Bayesian Spatio-Temporal Modeling of the Dynamics of COVID-19 Deaths in Peru.

Amid the COVID-19 pandemic, understanding the spatial and temporal dynamics of the disease is crucial for effective public health interventions. This study aims to analyze COVID-19 data in Peru using a Bayesian spatio-temporal generalized linear model to elucidate mortality patterns and assess the impact of vaccination efforts. Leveraging data from 194 provinces over 651 days, our analysis reveals heterogeneous spatial and temporal patterns in COVID-19 mortality rates. Higher vaccination coverage is associated with reduced mortality rates, emphasizing the importance of vaccination in mitigating the pandemic's impact. The findings underscore the value of spatio-temporal data analysis in understanding disease dynamics and guiding targeted public health interventions.

Improving the spatial and temporal resolution of burden of disease measures with Bayesian models.

This paper contributes to the field by addressing the critical issue of enhancing the spatial and temporal resolution of health data. Although Bayesian methods are frequently employed to address this challenge in various disciplines, the application of Bayesian spatio-temporal models to burden of disease (BOD) studies remains limited. Our novelty lies in the exploration of two existing Bayesian models that we show to be applicable to a wide range of BOD data, including mortality and prevalence, thereby providing evidence to support the adoption of Bayesian modeling in full BOD studies in the future. We illustrate the benefits of Bayesian modeling with an Australian case study involving asthma and coronary heart disease. Our results showcase the effectiveness of Bayesian approaches in increasing the number of small areas for which results are available and improving the reliability and stability of the results compared to using data directly from surveys or administrative sources.

Bayesian hierarchical spatiotemporal models for prediction of (under)reporting rates and cases: COVID-19 infection among the older people in the United States during the 2020-2022 pandemic.

The gap between the reported and actual COVID-19 infection cases has been an issue of concern. Here, we present Bayesian hierarchical spatiotemporal disease mapping models for state-level predictions of COVID-19 infection risks and (under)reporting rates among people aged 65 and above during the first two years of the pandemic in the United States. With prior elicitation based on recent prevalence studies, the study suggests that the median state-level reporting rate of COVID-19 infection was 90% (interquartile range: [78%, 96%]). Our study uncovers spatiotemporal variations and dynamics in state-level infection risks and (under)reporting rates, suggesting time-varying associations between higher population density, higher percentage of minorities, and higher percentage of vaccination and increased risks of COVID-19 infection, as well as an association between more easily accessible tests and higher reporting rates. With sensitivity analyses, we highlight the impact and importance of incorporating covariates information and objective prior references for evaluating the issue of underreporting.

A multi-resolution ensemble model of three decision-tree-based algorithms to predict daily NO2 concentration in France 2005-2022.

Understanding and managing the health effects of Nitrogen Dioxide (NO2) requires high resolution spatiotemporal exposure maps. Here, we developed a multi-stage multi-resolution ensemble model that predicts daily NO2 concentration across continental France from 2005 to 2022. Innovations of this work include the computation of daily predictions at a 200 m resolution in large urban areas and the use of a spatio-temporal blocking procedure to avoid data leakage and ensure fair performance estimation. Predictions were obtained after three cascading stages of modeling: (1) predicting NO2 total column density from Ozone Monitoring Instrument satellite; (2) predicting daily NO2 concentrations at a 1 km spatial resolution using a large set of potential predictors such as predictions obtained from stage 1, land-cover and road traffic data; and (3) predicting residuals from stage 2 models at a 200 m resolution in large urban areas. The latter two stages used a generalized additive model to ensemble predictions of three decision-tree algorithms (random forest, extreme gradient boosting and categorical boosting). Cross-validated performances of our ensemble models were overall very good, with a ten-fold cross-validated R2 for the 1 km model of 0.83, and of 0.69 for the 200 m model. All three basis learners participated in the ensemble predictions to various degrees depending on time and space. In sum, our multi-stage approach was able to predict daily NO2 concentrations with a relatively low error. Ensembling the predictions maximizes the chance of obtaining accurate values if one basis learner fails in a specific area or at a particular time, by relying on the other learners. To the best of our knowledge, this is the first study aiming to predict NO2 concentrations in France with such a high spatiotemporal resolution, large spatial extent, and long temporal coverage. Exposure estimates are available to investigate NO2 health effects in epidemiological studies.

Spatio-temporal characterization of phenotypic resistance in malaria vector species.

BACKGROUND: Malaria, a deadly disease caused by Plasmodium protozoa parasite and transmitted through bites of infected female Anopheles mosquitoes, remains a significant public health challenge in sub-Saharan Africa. Efforts to eliminate malaria have increasingly focused on vector control using insecticides. However, the emergence of insecticide resistance (IR) in malaria vectors pose a formidable obstacle, and the current IR mapping models remain static, relying on fixed coefficients. This study introduces a dynamic spatio-temporal approach to characterize phenotypic resistance in Anopheles gambiae complex and Anopheles arabiensis. We developed a cellular automata (CA) model and applied it to data collected from Ethiopia, Nigeria, Cameroon, Chad, and Burkina Faso. The data encompasses georeferenced records detailing IR levels in mosquito vector populations across various classes of insecticides. In characterizing the dynamic patterns of confirmed resistance, we identified key driving factors through correlation analysis, chi-square tests, and extensive literature review. RESULTS: The CA model demonstrated robustness in capturing the spatio-temporal dynamics of confirmed IR states in the vector populations. In our model, the key driving factors included insecticide usage, agricultural activities, human population density, Land Use and Land Cover (LULC) characteristics, and environmental variables. CONCLUSIONS: The CA model developed offers a robust tool for countries that have limited data on confirmed IR in malaria vectors. The embrace of a dynamical modeling approach and accounting for evolving conditions and influences, contribute to deeper understanding of IR dynamics, and can inform effective strategies for malaria vector control, and prevention in regions facing this critical health challenge.

Time Series Forecasting of Cardiovascular Mortality: Machine Learning Based on State Economic and Local Medical Data.

This study focuses on the complex interplay of healthcare, economic factors, and population dynamics, addressing a research gap in regional-level models that integrate diverse features within a temporal framework. Our primary objective is to develop an advanced temporal model for predicting cardiovascular mortality in Russian regions by integrating global and local healthcare features with economic and population dynamics. Utilizing a dataset from the Almazov Center's Department of Mortality Performance Monitoring, covering 94 regions and 752 records from January 1, 2015, to December 31, 2023, our analysis incorporates key parameters such as angioplasty procedures, population morbidity rates, Ischemic Heart Disease (IHD) and Cardiovascular Diseases (CVD) monitoring, and demographic data. Employing XGBoost and a regression model, our methodology ensures the model's robustness and generalizability.

AA-RGTCN: reciprocal global temporal convolution network with adaptive alignment for video-based person re-identification.

Person re-identification(Re-ID) aims to retrieve pedestrians under different cameras. Compared with image-based Re-ID, video-based Re-ID extracts features from video sequences that contain both spatial features and temporal features. Existing methods usually focus on the most attractive image parts, and this will lead to redundant spatial description and insufficient temporal description. Other methods that take temporal clues into consideration usually ignore misalignment between frames and only focus on a fixed length of one given sequence. In this study, we proposed a Reciprocal Global Temporal Convolution Network with Adaptive Alignment(AA-RGTCN). The structure could address the drawback of misalignment between frames and model discriminative temporal representation. Specifically, the Adaptive Alignment block is designed to shift each frame adaptively to its best position for temporal modeling. Then, we proposed the Reciprocal Global Temporal Convolution Network to model robust temporal features across different time intervals along both normal and inverted time order. The experimental results show that our AA-RGTCN can achieve 85.9% mAP and 91.0% Rank-1 on MARS, 90.6% Rank-1 on iLIDS-VID, and 96.6% Rank-1 on PRID-2011, indicating we could gain better performance than other state-of-the-art approaches.

Spatial-temporal Bayesian accelerated failure time models for survival endpoints with applications to prostate cancer registry data.

Prostate cancer is the most common cancer after non-melanoma skin cancer and the second leading cause of cancer deaths in US men. Its incidence and mortality rates vary substantially across geographical regions and over time, with large disparities by race, geographic regions (i.e., Appalachia), among others. The widely used Cox proportional hazards model is usually not applicable in such scenarios owing to the violation of the proportional hazards assumption. In this paper, we fit Bayesian accelerated failure time models for the analysis of prostate cancer survival and take dependent spatial structures and temporal information into account by incorporating random effects with multivariate conditional autoregressive priors. In particular, we relax the proportional hazards assumption, consider flexible frailty structures in space and time, and also explore strategies for handling the temporal variable. The parameter estimation and inference are based on a Monte Carlo Markov chain technique under a Bayesian framework. The deviance information criterion is used to check goodness of fit and to select the best candidate model. Extensive simulations are performed to examine and compare the performances of models in different contexts. Finally, we illustrate our approach by using the 2004-2014 Pennsylvania Prostate Cancer Registry data to explore spatial-temporal heterogeneity in overall survival and identify significant risk factors.

SF-TMN: SlowFast temporal modeling network for surgical phase recognition.

PURPOSE: Automatic surgical phase recognition is crucial for video-based assessment systems in surgical education. Utilizing temporal information is crucial for surgical phase recognition; hence, various recent approaches extract frame-level features to conduct full video temporal modeling. METHODS: For better temporal modeling, we propose SlowFast temporal modeling network (SF-TMN) for offline surgical phase recognition that can achieve not only frame-level full video temporal modeling but also segment-level full video temporal modeling. We employ a feature extraction network, pretrained on the target dataset, to extract features from video frames as the training data for SF-TMN. The Slow Path in SF-TMN utilizes all frame features for frame temporal modeling. The Fast Path in SF-TMN utilizes segment-level features summarized from frame features for segment temporal modeling. The proposed paradigm is flexible regarding the choice of temporal modeling networks. RESULTS: We explore MS-TCN and ASFormer as temporal modeling networks and experiment with multiple combination strategies for Slow and Fast Paths. We evaluate SF-TMN on Cholec80 and Cataract-101 surgical phase recognition tasks and demonstrate that SF-TMN can achieve state-of-the-art results on all considered metrics. SF-TMN with ASFormer backbone outperforms the state-of-the-art Swin BiGRU by approximately 1% in accuracy and 1.5% in recall on Cholec80. We also evaluate SF-TMN on action segmentation datasets including 50salads, GTEA, and Breakfast, and achieve state-of-the-art results. CONCLUSION: The improvement in the results shows that combining temporal information from both frame level and segment level by refining outputs with temporal refinement stages is beneficial for the temporal modeling of surgical phases.

Predicting COVID-19 hospitalizations: The importance of healthcare hotlines, test positivity rates and vaccination coverage.

In this study, we developed a negative binomial regression model for one-week ahead spatio-temporal predictions of the number of COVID-19 hospitalizations in Uppsala County, Sweden. Our model utilized weekly aggregated data on testing, vaccination, and calls to the national healthcare hotline. Variable importance analysis revealed that calls to the national healthcare hotline were the most important contributor to prediction performance when predicting COVID-19 hospitalizations. Our results support the importance of early testing, systematic registration of test results, and the value of healthcare hotline data in predicting hospitalizations. The proposed models may be applied to studies modeling hospitalizations of other viral respiratory infections in space and time assuming count data are overdispersed. Our suggested variable importance analysis enables the calculation of the effects on the predictive performance of each covariate. This can inform decisions about which types of data should be prioritized, thereby facilitating the allocation of healthcare resources.

Modeling the temporal evolution of plasma p-tau in relation to amyloid beta and tau PET.

INTRODUCTION: The timing of plasma biomarker changes is not well understood. The goal of this study was to evaluate the temporal co-evolution of plasma and positron emission tomography (PET) Alzheimer's disease (AD) biomarkers. METHODS: We included 1408 Mayo Clinic Study of Aging and Alzheimer's Disease Research Center participants. An accelerated failure time (AFT) model was fit with amyloid beta (Aß) PET, tau PET, plasma p-tau217, p-tau181, and glial fibrillary acidic protein (GFAP) as endpoints. RESULTS: Individual timing of plasma p-tau progression was strongly associated with Aß PET and GFAP progression. In the population, GFAP became abnormal first, then Aß PET, plasma p-tau, and tau PET temporal meta-regions of interest when applying cut points based on young, cognitively unimpaired participants. DISCUSSION: Plasma p-tau is a stronger indicator of a temporally linked response to elevated brain Aß than of tau pathology. While Aß deposition and a rise in GFAP are upstream events associated with tau phosphorylation, the temporal link between p-tau and Aß PET was the strongest. HIGHLIGHTS: Plasma p-tau progression was more strongly associated with Aß than tau PET. Progression on plasma p-tau was associated with Aß PET and GFAP progression. P-tau181 and p-tau217 become abnormal after Aß PET and before tau PET. GFAP became abnormal first, before plasma p-tau and Aß PET.

Improving mixed-integer temporal modeling by generating synthetic data using conditional generative adversarial networks: A case study of fluid overload prediction in the intensive care unit.

OBJECTIVE: The challenge of mixed-integer temporal data, which is particularly prominent for medication use in the critically ill, limits the performance of predictive models. The purpose of this evaluation was to pilot test integrating synthetic data within an existing dataset of complex medication data to improve machine learning model prediction of fluid overload. MATERIALS AND METHODS: This retrospective cohort study evaluated patients admitted to an ICU ≥ 72 h. Four machine learning algorithms to predict fluid overload after 48-72 h of ICU admission were developed using the original dataset. Then, two distinct synthetic data generation methodologies (synthetic minority over-sampling technique (SMOTE) and conditional tabular generative adversarial network (CTGAN)) were used to create synthetic data. Finally, a stacking ensemble technique designed to train a meta-learner was established. Models underwent training in three scenarios of varying qualities and quantities of datasets. RESULTS: Training machine learning algorithms on the combined synthetic and original dataset overall increased the performance of the predictive models compared to training on the original dataset. The highest performing model was the meta-model trained on the combined dataset with 0.83 AUROC while it managed to significantly enhance the sensitivity across different training scenarios. DISCUSSION: The integration of synthetically generated data is the first time such methods have been applied to ICU medication data and offers a promising solution to enhance the performance of machine learning models for fluid overload, which may be translated to other ICU outcomes. A meta-learner was able to make a trade-off between different performance metrics and improve the ability to identify the minority class.

Spatio-temporal distribution and contributing factors of tegumentary and visceral leishmaniasis: A comparative study in Bahia, Brazil.

Tegumentary (TL) and visceral (VL) leishmaniasis are neglected zoonotic diseases in Brazil, caused by different parasites and transmitted by various vector species. This study investigated and compared spatio-temporal patterns of TL and VL from 2007 to 2020 in the state of Bahia, Brazil, and their correlations with extrinsic factors. The results showed that the total number of cases of both TL and VL were decreasing. The number of municipalities with reported cases reduced for TL over time but remained almost unchanged for VL. There were few municipalities with reported both diseases. Statistical analysis showed that local TL incidence was associated positively with natural forest. Local VL incidence was associated positively with Cerrado (Brazilian savannah) vegetation. This study identified different patterns of occurrence of VL and TL and the risk areas that could be prioritized for epidemiological surveillance.

Action Recognition Based on Multi-Level Topological Channel Attention of Human Skeleton.

In action recognition, obtaining skeleton data from human poses is valuable. This process can help eliminate negative effects of environmental noise, including changes in background and lighting conditions. Although GCN can learn unique action features, it fails to fully utilize the prior knowledge of human body structure and the coordination relations between limbs. To address these issues, this paper proposes a Multi-level Topological Channel Attention Network algorithm: Firstly, the Multi-level Topology and Channel Attention Module incorporates prior knowledge of human body structure using a coarse-to-fine approach, effectively extracting action features. Secondly, the Coordination Module utilizes contralateral and ipsilateral coordinated movements in human kinematics. Lastly, the Multi-scale Global Spatio-temporal Attention Module captures spatiotemporal features of different granularities and incorporates a causal convolution block and masked temporal attention to prevent non-causal relationships. This method achieved accuracy rates of 91.9% (Xsub), 96.3% (Xview), 88.5% (Xsub), and 90.3% (Xset) on NTU-RGB+D 60 and NTU-RGB+D 120, respectively.

Spatio-temporal modeling of human leptospirosis prevalence using the maximum entropy model.

BACKGROUND: Leptospirosis, a zoonotic disease, stands as one of the prevailing health issues in some tropical areas of Iran. Over a decade, its incidence rate has been estimated at approximately 2.33 cases per 10,000 individuals. Our research focused on analyzing the spatiotemporal clustering of Leptospirosis and developing a disease prevalence model as an essential focal point for public health policymakers, urging targeted interventions and strategies. METHODS: The SaTScan and Maximum Entropy (MaxEnt) modeling methods were used to find the spatiotemporal clusters of the Leptospirosis and model the disease prevalence in Iran. We incorporated nine environmental covariates by employing a spatial resolution of 1 km x 1 km, the finest resolution ever implemented for modeling Human Leptospirosis in Iran. These covariates encompassed the Digital Elevation Model (DEM), slope, displacement areas, water bodies, and land cover, monthly recorded Normalized Difference Vegetation Index (NDVI), monthly recorded precipitation, monthly recorded mean and maximum temperature, contributing significantly to our disease modeling approach. The analysis using MaxEnt yielded the Area Under the Receiver Operating Characteristic Curve (AUC) metrics for the training and test data, to evaluate the accuracy of the implemented model. RESULTS: The findings reveal a highly significant primary cluster (p-value < 0.05) located in the western regions of the Gilan province, spanning from July 2013 to July 2015 (p-value < 0.05). Moreover, there were four more clusters (p-value < 0.05) identified near Someh Sara, Neka, Gorgan and Rudbar. Furthermore, the risk mapping effectively illustrates the potential expansion of the disease into the western and northwestern regions. The AUC metrics of 0.956 and 0.952 for the training and test data, respectively, underscoring the robust accuracy of the implemented model. Interestingly, among the variables considered, the influence of slope and distance from water bodies appears to be minimal. However, altitude and precipitation stand out as the primary determinants that significantly contribute to the prevalence of the disease. CONCLUSIONS: The risk map generated through this study carries significant potential to enhance public awareness and inform the formulation of impactful policies to combat Leptospirosis. These maps also play a crucial role in tracking disease incidents and strategically directing interventions toward the regions most susceptible.

Malaria Risk Drivers in the Brazilian Amazon: Land Use-Land Cover Interactions and Biological Diversity.

Malaria is a prevalent disease in several tropical and subtropical regions, including Brazil, where it remains a significant public health concern. Even though there have been substantial efforts to decrease the number of cases, the reoccurrence of epidemics in regions that have been free of cases for many years presents a significant challenge. Due to the multifaceted factors that influence the spread of malaria, influencing malaria risk factors were analyzed through regional outbreak cluster analysis and spatio-temporal models in the Brazilian Amazon, incorporating climate, land use/cover interactions, species richness, and number of endemic birds and amphibians. Results showed that high amphibian and bird richness and endemism correlated with a reduction in malaria risk. The presence of forest had a risk-increasing effect, but it depended on its juxtaposition with anthropic land uses. Biodiversity and landscape composition, rather than forest formation presence alone, modulated malaria risk in the period. Areas with low endemic species diversity and high human activity, predominantly anthropogenic landscapes, posed high malaria risk. This study underscores the importance of considering the broader ecological context in malaria control efforts.

Spatio-temporal interpolation and delineation of extreme heat events in California between 2017 and 2021.

Robust spatio-temporal delineation of extreme climate events and accurate identification of areas that are impacted by an event is a prerequisite for identifying population-level and health-related risks. In prior research, attributes such as temperature and humidity have often been linearly assigned to the population of the study unit from the closest weather station. This could result in inaccurate event delineation and biased assessment of extreme heat exposure. We have developed a spatio-temporal model to dynamically delineate boundaries for Extreme Heat Events (EHE) across space and over time, using a relative measure of Apparent Temperature (AT). Our surface interpolation approach offers a higher spatio-temporal resolution compared to the standard nearest-station (NS) assignment method. We show that the proposed approach can provide at least 80.8 percent improvement in identification of areas and populations impacted by EHEs. This improvement in average adjusts the misclassification of about one million Californians per day of an extreme event, who would be either unidentified or misidentified under EHEs between 2017 and 2021.

A Bayesian multistage spatio-temporally dependent model for spatial clustering and variable selection.

In spatio-temporal epidemiological analysis, it is of critical importance to identify the significant covariates and estimate the associated time-varying effects on the health outcome. Due to the heterogeneity of spatio-temporal data, the subsets of important covariates may vary across space and the temporal trends of covariate effects could be locally different. However, many spatial models neglected the potential local variation patterns, leading to inappropriate inference. Thus, this article proposes a flexible Bayesian hierarchical model to simultaneously identify spatial clusters of regression coefficients with common temporal trends, select significant covariates for each spatial group by introducing binary entry parameters and estimate spatio-temporally varying disease risks. A multistage strategy is employed to reduce the confounding bias caused by spatially structured random components. A simulation study demonstrates the outperformance of the proposed method, compared with several alternatives based on different assessment criteria. The methodology is motivated by two important case studies. The first concerns the low birth weight incidence data in 159 counties of Georgia, USA, for the years 2007 to 2018 and investigates the time-varying effects of potential contributing covariates in different cluster regions. The second concerns the circulatory disease risks across 323 local authorities in England over 10 years and explores the underlying spatial clusters and associated important risk factors.

A multivariate spatio-temporal model for the incidence of imported COVID-19 cases and COVID-19 deaths in Cuba.

To monitor the COVID-19 epidemic in Cuba, data on several epidemiological indicators have been collected on a daily basis for each municipality. Studying the spatio-temporal dynamics in these indicators, and how they behave similarly, can help us better understand how COVID-19 spread across Cuba. Therefore, spatio-temporal models can be used to analyze these indicators. Univariate spatio-temporal models have been thoroughly studied, but when interest lies in studying the association between multiple outcomes, a joint model that allows for association between the spatial and temporal patterns is necessary. The purpose of our study was to develop a multivariate spatio-temporal model to study the association between the weekly number of COVID-19 deaths and the weekly number of imported COVID-19 cases in Cuba during 2021. To allow for correlation between the spatial patterns, a multivariate conditional autoregressive prior (MCAR) was used. Correlation between the temporal patterns was taken into account by using two approaches; either a multivariate random walk prior was used or a multivariate conditional autoregressive prior (MCAR) was used. All models were fitted within a Bayesian framework.