Spatial Dependence Research Articles

A spatial dependency based reinforcement learning model for selecting features in spatial classification

A spatial dependency based reinforcement learning model for selecting features in spatial classification

Neighbourhood factors and tuberculosis incidence in Cape Town: A negative binomial regression and spatial analysis.

Although the link between poverty and tuberculosis (TB) is widely recognised, limited studies have investigated the association between neighbourhood factors and TB incidence. Since the factors influencing different episodes of TB might be different, this study focused on the first episode of TB disease (first-episode TB). All first episodes in previously linked and geocoded TB notification data from 2007 to 2015 in Cape Town, South Africa, were aggregated at the neighbourhood level and merged with the 2011 census data. We conducted an ecological study to assess the association between neighbourhood incidence of first-episode TB and neighbourhood factors (total TB burden [all episodes] in the previous year, socioeconomic index, mean household size, mean age, and percentage males) using a negative binomial regression. We also examined the presence of hotspots in neighbourhood TB incidence with the Global Moran's I statistic and assessed spatial dependency in the association between neighbourhood factors and TB incidence using a spatial lag model. The study included 684 neighbourhoods with a median first-episode TB incidence rate of 114 (IQR: 0-345) per 100,000 people. We found lower neighbourhood socioeconomic index (SEI), higher neighbourhood total TB burden, lower neighbourhood mean household size, and lower neighbourhood mean age were associated with increased neighbourhood first-episode TB incidence. Our findings revealed a hotspot of first-episode TB incidence in Cape Town and evidence of spatial dependency in the association between neighbourhood factors and TB incidence. Neighbourhood TB burden and SEI were associated with first-episode TB incidence, and there was spatial dependency in this association. Our findings can inform targeted interventions to reduce TB in high-risk neighbourhoods, thereby reducing health disparities and promoting health equity.

Differential gene expression analysis of spatial transcriptomic experiments using spatial mixed models

Spatial transcriptomics (ST) assays represent a revolution in how the architecture of tissues is studied by allowing for the exploration of cells in their spatial context. A common element in the analysis is delineating tissue domains or “niches” followed by detecting differentially expressed genes to infer the biological identity of the tissue domains or cell types. However, many studies approach differential expression analysis by using statistical approaches often applied in the analysis of non-spatial scRNA data (e.g., two-sample t-tests, Wilcoxon’s rank sum test), hence neglecting the spatial dependency observed in ST data. In this study, we show that applying linear mixed models with spatial correlation structures using spatial random effects effectively accounts for the spatial autocorrelation and reduces inflation of type-I error rate observed in non-spatial based differential expression testing. We also show that spatial linear models with an exponential correlation structure provide a better fit to the ST data as compared to non-spatial models, particularly for spatially resolved technologies that quantify expression at finer scales (i.e., single-cell resolution).

Deep knowledge distillation: A self-mutual learning framework for traffic prediction

Traffic flow prediction in spatio-temporal networks is a crucial aspect of Intelligent Transportation Systems (ITS). Existing traffic flow forecasting methods, particularly those utilizing graph neural networks, encounter limitations. When processing large-scale graph data, the depth of these models can restrict their ability to effectively capture complex relationships and patterns. Additionally, these methods often focus mainly on local neighborhood information, which can limit their capability to recognize and analyze global relationships and patterns within the graph data. Therefore, we proposed a deep knowledge distillation model, tailored to effectively capture spatio-temporal patterns in traffic flow prediction. This model incorporates a bidirectional random walk process on a directed graph, enabling it to effectively capture both spatial and temporal dependencies. Utilizing a blend of mutual learning and self-distillation, our approach enhances the detection of spatio-temporal relationships within traffic data and improves the feature perception ability at both local and global levels. We tested our model on two real-world datasets, achieving notable improvements in prediction accuracy, especially for predictions within a one-hour timeframe. In comparison to the baseline model, our proposed model achieved accuracy improvements of 0.19 and 0.18 on the respective datasets. These results highlight the success of using mutual learning and self-distillation to transfer knowledge effectively within and between models and to improve the model’s capability in identifying and extracting features.

Fintech and financial stability: Evidence from spatial analysis for 25 countries

Fintech has experienced rapid advances in recent years. This study examines the impact of Fintech on financial stability for a group of 25 countries during 2013–2020. We adopt the novel Fintech-enabled financing volume to directly measure Fintech development. We utilise both the aggregate and disaggregated level of Fintech financing; the latter includes crowdfunding, business lending and consumer lending, each has a different funding process and default rates. We account for spatial dependence in financial stability across countries by employing various spatial models. Our findings first reveal that there is positive spatial dependence of financial stability across countries. It implies that financial stability has a positive spillover to neighbouring countries and validates the necessity of spatial analysis. Second, based on the Spatial Durbin Model which best describes our data, Fintech financing makes a positive local and cross-border contribution towards financial stability, irrespective of alternative weight matrices and sample sizes. Such positive impact is more profound in countries with smaller sizes of Fintech financing volume, and the cross-border effect is stronger with closer geographic proximity. Finally, crowdfunding enhances financial stability, whilst consumer lending has a contrasting destabilising effect.

Task-Dependent Brain Activity in Generalized Anxiety Disorder Determined By Bayesian Spatio-Temporal Single Case Model

Introduction: Data obtained from functional magnetic resonance imaging (fMRI) have a complex structure. Considering the special features of this type of data in analyses is of particular importance. Previous studies on generalized anxiety disorder (GAD) as a prevalent mental disorder using functional neuroimaging have had conflicting results. In this study, we apply a Bayesian spatiotemporal model to this type of data which considers both spatial and temporal dependence among regions which is one of the most essential features to consider. Methods: In this single-subject study, we analyze data from a patient with GAD and a healthy participant. Both participants are 24-year-old women who are assigned an emotion reactivity task (matching neutral and negative facial expressions) inside a scanner. The spatial Bayesian variable selection method is used to detect blood oxygen level-dependent activation in fMRI data. Results: Activation areas in neutral and negative facial expressions are provided for both participants by posterior probability map. The results of our study show a greater level of activity in the GAD participant in comparison to the healthy participant in responding to the negative matching task. Conclusion: the GAD patient showed more neural activity in response to negative facial expressions than the healthy participant in brain regions related to emotional response in the areas of the frontal Pole, middle frontal gyrus, insular cortex, and frontal orbital cortex. Moreover, the inferior frontal gyrus in the patient with GAD showed more reaction to negative emotional stimuli.

The impact of agricultural credit on the cattle inventory and deforestation in Colombia: a spatial analysis

The relationship between agricultural credit and agricultural production as well as the impacts on sustainable development (i.e., poverty alleviation, reduction of inequalities, food and nutrition security, and stimulation of economic growth) have been widely documented. The objective of this study is to analyze the impact of credit on cattle production and deforestation in Colombia through spatial panel data models. For this purpose, a departmental data panel for the period 2011–2020 was built, based on available information from public entities. The results suggest that in Colombia, the relationship between access to credit and cattle production is significant and can be either negative or positive. In addition, there is evidence of spatial dependence, meaning that cattle production in one department is being affected by cattle production in a neighboring department or by all the departments that make up the national territory. Regarding deforestation, results show that, although the number of cattle present in a department does affect its annual deforestation rate due to a poor coverage of sustainably intensified cattle ranching systems, there is no relationship between deforestation and the access to credit nor any spatial correlations.

Time and frequency-domain feature fusion network for multivariate time series classification

Multivariate time series classification is a significant research topic in the realm of data mining, which encompasses a wide array of practical applications in domains such as healthcare, energy systems, and traffic. The complex temporal and spatial dependencies inherent in multivariate time series pose challenges to classification tasks. Previous studies have usually focused on the time-domain information of multivariate time series. However, achieving accurate classification using only the time-domain information may be difficult. To address this challenge, a time and frequency-domain feature fusion network (TF-Net) for multivariate time series classification is proposed in this paper. Our model contains two modules, the time-domain module and the frequency-domain module. The time-domain module is used to capture the time-domain features of multivariate time series. It is constructed using CNNs and GCNs, enabling the capture of both temporal and spatial dependencies within the time-domain. The frequency-domain module is used to capture the frequency-domain features of multivariate time series data. In this module, we treat the frequency-domain features as images and innovatively transform the multivariate time series classification task into an image classification task. Our method is able to classify multivariate time series from both the time-domain and frequency-domain, which provides a new perspective for multivariate time series analysis. We conduct extensive experiments on the UAE archive, and the experimental results show that our model achieves the best performance compared to the ten state-of-the-art methods.

Short-term multi-step-ahead sector-based traffic flow prediction based on the attention-enhanced graph convolutional LSTM network (AGC-LSTM)

AbstractAccurate sector-based air traffic flow predictions are essential for ensuring the safety and efficiency of the air traffic management (ATM) system. However, due to the inherent spatial and temporal dependencies of air traffic flow, it is still a challenging problem. To solve this problem, some methods are proposed considering the relationship between sectors, while the complicated spatiotemporal dynamics and interdependencies between traffic flow of route segments related to the sector are not taken into account. To address this challenge, the attention-enhanced graph convolutional long short-term memory network (AGC-LSTM) model is applied to improve the short-term sector-based traffic flow prediction, in which spatial structures of route segments related to the sector are considered for the first time. Specifically, the graph convolutional networks (GCN)-LSTM network model was employed to capture spatiotemporal dependencies of the flight data, and the attention mechanism is designed to concentrate on the informative features from key nodes at each layer of the AGC-LSTM model. The proposed model is evaluated through a case study of the typical enroute sector in the central–southern region of China. The prediction results show that MAE reduces by 14.4% compared to the best performing GCN-LSTM model among the other five models. Furthermore, the study involves comparative analyses to assess the influence of route segment range, input and output sequence lengths, and time granularities on prediction performance. This study helps air traffic managers predict flight situations more accurately and avoid implementing overly conservative or excessively aggressive flow management measures for the sectors.

Machine learning prediction of health risk and spatial dependence of geogenic contaminated groundwater from the Hetao Basin, China

Geogenic contaminated groundwater (GCG), characterized by elevated arsenic, fluoride, and iodine levels, present a significant challenge to public health and government management. Conventional survey-based approaches of collecting groundwater samples, conducting physicochemical tests, and performing spatial interpolation to obtain regional groundwater chemical component maps are inefficient and costly. More importantly, it does not take into account the actual hydrogeological conditions or the characteristics of pollutant transport and enrichment. To address this issue, we utilized Support Vector Machine (SVM), Random Forest (RF), Adaptive Boosting (AdaBoost), and Extreme Gradient Boosting (XGBoost) to analyze the likelihood of occurrence of arsenic, fluoride, and iodine as well as their spatial distribution in shallow groundwater from the Hetao Basin. Our study incorporated 20 indicators related to meteorology, soil physicochemical properties, and groundwater conditions, along with 1505 labeled samples consisting of groundwater arsenic, fluoride, and iodine concentrations and their corresponding coordinates. Subsequently, the study automatically analyzed the meteorological, soil physicochemical properties and groundwater conditions by constructing a machine learning model using the available data. In order to optimise and select the best prediction model, this paper presents a quantitative evaluation of the prediction performance of various machine learning models. The accuracy (AC), area under curve (AUC) and mean squared error (MSE) were calculated to predict the spatial distribution of CGC. Subsequently, the optimized model for predicting the spatial distribution of GCG was selected. The results showed that the XGBoost algorithm provided optimal predictions for groundwater with arsenic concentrations above 10 μg/L and fluoride concentrations exceeding 1.5 mg/L, whereas the RF model provided the best predictions for groundwater with arsenic concentrations surpassing 50 μg/L and iodine concentrations exceeding 100 μg/L. Subsequently, groundwater health risk zones were delineated based on an optimal prediction model, and demographic analysis was conducted in both the direct and potential groundwater risk zones. Model predictions indicated that hundreds of thousands of people in the Hetao Basin were facing a public health crisis caused by high concentrations of arsenic, fluoride and iodine in groundwater. These findings underscore the significant health challenge in the study area. Considering the agricultural development and increasing groundwater use in the area, our findings can guide local governments in managing the extent of groundwater development, establishing control zones, and enhancing protection measures for populations at risk from groundwater contamination.

Enhancing Fetal Electrocardiogram Signal Extraction Accuracy through a CycleGAN Utilizing Combined CNN-BiLSTM Architecture.

The fetal electrocardiogram (FECG) records changes in the graph of fetal cardiac action potential during conduction, reflecting the developmental status of the fetus in utero and its physiological cardiac activity. Morphological alterations in the FECG can indicate intrauterine hypoxia, fetal distress, and neonatal asphyxia early on, enhancing maternal and fetal safety through prompt clinical intervention, thereby reducing neonatal morbidity and mortality. To reconstruct FECG signals with clear morphological information, this paper proposes a novel deep learning model, CBLS-CycleGAN. The model's generator combines spatial features extracted by the CNN with temporal features extracted by the BiLSTM network, thus ensuring that the reconstructed signals possess combined features with spatial and temporal dependencies. The model's discriminator utilizes PatchGAN, employing small segments of the signal as discriminative inputs to concentrate the training process on capturing signal details. Evaluating the model using two real FECG signal databases, namely "Abdominal and Direct Fetal ECG Database" and "Fetal Electrocardiograms, Direct and Abdominal with Reference Heartbeat Annotations", resulted in a mean MSE and MAE of 0.019 and 0.006, respectively. It detects the FQRS compound wave with a sensitivity, positive predictive value, and F1 of 99.51%, 99.57%, and 99.54%, respectively. This paper's model effectively preserves the morphological information of FECG signals, capturing not only the FQRS compound wave but also the fetal P-wave, T-wave, P-R interval, and ST segment information, providing clinicians with crucial diagnostic insights and a scientific foundation for developing rational treatment protocols.

The spatial dimensions of cultural consumption: how distance influences consumption levels in a spatial setting

AbstractCultural consumption has increasingly acquired a fundamental role in urban policy frameworks, thanks to its empirically proven positive effects on individuals and on societies. Although several theoretical and empirical contributions have examined the main socio-economic determinants that explain cultural consumption; its spatial dimensions remain significantly under-considered and require further analysis. The aim of this paper is to analyse the factors influencing cultural consumption with a focus on spatial distance and spatial dependence. Specifically, it seeks to inquire into the extent to which spatial distance between consumers and cultural institutions plays a role in neighbourhood’s levels of cultural consumption. To do this, human mobility data towards cultural institutions are used as a proxy for individual cultural consumption levels in six French cities. Results show that spatial proximity with the cultural offer matters in explaining consumption patterns, but that traditional socioeconomic determinants have a higher explanatory value.

Assessing life expectancy disparities in Chicago with a deep dive into green space

BackgroundFew studies have examined the association between green space and life expectancy, and only one study has looked at that data for small areas in the United States. These studies have found a positive association between green space and life expectancy. This study investigates the association between green space and small-area life expectancy in Chicago, a city in the United States with distinct environmental and socioeconomic characteristics and the largest U.S. life expectancy gap between neighborhoods. MethodsWe conducted a spatial analysis using the green space metrics of normalized difference vegetation index (NDVI), park access, and tree canopy. Additionally, we included 18 lifestyle, environmental, and social covariates. Diagnostic analytics included multiple linear regression, Moran I test for spatial dependence, and principal component analysis with CAR Gaussian model. We performed a simulation to predict life expectancy gain if NDVI was brought to median levels. ResultsOur study revealed that after adjusting for covariates and spatial dependence (Moran I test p = 0.012) with a CAR Gaussian model, NDVI (credible interval = 0.292-0.894) demonstrated significance at a 5 % level, while tree canopy (credible interval = -0.404-0.109) and park access (credible interval = -0.258-0.161) did not. Excluding non-significant variables (tree canopy and park access) from the CAR Gaussian model, the median effect of NDVI on life expectancy was 0.491 (95 % credible interval: 0.24–0.718). Projecting this effect to tracts below the median NDVI predicted a gain of 476,000 years of life expectancy for Chicago residents. ConclusionThis study provides evidence that green space and LE are positively correlated, suggesting the potential to increase green space to address current LE gaps. Future research should investigate the mechanisms by which green space correlates to lifespan.

Spatial profiling of in vivo diffusion-weighted MRI parameters in the healthy human kidney.

Diffusion-weighted MRI is a technique that can infer microstructural and microcirculatory features from biological tissue, with particular application to renal tissue. There is extensive literature on diffusion tensor imaging (DTI) of anisotropy in the renal medulla, intravoxel incoherent motion (IVIM) measurements separating microstructural from microcirculation effects, and combinations of the two. However, interpretation of these features and adaptation of more specific models remains an ongoing challenge. One input to this process is a whole organ distillation of corticomedullary contrast of diffusion metrics, as has been explored for other renal biomarkers. In this work, we probe the spatial dependence of diffusion MRI metrics with concentrically layered segmentation in 11 healthy kidneys at 3 T. The metrics include those from DTI, IVIM, a combined approach titled "REnal Flow and Microstructure AnisotroPy (REFMAP)", and a multiply encoded model titled "FC-IVIM" providing estimates of fluid velocity and branching length. Fractional anisotropy decreased from the inner kidney to the outer kidney with the strongest layer correlation in both parenchyma (including cortex and medulla) and medulla with Spearman correlation coefficients and p-values (r, p) of (0.42, <0.001) and (0.37, <0.001), respectively. Also, dynamic parameters derived from the three models significantly decreased with a high correlation from the inner to the outer parenchyma or medulla with (r, p) ranges of (0.46-0.55, <0.001). These spatial trends might find implications for indirect assessments of kidney physiology and microstructure using diffusion MRI.

Sign Speak: Recogninzing Sign Language with Machine Learning

Sign language serves as a critical means of communication for individuals with hearing impairments, enabling them to integrate into society effectively and express themselves. However, interpreting and recognizing sign language gestures present unique challenges due to the dynamic nature of gestures and spatial dependencies inherent in sign language communication. As a response, the SignSpeak project employs advanced machine learning techniques to address these challenges and enhance accessibility for the deaf and hard of hearing community. The project leverages a diverse dataset sourced from Kaggle, comprising images of sign language gestures captured in various contexts. The integration of advanced algorithms, such as 3D Convolutional Neural Networks (CNNs), and Gated Recurrent Units (GRUs), enables SignSpeak to recognize and interpret sign language gestures accurately and in real-time. This integration allows the model to capture both spatial and temporal features inherent in sign language, thus enabling more robust and accurate recognition. The project encompasses several critical stages, including data preprocessing, model development, training, and evaluation. Data preprocessing involves converting the image data into a suitable format and applying augmentation techniques to enhance the diversity and robustness of the dataset. Model development entails designing a deep learning architecture that combines CNNs and GRUs to effectively capture spatial and temporal dependencies in sign language gestures. Training the model involves optimizing parameters and hyperparameters to achieve optimal performance. Evaluation metrics such as accuracy, F1 score, and recall are utilized to assess the model's performance on both training and validation datasets. The trained model is then tested on a separate test dataset to evaluate its real- world performance and generalization ability. Experimental results demonstrate the efficacy of the SignSpeak approach in accurately recognizing and interpreting sign language gestures. The model achieves high accuracy scores, demonstrating its potential to enhance accessibility and inclusion for individuals with hearing impairments. By providing real-time translation of sign language into text or speech, SignSpeak contributes to breaking down communication barriers and promoting equal participation for all members of society.

Exploring the Association Between Structural Racism and Mental Health: Geospatial and Machine Learning Analysis.

Structural racism produces mental health disparities. While studies have examined the impact of individual factors such as poverty and education, the collective contribution of these elements, as manifestations of structural racism, has been less explored. Milwaukee County, Wisconsin, with its racial and socioeconomic diversity, provides a unique context for this multifactorial investigation. This research aimed to delineate the association between structural racism and mental health disparities in Milwaukee County, using a combination of geospatial and deep learning techniques. We used secondary data sets where all data were aggregated and anonymized before being released by federal agencies. We compiled 217 georeferenced explanatory variables across domains, initially deliberately excluding race-based factors to focus on nonracial determinants. This approach was designed to reveal the underlying patterns of risk factors contributing to poor mental health, subsequently reintegrating race to assess the effects of racism quantitatively. The variable selection combined tree-based methods (random forest) and conventional techniques, supported by variance inflation factor and Pearson correlation analysis for multicollinearity mitigation. The geographically weighted random forest model was used to investigate spatial heterogeneity and dependence. Self-organizing maps, combined with K-means clustering, were used to analyze data from Milwaukee communities, focusing on quantifying the impact of structural racism on the prevalence of poor mental health. While 12 influential factors collectively accounted for 95.11% of the variability in mental health across communities, the top 6 factors-smoking, poverty, insufficient sleep, lack of health insurance, employment, and age-were particularly impactful. Predominantly, African American neighborhoods were disproportionately affected, which is 2.23 times more likely to encounter high-risk clusters for poor mental health. The findings demonstrate that structural racism shapes mental health disparities, with Black community members disproportionately impacted. The multifaceted methodological approach underscores the value of integrating geospatial analysis and deep learning to understand complex social determinants of mental health. These insights highlight the need for targeted interventions, addressing both individual and systemic factors to mitigate mental health disparities rooted in structural racism.

Improving distribution models of sparsely documented disease vectors by incorporating information on related species via joint modeling

A necessary component of understanding vector‐borne disease risk is accurate characterization of the distributions of their vectors. Species distribution models have been successfully applied to data‐rich species but may produce inaccurate results for sparsely documented vectors. In light of global change, vectors that are currently not well‐documented could become increasingly important, requiring tools to predict their distributions. One way to achieve this could be to leverage data on related species to inform the distribution of a sparsely documented vector based on the assumption that the environmental niches of related species are not independent. Relatedly, there is a natural dependence of the spatial distribution of a disease on the spatial dependence of its vector. Here, we propose to exploit these correlations by fitting a hierarchical model jointly to data on multiple vector species and their associated human diseases to improve distribution models of sparsely documented species. To demonstrate this approach, we evaluated the ability of twelve models – which differed in their pooling of data from multiple vector species and inclusion of disease data – to improve distribution estimates of sparsely documented vectors. We assessed our models on two simulated datasets, which allowed us to generalize our results and examine their mechanisms. We found that when the focal species is sparsely documented, incorporating data on related vector species reduces uncertainty and improves accuracy by reducing overfitting. When data on vector species are already incorporated, disease data only marginally improve model performance. However, when data on other vectors are not available, disease data can improve model accuracy and reduce overfitting and uncertainty. We then assessed the approach on empirical data on ticks and tick‐borne diseases in Florida and found that incorporating data on other vector species improved model performance. This study illustrates the value in exploiting correlated data via joint modeling to improve distribution models of data‐limited species.

Measuring the productivity of urban commercial banks in China

This paper contributes to the productivity analysis literature by measuring the productivity of 76 urban commercial banks operating in China's 27 regions from 2013 to 2018. A spatial total factor productivity index is constructed using a spatial autoregressive production function. The index is decomposed into the efficiency change and the returns to scale change, and it identifies both spatial direct (feedback) and spatial indirect (spillovers) components. The result confirms positive spatial dependence in the production of urban commercial banks. Additionally, productivity of China's urban commercial banks increases, which is mainly contributed to by the returns to scale change.

The role of tourism in European regions’ economic recovery: A spatial perspective

The study searches for the presence of a tourism-led growth path after a negative shock. Both the presence of a tourism role in the recovery process and the existence of a spatial interdependence across regions are investigated. The purpose of the study is to look at insights, provided in particular by the experience of the 2007–2008 crisis in European regions, that can be useful to build a recovery after any future shocks, as for example the COVID-19 pandemic resulted to be. The findings show the presence of a strong spatial dependence in the recovery process and, furthermore, the positive potential that tourism activities exert on the ability to recover favouring positive spillovers across regions.

Estimating Spatio-Temporal Building Power Consumption Based on Graph Convolution Network Method

Buildings are responsible for around 30% and 42% of the consumed energy at the global and European levels, respectively. Accurate building power consumption estimation is crucial for resource saving. This research investigates the combination of graph convolutional networks (GCNs) and long short-term memory networks (LSTMs) to analyze power building consumption, thereby focusing on predictive modeling. Specifically, by structuring graphs based on Pearson’s correlation and Euclidean distance methods, GCNs are employed to discern intricate spatial dependencies, and LSTM is used for temporal dependencies. The proposed models are applied to data from a multistory, multizone educational building, and they are then compared with baseline machine learning, deep learning, and statistical models. The performance of all models is evaluated using metrics such as the mean absolute error (MAE), mean squared error (MSE), R-squared (R2), and the coefficient of variation of the root mean squared error (CV(RMSE)). Among the proposed computation models, one of the Euclidean-based models consistently achieved the lowest MAE and MSE values, thus indicating superior prediction accuracy. The suggested methods seem promising and highlight the effectiveness of GCNs in improving accuracy and reliability in predicting power consumption. The results could be useful in the planning of building energy policies by engineers, as well as in the evaluation of the energy management of structures.