTM, Bidirectional Encoder Representations from Transformers (BERT)-based LSTM, Deep Transformer Model, and GRU, among others, in terms of forecasting accuracy. This underscores the utility of the hybrid algorithm, particularly in predicting stock market trends during the COVID-19 pandemic period and its associated market dynamics.

Maternal investment and early thermal conditions affect performance and antipredator responses.

Exposure to increased temperatures during early development can lead to phenotypic plasticity in morphology, physiology, and behavior across a range of ectothermic animals. In addition, maternal effects are known to be important contributors to phenotypic variation in offspring. Whether the 2 factors interact to shape offspring morphology and behavior is rarely explored. This is critical because climate change is expected to impact both incubation temperature and maternal stress and resource allocation. Using a fully factorial design, and Bayesian multivariate mixed models, we explored how the manipulation of early thermal environment and yolk-quantity in eggs affected the morphology, performance, and antipredator behavior of 2 sympatric Australian skink species (Lampropholis delicata and L. guichenoti). We found that juveniles from the hot treatment were larger than those on the cold treatment in L. guichenoti but not L. delicata. Using repeated behavioral measures for individual lizards, we found an interaction between incubation temperature and maternal investment in performance, with running speed being affected in a species-specific way by the treatment. We predicted that changes in performance should influence antipredator responses. In support of this prediction, we found that maternal investment impacted antipredator behavior, with animals from the yolk-reduced and cold treatment resuming activity faster after a simulated predatory attack in L. delicata. However, the prediction was not supported in L. guichenoti. Our results highlight the importance of exploring the multifaceted role that environments play across generations to understand how different anthropogenic factors will impact wildlife in the future.

Inferring personal intake recommendations of phosphorous and potassium for end-stage renal failure patients by simulating with Bayesian hierarchical multivariate model.

Most end-stage renal disease (ESRD) patients face a risk of malnutrition, partly due to dietary restrictions on phosphorous and, in some cases, potassium intake. These restrictions aim to regulate plasma phosphate and potassium concentrations and prevent the adverse effects of hyperphosphatemia or hyperkalemia. However, individual responses to nutrition are known to vary, highlighting the need for personalized recommendations rather than relying solely on general guidelines. In this study, our objective was to develop a Bayesian hierarchical multivariate model that estimates the individual effects of nutrients on plasma concentrations and to present a recommendation algorithm that utilizes this model to infer personalized dietary intakes capable of achieving normal ranges for all considered concentrations. Considering the limited research on the reactions of ESRD patients, we collected dietary intake data and corresponding laboratory analyses from a cohort of 37 patients. The collected data were used to estimate the common hierarchical model, from which personalized models of the patients' diets and individual reactions were extracted. The application of our recommendation algorithm revealed substantial variations in phosphorus and potassium intakes recommended for each patient. These personalized recommendations deviate from the general guidelines, suggesting that a notably richer diet may be proposed for certain patients to mitigate the risk of malnutrition. Furthermore, all the participants underwent either hospital, home, or peritoneal dialysis treatments. We explored the impact of treatment type on nutritional reactions by incorporating it as a nested level in the hierarchical model. Remarkably, this incorporation improved the fit of the nutritional effect model by a notable reduction in the normalized root mean square error (NRMSE) from 0.078 to 0.003. These findings highlight the potential for personalized dietary modifications to optimize nutritional status, enhance patient outcomes, and mitigate the risk of malnutrition in the ESRD population.

Consistent individual differences in cattle grazing patterns

Cattle grazing patterns can have a significant impact on the economic and environmental sustainability of rangelands and cattle production. Grazing patterns appear to be driven by consistent individual differences in cattle, deemed ‘grazing personalities’. Optimal distribution of cattle on rangelands can be manipulated by alteration of the environment, like adding water, supplement, or fencing to targeted sites, but few studies address how these management tools are impacting grazing patterns of individual cattle. Our study sought to fill this gap in knowledge by investigating consistency of grazing patterns among fifty cattle, fitted with GPS collars, across two years given the addition of a water site at higher elevation prior to year two. We used Bayesian multivariate mixed models to evaluate consistency, evidenced by among-individual correlations across years with confidence intervals that do not cross 0. We found that cattle were consistent in all rangeland use metrics across years after the addition of the new water site: elevation (Estimate = 0.38, CI = [0.07,0.64]), slope (Estimate = 0.59, CI = [0.21,0.86]), distance to water (Estimate = 0.38 CI = [0.03,0.67]), distance to supplement (Estimate = 0.42, CI = [0.04,0.72]), distance to loafing sites (Estimate = 0.68, CI = [0.24,0.95]), except for distance traveled (Estimate = 0.16,CI = [−0.18, 0.48]). Cows that ventured higher on rangeland in year one (Estimate = 0.84 CI = [0.59,0.98]) and year two (Estimate = 0.93 CI = [0.79,0.99]) and further from watering sites in year two (Estimate = 0.72 CI = [0.41,0.92]) also had more within-individual variability in their grazing patterns than those that stayed at lower elevation and closer to water. These results indicate differences in plasticity or variability can be linked to grazing personalities. There was strong evidence that temperature affected grazing distribution such that on hotter days, cows clumped closer to water and loafing sites, were found at lower elevation, and traveled less far, not reaching supplement sites that were placed at the higher elevations. Based on results of this study, ranchers should consider the effectiveness of management tools on individual cattle and conduct cost-benefit analyses of tools implemented for specific behavior types of cattle. Cattle with behavior types demonstrating more within-cow variability and more ‘hill-climbing’ will impact how cattle graze especially patchy landscapes where resources are distributed across elevation, thus ranchers could potentially match cattle types to rangeland characteristics.

Spatial source apportionment of airborne coarse particulate matter using PMF-Bayesian receptor model

Ambient particulate matter (PM2.5 and PM10), has been extensively monitored in numerous urban areas across the globe. Over the past decade, there has been a significant improvement in PM2.5 air quality, while improvements in PM10 levels have been comparatively modest, primarily due to the limited reduction in coarse particle (PM2.5–10) pollution. Unlike PM2.5, PM2.5–10 predominantly originates from local emissions and is often characterized by pronounced spatial heterogeneity. In this study, we utilized over one million data points on PM concentrations, collected from >100 monitoring sites within a Chinese megacity, to perform spatial source apportionment of PM2.5–10. Despite the widespread availability of such data, it has rarely been employed for this purpose. We employed an enhanced positive matrix factorization approach, capable of handling large datasets, in conjunction with a Bayesian multivariate receptor model to deduce spatial source impacts. Four primary sources were successfully identified and interpreted, including residential burning, industrial processes, road dust, and meteorology-related sources. This interpretation was supported by a considerable body of prior knowledge concerning emission sources, which is usually unavailable in most cases. The methodology proposed in this study demonstrates significant potential for generalization to other regions, thereby contributing to the development of air quality management strategies.

Associations between Predictors of PTSD and Psychosocial Functioning in Veterans: Results from a Longitudinal Assessment Study

Impairments in psychosocial functioning are common in veterans, especially in those with significant mental health symptoms. Although available treatments are aimed at alleviating these symptoms, impairments in psychosocial functioning do not appear to be fully addressed. To achieve rehabilitation and full societal participation, there is a need to identify longitudinal associations of both symptoms and functional outcomes which can be targeted in treatment. United States veterans (N=491) of the Iraq and Afghanistan wars were recruited as part of a longitudinal assessment study which examined predictors of postdeployment adjustment. Veterans were assessed at four timepoints over the course of a two-year period. A Bayesian multivariate multilevel model was used to estimate the association of predictors of PTSD (depression, alcohol use, suicidal ideation, and sleep) on psychosocial functioning as encompassed by quality of life (Quality of Life Scale (QLS)) and disability (World Health Organization Disability Assessment Schedule (WHODAS)) scores over time. As female veterans have unique environmental exposures and functional demands, interactions between predictors and gender were included in all models. There was significant overlap between predictors of PTSD and predictors of disability across domains and quality of life. Depressive symptoms and social support emerged as the strongest predictors of psychosocial functioning. Additionally, suicidality and alcohol use emerged as predictors of quality of life, but not disability. As expected, increases in PTSD symptoms predicted increased disability and decreased quality of life. The effect of depressive symptoms on quality of life was more pronounced for male veterans, and the effect of PTSD and alcohol use on quality of life was more pronounced for female veterans. Findings highlight various treatment targets which have the potential to improve symptoms of PTSD and functional outcomes. Findings highlight an opportunity to leverage intervention and prevention efforts focused on decreasing depression and increasing social support to improve trauma symptoms and maximize rehabilitation and functional recovery in veterans.

Greek GDP forecasting using Bayesian multivariate models

Greek GDP forecasting using Bayesian multivariate models

Multivariate Bayesian Dynamic Modeling for Causal Prediction

Multivariate Bayesian Dynamic Modeling for Causal Prediction

Every tenth malignant solid tumor attributed to overweight and alcohol consumption: A population-based cohort study

BackgroundRecent studies have shown that some four in ten cancers are attributable to a few key risk factors. The aim of this study was to estimate cohort-based population attributable fractions (PAFs) in Finland for potentially modifiable cancer risk factors. MethodsData from eight health studies including 253,953 subjects with 29,802 incident malignant solid tumors were analysed using Bayesian multivariate regression model with multiplicative risk factor effects. We estimated the effects of smoking, excess body weight, alcohol consumption, physical activity, parity and education on cancer incidence and related PAFs by cancer site, accounting for competing mortality. ResultsPAF for all cancer sites and exposures combined was 34% (95% credible interval 29%−39%) in men and 24% (19%−28%) in women. In men, 23% (21%−27%) and in women 8% (6%−9%) of all cancers were attributed to smoking. PAF related to excess body weight was 4% (2%−6%) in men and 5% (2%−7%) in women, to alcohol 7% (3%−10%) in men and 4% (0%−7%) in women, and to excess body weight and alcohol combined 10% (6%−15%) in men and 9% (4%−13%) in women. ConclusionSmoking was the most important factor contributing to cancer burden in Finnish men and women over the last 40 years. The contribution of excess body weight and alcohol consumption together outweighed the role of smoking in women. As the prevalence of overweight is expected to increase, more efficient public health measures supporting adherence to healthy weight are essential to reduce cancer burden.

Cross-regional green certificate transaction strategies based on a double-layer game model

With the implementation of the renewable energy quota system and the development of the green certificate market, the diversification of cross-region trading subjects has brought about the problem of complicating the competition and cooperative relationship between multiple subjects. This problem makes it difficult to balance the conflict of interests of multiple participating subjects in green certificate transactions, and reduces the enthusiasm for trading. To this end, this paper proposes a multi-participant decision model of cross-regional green certificate trading behavior that takes into account the uncertainty of renewable energy generation. In the model, in order to explore the potential full set of strategies of the participating subjects, the cooperative game theory is utilized to analyze the complex interaction between multiple participating subjects in the cross-regional transaction, and the cooperative game model based on green certificates is established. As the participating subjects' information about other participants and their knowledge of the game environment are limited, the decision-making behavior is uncertain. In order to more reasonably portray the subject's game behavior in the actual decision-making scenario and to solve the defect that the traditional evolutionary game model can not reflect the influence of uncertain factors, we use the stochastic evolutionary game theory to dynamically deduce the cooperative strategies of the participating subjects under the limited rationality, and at the same time, introduce the strategy influencing factors to explore the dependence relationship between strategies. In addition, in order to cope with the impact of power generation uncertainty on cross-regional trading and to solve the defects of distributionally robust fuzzy set construction which is subjective and only for the analysis of uncertainty in a single time section, it is proposed to use Bayesian multivariate nonlinear model and mixed-effects model to construct probability distribution fuzzy sets of wind and light power generation uncertainty on the basis of distributionally robust theory. Finally, based on the arithmetic example, the cooperative strategy of multi-participants in cross-regional transactions under finite rationality is deduced, and the impact of uncertainty factors on the cooperative strategy is analyzed.

Interactions among acute respiratory viruses in Beijing, Chongqing, Guangzhou, and Shanghai, China, 2009-2019.

A viral infection can modify the risk to subsequent viral infections via cross-protective immunity, increased immunopathology, or disease-driven behavioral change. There is limited understanding of virus-virus interactions due to lack of long-term population-level data. Our study leverages passive surveillance data of 10 human acute respiratory viruses from Beijing, Chongqing, Guangzhou, and Shanghai collected during 2009 to 2019: influenza A and B viruses; respiratory syncytial virus A and B; human parainfluenza virus (HPIV), adenovirus, metapneumovirus (HMPV), coronavirus, bocavirus (HBoV), and rhinovirus (HRV). We used a multivariate Bayesian hierarchical model to evaluate correlations in monthly prevalence of test-positive samples between virus pairs, adjusting for potential confounders. Of 101,643 lab-tested patients, 33,650 tested positive for any acute respiratory virus, and 4,113 were co-infected with multiple viruses. After adjusting for intrinsic seasonality, long-term trends and multiple comparisons, Bayesian multivariate modeling found positive correlations for HPIV/HRV in all cities and for HBoV/HRV and HBoV/HMPV in three cities. Models restricted to children further revealed statistically significant associations for another ten pairs in three of the four cities. In contrast, no consistent correlation across cities was found among adults. Most virus-virus interactions exhibited substantial spatial heterogeneity. There was strong evidence for interactions among common respiratory viruses in highly populated urban settings. Consistent positive interactions across multiple cities were observed in viruses known to typically infect children. Future intervention programs such as development of combination vaccines may consider spatially consistent virus-virus interactions for more effective control.