Addressing issues in sparseness, ecological bias and formulation of the adjacency matrix in Bayesian spatio-temporal analysis of disease counts

The main objective of this PhD was to further develop Bayesian spatio-temporal models (specifically the Conditional Autoregressive (CAR) class of models), for the analysis of sparse disease outcomes such as birth defects. The motivation for the thesis arose from problems encountered when analyzing a large birth defect registry in New South Wales. The specific components and related research objectives of the thesis were developed from gaps in the literature on current formulations of the CAR model, and health service planning requirements. Data from a large probabilistically-linked database from 1990 to 2004, consisting of fields from two separate registries: the Birth Defect Registry (BDR) and Midwives Data Collection (MDC) were used in the analyses in this thesis. The main objective was split into smaller goals. The first goal was to determine how the specification of the neighbourhood weight matrix will affect the smoothing properties of the CAR model, and this is the focus of chapter 6. Secondly, I hoped to evaluate the usefulness of incorporating a zero-inflated Poisson (ZIP) component as well as a shared-component model in terms of modeling a sparse outcome, and this is carried out in chapter 7. The third goal was to identify optimal sampling and sample size schemes designed to select individual level data for a hybrid ecological spatial model, and this is done in chapter 8. Finally, I wanted to put together the earlier improvements to the CAR model, and along with demographic projections, provide forecasts for birth defects at the SLA level. Chapter 9 describes how this is done. For the first objective, I examined a series of neighbourhood weight matrices, and showed how smoothing the relative risk estimates according to similarity by an important covariate (i.e. maternal age) helped improve the model’s ability to recover the underlying risk, as compared to the traditional adjacency (specifically the Queen) method of applying weights. Next, to address the sparseness and excess zeros commonly encountered in the analysis of rare outcomes such as birth defects, I compared a few models, including an extension of the usual Poisson model to encompass excess zeros in the data. This was achieved via a mixture model, which also encompassed the shared component model to improve on the estimation of sparse counts through borrowing strength across a shared component (e.g. latent risk factor/s) with the referent outcome (caesarean section was used in this example). Using the Deviance Information Criteria (DIC), I showed how the proposed model performed better than the usual models, but only when both outcomes shared a strong spatial correlation. The next objective involved identifying the optimal sampling and sample size strategy for incorporating individual-level data with areal covariates in a hybrid study design. I performed extensive simulation studies, evaluating thirteen different sampling schemes along with variations in sample size. This was done in the context of an ecological regression model that incorporated spatial correlation in the outcomes, as well as accommodating both individual and areal measures of covariates. Using the Average Mean Squared Error (AMSE), I showed how a simple random sample of 20% of the SLAs, followed by selecting all cases in the SLAs chosen, along with an equal number of controls, provided the lowest AMSE. The final objective involved combining the improved spatio-temporal CAR model with population (i.e. women) forecasts, to provide 30-year annual estimates of birth defects at the Statistical Local Area (SLA) level in New South Wales, Australia. The projections were illustrated using sixteen different SLAs, representing the various areal measures of socio-economic status and remoteness. A sensitivity analysis of the assumptions used in the projection was also undertaken. By the end of the thesis, I will show how challenges in the spatial analysis of rare diseases such as birth defects can be addressed, by specifically formulating the neighbourhood weight matrix to smooth according to a key covariate (i.e. maternal age), incorporating a ZIP component to model excess zeros in outcomes and borrowing strength from a referent outcome (i.e. caesarean counts). An efficient strategy to sample individual-level data and sample size considerations for rare disease will also be presented. Finally, projections in birth defect categories at the SLA level will be made.

Inferring lung cancer risk factor patterns through joint Bayesian spatio-temporal analysis

Background: Preventing risk factor exposure is vital to reduce the high burden from lung cancer. The leading risk factor for developing lung cancer is tobacco smoking. In Australia, despite apparent success in reducing smoking prevalence, there is limited information on small area patterns and small area temporal trends. We sought to estimate spatio-temporal patterns for lung cancer risk factors using routinely collected population-based cancer data. Methods: The analysis used a Bayesian shared component spatio-temporal model, with male and female lung cancer included separately. The shared component reflected exposure to lung cancer risk factors, and was modelled over 477 statistical local areas (SLAs) and 15 years in Queensland, Australia. Analyses were also run adjusting for area-level socioeconomic disadvantage, Indigenous population composition, or remoteness. Results: Strong spatial patterns were observed in the underlying risk factor exposure for both males (median Relative Risk (RR) across SLAs compared to the Queensland average ranged from 0.48-2.00) and females (median RR range across SLAs 0.53-1.80), with high exposure observed in many remote areas. Strong temporal trends were also observed. Males showed a decrease in the underlying risk across time, while females showed an increase followed by a decrease in the final two years. These patterns were largely consistent across each SLA. The high underlying risk estimates observed among disadvantaged, remote and indigenous areas decreased after adjustment, particularly among females. Conclusion: The modelled underlying exposure appeared to reflect previous smoking prevalence, with a lag period of around 30 years, consistent with the time taken to develop lung cancer. The consistent temporal trends in lung cancer risk factors across small areas support the hypothesis that past interventions have been equally effective across the state. However, this also means that spatial inequalities have remained unaddressed, highlighting the potential for future interventions, particularly among remote areas.

Spatio-temporal analysis of DNA damage repair using the X-ray microbeam

Cellular response to radiation damage is made by a complex network of pathways and feedback loops whose spatiotemporal organization is still unclear despite its decisive role in determining the fate of the damaged cell. The single-cell approach and the high spatial resolution offered by microbeams provide the perfect tool to study and quantify the dynamic processes associated with the induction and repair of DNA damage. The soft X-ray microbeam has been used to follow the development of radiation induced foci in live cells by monitoring their size and intensity as a function of dose and time using yellow fluorescent protein (YFP) tagging techniques. Preliminary data indicate a delayed and linear rising of the intensity signal indicating a slow kinetic for the accumulation of DNA repair protein 53BP1. A slow and limited foci diffusion has also been observed. Further investigations are required to assess whatever such diffusion is consistent with a random walk pattern or if it is the result of a more structured lesion processing phenomenon. In conclusion, our data indicates that the use of microbeams coupled to live cell microscopy represent a sophisticated approach for visualizing and quantifying the dynamics changes of DNA proteins at the damaged sites.

Spatio-temporal analysis of an invasive plant pathogen (Phytophthora ramorum) in England and Wales

Phytophthora ramorum is a damaging invasive plant pathogen and was first discovered in the UK in 2002. Spatial point analyses were applied to the occurrence of this disease in England and Wales during the period of 2003-2006 in order to assess its spatio-temporal spread. Out of the 4301 garden centres and nurseries (GCN) surveyed, there were 164, 105, 123 and 41 sites with P. ramorum in 2003, 2004, 2005 and 2006, respectively. Spatial analysis of the observed point patterns of GCN outbreaks suggested that these sites were significantly clumped within a radius of ca 60 km in 2003, but not in later years. Further analyses were conducted to determine the relationship of GCN outbreak sites over two consecutive years and thus to infer possible disease spread over time. This analysis suggested that disease spread among GCN sites was most likely to have occurred within a distance of 60 km for 2003-2004, but not for the later years. There were 35, 63, 81 and 58 sites with P. ramorum in the semi-natural environment (SNE). Analyses were carried out to assess whether infected GCN sites could act as an inoculum source of infected SNE plants or vice versa. In all years, there was a significant spatial closeness among GCN and SNE outbreak sites within a distance of 1 km. But a significant relationship over a longer distance (within 60 km) was only observed between cases in 2003 and 2004. These analyses suggest that statutory actions taken so far appear to have reduced the extent of long-distance spread of P. ramorum among garden centres and nurseries, but not the disease spread at a shorter distance between GCN and SNE sites.

Spatio-temporal analysis of house price convergence based on a demographical distance

Convergences of house prices have been studied for over three decades, but yet have been confirmed because of spatial heterogeneity and autocorrelations in house prices. A spatio-temporal approach was recently proposed to address the spatial and temporal issues related to house prices. However, most previous studies placed the focus on the spatial heterogeneity and autocorrelations from geographical locations, which neglected other spatial factors. In order to overcome this shortfall, this research argued a demographical distance, constructed by demographical structure and housing market scales, to investigate the house price convergences in Australian capital cities. The results confirmed the house price levels in Canberra, Brisbane and Perth converged to the house price level in Sydney.

Mixed Spatial and Nonspatial Problems in Location Based Services

With hundreds of millions of users reporting locations and embracing mobile technologies, Location Based Services (LBSs) are raising new challenges. In this dissertation, we address three emerging problems in location services, where geolocation data plays a central role. First, to handle the unprecedented growth of generated geolocation data, existing location services rely on geospatial database systems. However, their inability to leverage combined geographical and textual information in analytical queries (e.g. spatial similarity joins) remains an open problem. To address this, we introduce SpsJoin, a framework for computing spatial set-similarity joins. SpsJoin handles combined similarity queries that involve textual and spatial constraints simultaneously. LBSs use this system to tackle different types of problems, such as deduplication, geolocation enhancement and record linkage. We define the spatial set-similarity join problem in a general case and propose an algorithm for its efficient computation. Our solution utilizes parallel computing with MapReduce to handle scalability issues in large geospatial databases. Second, applications that use geolocation data are seldom concerned with ensuring the privacy of participating users. To motivate participation and address privacy concerns, we propose iSafe, a privacy preserving algorithm for computing safety snapshots of co-located mobile devices as well as geosocial network users. iSafe combines geolocation data extracted from crime datasets and geosocial networks such as Yelp. In order to enhance iSafe's ability to compute safety recommendations, even when crime information is incomplete or sparse, we need to identify relationships between Yelp venues and crime indices at their locations. To achieve this, we use SpsJoin on two datasets (Yelp venues and geolocated businesses) to find venues that have not been reviewed and to further compute the crime indices of their locations. Our results show a statistically significant dependence between location crime indices and Yelp features. Third, review centered LBSs (e.g., Yelp) are increasingly becoming targets of malicious campaigns that aim to bias the public image of represented businesses. Although Yelp actively attempts to detect and filter fraudulent reviews, our experiments showed that Yelp is still vulnerable. Fraudulent LBS information also impacts the ability of iSafe to provide correct safety values. We take steps toward addressing this problem by proposing SpiDeR, an algorithm that takes advantage of the richness of information available in Yelp to detect abnormal review patterns. We propose a fake venue detection solution that applies SpsJoin on Yelp and U.S. housing datasets. We validate the proposed solutions using ground truth data extracted by our experiments and reviews filtered by Yelp.

Insights to urban dynamics through landscape spatial pattern analysis

Urbanisation is a dynamic complex phenomenon involving large scale changes in the land uses at local levels. Analyses of changes in land uses in urban environments provide a historical perspective of land use and give an opportunity to assess the spatial patterns, correlation, trends, rate and impacts of the change, which would help in better regional planning and good governance of the region. Main objective of this research is to quantify the urban dynamics using temporal remote sensing data with the help of well-established landscape metrics. Bangalore being one of the rapidly urbanising landscapes in India has been chosen for this investigation. Complex process of urban sprawl was modelled using spatio temporal analysis. Land use analyses show 584% growth in built-up area during the last four decades with the decline of vegetation by 66% and water bodies by 74%. Analyses of the temporal data reveals an increase in urban built up area of 342.83% (during 1973-1992), 129.56% (during 1992-1999), 106.7% (1999-2002), 114.51% (2002-2006) and 126.19% from 2006 to 2010. The Study area was divided into four zones and each zone is further divided into 17 concentric circles of 1 km incrementing radius to understand the patterns and extent of the urbanisation at local levels. The urban density gradient illustrates radial pattern of urbanisation for the period 1973-2010. Bangalore grew radially from 1973 to 2010 indicating that the urbanisation is intensifying from the central core and has reached the periphery of the Greater Bangalore. Shannon's entropy, alpha and beta population densities were computed to understand the level of urbanisation at local levels. Shannon's entropy values of recent time confirms dispersed haphazard urban growth in the city, particularly in the outskirts of the city. This also illustrates the extent of influence of drivers of urbanisation in various directions. Landscape metrics provided in depth knowledge about the sprawl. Principal component analysis helped in prioritizing the metrics for detailed analyses. The results clearly indicates that whole landscape is aggregating to a large patch in 2010 as compared to earlier years which was dominated by several small patches. The large scale conversion of small patches to large single patch can be seen from 2006 to 2010. In the year 2010 patches are maximally aggregated indicating that the city is becoming more compact and more urbanised in recent years. Bangalore was the most sought after destination for its climatic condition and the availability of various facilities (land availability, economy, political factors) compared to other cities. The growth into a single urban patch can be attributed to rapid urbanisation coupled with the industrialisation. Monitoring of growth through landscape metrics helps to maintain and manage the natural resources. (C) 2012 Elsevier B.V. All rights reserved.

Analysis of rain quality data from the South African interior

Rain acidity may be ascribed to emissions from power station stacks, as well as emissions from other industry, biomass burning, maritime influences, agricultural influences, etc. Rain quality data are available for 30 sites in the South African interior, some from as early as 1985 for up to 14 rainfall seasons, while others only have relatively short records. The article examines trends over time in the raw and volume weighted concentrations of the parameters measured, separately for each of the sites for which sufficient data are available. The main thrust, however, is to examine the inter-relationship structure between the concentrations within each rain event (unweighted data), separately for each site, and to examine whether these inter-relationships have changed over time. The rain events at individual sites can be characterized by approximately eight combinations of rainfall parameters (or rain composition signatures), and these are common to all sites. Some sites will have more events from one signature than another, but there appear to be no signatures unique to a single site. Analysis via factor and cluster analysis, with a correspondence analysis of the results, also aid interpretation of the patterns. This spatio-temporal analysis, performed by pooling all rain event data, irrespective of site or time period, results in nine combinations of rainfall parameters being sufficient to characterize the rain events. The sites and rainfall seasons show patterns in these combinations of parameters, with some combinations appearing more frequently during certain rainfall seasons. In particular, the presence of the combination of low acetate and formate with high magnesium appears to be increasing in the later rainfall seasons, as does this combination together with calcium, sodium, chloride, potassium and fluoride. As expected, sites close together exhibit similar signatures. Copyright © 2002 John Wiley & Sons, Ltd.

Twenty-year advanced DInSAR analysis of severe land subsidence: The Alto Guadalentín Basin (Spain) case study

A twenty-year period of severe land subsidence evolution in the Alto Guadalentín Basin (southeast Spain) is monitored using multi-sensor SAR images, processed by advanced differential interferometric synthetic aperture radar (DInSAR) techniques. The SAR images used in this study consist of four datasets acquired by ERS-1/2, ENVISAT, ALOS and COSMO-SkyMed satellites between 1992 and 2012. The integration of ground surface displacement maps retrieved for different time periods allows us to quantify up to 2.50 m of cumulated displacements that occurred between 1992 and 2012 in the Alto Guadalentín Basin. DInSAR results were locally compared with global positioning system (GPS) data available for two continuous stations located in the study area, demonstrating the high consistency of local vertical motion measurements between the two different surveying techniques. An average absolute error of 4.6 ± 4 mm for the ALOS data and of 4.8 ± 3.5 mm for the COSMO-SkyMed data confirmed the reliability of the analysis. The spatial analysis of DInSAR ground surface displacement reveals a direct correlation with the thickness of the compressible alluvial deposits. Detected ground subsidence in the past 20 years is most likely a consequence of a 100–200 m groundwater level drop that has persisted since the 1970s due to the overexploitation of the Alto Guadalentín aquifer system. The negative gradient of the pore pressure is responsible for the extremely slow consolidation of a very thick (> 100 m) layer of fine-grained silt and clay layers with low vertical hydraulic permeability (approximately 50 mm/h) wherein the maximum settlement has still not been reached.

Spatio-temporal analyses of West Nile virus transmission in Culex mosquitoes in northern Illinois, USA, 2004

After a severe outbreak of West Nile virus (WNV) in Cook County, Illinois, in 2002, detections of WNV in mosquitoes were frequent across the state in the following years despite small numbers of human cases. We conducted a spatio-temporal analysis of Culex (subgenus Culex) mosquitoes collected in 2004 in three mosquito abatement districts (MAD) in Cook County by calculating monthly estimates of mosquito density, prevalence of infected mosquitoes, and exposure intensity, which in turn is a product of mosquito density and infection rates. Mosquito infections were detected early at three sites in late May and were widely detected throughout the three MADs in the summer with infection rates as high as 13 per 1000 in August. Exposure intensities were higher at sites adjacent to the Des Plaines River, especially in August and September. The aggregated pattern of WNV transmission along the river might be related to the existence of substantial forest preserves and wetlands that might produce ecological conditions favorable for mosquito proliferation and interactions between mosquitoes and birds.

SAQUA - sistema baseado em software livre para análise do comportamento da qualidade das águas fluviais: aplicado à bacia do rio Paraíba do Sul.

Nesta dissertação foi desenvolvido o sistema SAQUA (Sistema para Análise da Qualidade das Águas Fluviais), que permite o acompanhamento dos dados de séries históricas de parâmetros físico-químicos para análise da qualidade de águas fluviais. A alimentação do sistema SAQUA se dá a partir do arquivo tipo texto gerado no Hidroweb, sistema de banco de dados hidrológicos da ANA (Agência Nacional de Águas), disponibilizado na internet. O SAQUA constitui uma interface que permite a análise espaço-temporal de parâmetros de qualidade da água específicos definidos pelo usuário. A interface foi construída utilizando o servidor de mapas Mapserver, as linguagens HTML e PHP, além de consultas SQL e o uso do servidor Web Apache. A utilização de uma linguagem dinâmica como o PHP permitiu usar recursos internos do Mapserver por meio de funções que interagem de forma mais flexível com códigos presentes e futuros, além de interagir com o código HTML. O Sistema apresenta como resultado a representação gráfica da série histórica por parâmetro e, em mapa, a localização das estações em análise também definidas pelo usuário, geralmente associadas a uma determinada região hidrográfica. Tanto na representação gráfica da série temporal quanto em mapa, são destacados a partir de código de cores a estação de monitoramento e a observação em que os limites estabelecidos na resolução CONAMA 357/05 não foi atendido. A classe de uso da resolução CONAMA que será usada na análise também pode ser definida pelo usuário. Como caso de estudo e demonstração das funções do SAQUA foi escolhida a bacia hidrográfica do rio Paraíba do Sul, localizada na região hidrográfica Atlântico Sudeste do Brasil. A aplicação do sistema demonstrou ótimos resultados e o potencial da ferramenta computacional como apoio ao planejamento e à gestão dos recursos hídricos. Ressalta-se ainda, que todo o sistema foi desenvolvido a partir de softwares disponibilizados segundo a licença GPL de software livre, ou seja, sem custo na aquisição de licenças, demonstrando o potencial da aplicação destas ferramentas no campo dos recursos hídricos.