Direct fitting of dynamic models using integrated nested Laplace approximations — INLA

The emergence and shift in seasonality of Lyme borreliosis in Northern Europe

Climate change has had a major impact on seasonal weather patterns, resulting in marked phenological changes in a wide range of taxa. However, empirical studies of how changes in seasonality impact the emergence and seasonal dynamics of vector-borne diseases have been limited. Lyme borreliosis, a bacterial infection spread by hard-bodied ticks, is the most common vector-borne disease in the northern hemisphere and has been rapidly increasing in both incidence and geographical distribution in many regions of Europe and North America. By analysis of long-term surveillance data (1995-2019) from across Norway (latitude 57°58'-71°08' N), we demonstrate a marked change in the within-year timing of Lyme borreliosis cases accompanying an increase in the annual number of cases. The seasonal peak in cases is now six weeks earlier than 25 years ago, exceeding seasonal shifts in plant phenology and previous model predictions. The seasonal shift occurred predominantly in the first 10 years of the study period. The concurrent upsurgence in case number and shift in case timing indicate a major change in the Lyme borreliosis disease system over recent decades. This study highlights the potential for climate change to shape the seasonal dynamics of vector-borne disease systems.

Under age five children survival times in Nigeria: a Bayesian spatial modeling approach

BACKGROUND

Nigeria is among the top five countries in the world with the highest under-five mortality rates. In addition to the general leading causes of under-five mortality, studies have shown that disparity in sociocultural values and practices across ethnic groups in Nigeria influence child survival, thus there is a need for scientific validation. This study quantified the survival probabilities and the impact of socioeconomic and demographic factors, proximate and biological determinants, and environmental factors on the risk of under-five mortality in Nigeria.

METHODS

The Kaplan-Meier survival curve, Nelson Aalen hazard curve, and components survival probabilities were estimated. The Exponential, Gamma, Log-normal, Weibull, and Cox hazard models in a Bayesian mixed effect hierarchical hazard modeling framework with spatial components were considered, and the Deviance and Watanabe Akaike information criteria were used to select the best model for inference. A [Formula: see text] level of significance was assumed throughout this work. The 2018 Nigeria Demographic and Health Survey dataset was used, and the outcome variable was the time between birth and death or birth and the date of interview for children who were alive on the day of the interview.

RESULTS

Findings show that the probability of a child dying within the first two months is 0.04, and the probability of a boy child dying before attaining age five is 0.106, while a girl child is 0.094 probability. Gender, maternal education, household wealth status, source of water and toilet facility, residence, mass media, frequency of antenatal and postnatal visits, marital status, place of delivery, multiple births, who decide healthcare use, use of bednet are significant risk factors of child mortality in Nigeria. The mortality risk is high among the maternal age group below 24 and above 44years, and birth weight below 2.5Kg and above 4.5Kg. The under-five mortality risk is severe in Kebbi, Kaduna, Jigawa, Adamawa, Gombe, Kano, Kogi, Nasarawa, Plateau, and Sokoto states in Nigeria.

CONCLUSION

This study accentuates the need for special attention for the first two months after childbirth as it is the age group with the highest expected mortality. A practicable way to minimize death in the early life of children is to improve maternal healthcare service, promote maternal education, encourage delivery in healthcare facilities, positive parental attitude to support multiple births, poverty alleviation programs for the less privileged, and a prioritized intervention to Northern Nigeria.

Length, width, shape regularity, and chain structure: time series analysis of phytoplankton morphology from imagery

Functional traits are increasingly used to assess changes in phytoplankton community structure and to link individual characteristics to ecosystem functioning. However, they are usually inferred from taxonomic identification or manually measured for each organism, both time consuming approaches. Instead, we focus on high throughput imaging to describe the main temporal variations of morphological changes of phytoplankton in Narragansett Bay, a coastal time-series station. We analyzed a 2-yr dataset of morphological features automatically extracted from continuous imaging of individual phytoplankton images (~ 105 million images collected by an Imaging FlowCytobot). We identified synthetic morphological traits using multivariate analysis and revealed that morphological variations were mainly due to changes in length, width, shape regularity, and chain structure. Morphological changes were especially important in winter with successive peaks of larger cells with increasing complexity and chains more clearly connected. Small nanophytoplankton were present year-round and constituted the base of the community, especially apparent during the transitions between diatom blooms. High inter-annual variability was also observed. On a weekly timescale, increases in light were associated with more clearly connected chains while more complex shapes occurred at lower nitrogen concentrations. On an hourly timescale, temperature was the determinant variable constraining cell morphology, with a general negative influence on length and a positive one on width, shape regularity, and chain structure. These first insights into the phytoplankton morphology of Narragansett Bay highlight the possible morphological traits driving the phytoplankton succession in response to light, temperature, and nutrient changes.

Bayesian explanatory additive IRT models

In this article we extend the framework of explanatory mixed IRT models to a more general class called explanatory additive IRT models. We do this by augmenting the linear predictors in terms of smooth functions. This development offers many new modeling options such as the inclusion of nonlinear covariate effects, the specification of various temporal and spatial dependency patterns, and parameter partitioning across covariates. We use integrated nested Laplace approximation (INLA) for accurate and computationally efficient estimation of the parameters. Uninformative, weakly informative, and informative prior settings for the hyperparameters are discussed. Running time experiments and Monte Carlo parameter recovery simulations are performed in order to study the accuracy and computational efficiency of INLA when applied to the proposed explanatory additive IRT model class. Using a real-life dataset, a variety of application scenarios is explored, and the results are compared with classical maximum likelihood estimation when possible. R code is included in the supplemental materials to allow readers to fully reproduce the examples computed in the paper.

Variability and stability in child maltreatment risk across time and space and its association with neighborhood social & housing vulnerability in New Mexico: A bayesian space-time model

BACKGROUND

Modeling the spatio-temporal characteristics of substantiated child maltreatment risk has significant implications for child welfare policy.

OBJECTIVE

This study quantifies the spatiotemporal risk of child abuse and neglect in New Mexico at the census tract level over 9 years, identifies areas of increased risk, and evaluates the role of multiple measures of social and housing insecurity on substantiated child maltreatment referrals.

PARTICIPANTS AND SETTING

Child maltreatment substantiation data across 499 census tracts from 2007 to 2015 were obtained from the New Mexico Department of Public Health.

METHODS

Substantiated referral counts were analyzed within census tracts with Bayesian hierarchical space-time models using Laplace approximation. Standardized incidence ratios, spatial risk, and probability exceedances were calculated and mapped.

RESULTS

Multiple neighborhood structural factors were associated with an increased risk of substantiated child maltreatment, including the eviction rate (Incidence Density Ratio [IDR] = 1.09 [95 % CrI = 1.01-1.12]), rent burden (IDR = 1.11 [95 % CrI = 1.01-1.13]), urban tracts (IDR = 1.36 [95 % CrI = 1.05-1.77]), food desert tracts (IDR = 1.21 [95 % CrI = 1.04-1.41]), low income tracts (IDR = 1.27 [95 % CrI = 1.09-1.49]), percent of households with no vehicle access ([IDR] = 1.27 [95 % CrI = .247-6.47]), and percent of persons with a disability (IDR = 1.05 [95 % CrI = 1.03-1.06]). The racial/ethnic diversity ratio, however, was associated with lower incidence of child maltreatment allegation risk (IDR = .988 [95 % CrI = .982-.995]).

CONCLUSIONS

Population-based child abuse and neglect prevention and intervention efforts should be aided by the characteristics of neighborhoods that demonstrate strong spatial patterns of household and housing vulnerability, particularly in low income, racially segregated neighborhoods.

Statistical tests for detecting variance effects in quantitative trait studies

Motivation

Identifying variants, both discrete and continuous, that are associated with quantitative traits, or QTs, is the primary focus of quantitative genetics. Most current methods are limited to identifying mean effects, or associations between genotype or covariates and the mean value of a quantitative trait. It is possible, however, that a variant may affect the variance of the quantitative trait in lieu of, or in addition to, affecting the trait mean. Here, we develop a general methodology to identify covariates with variance effects on a quantitative trait using a Bayesian heteroskedastic linear regression model (BTH). We compare BTH with existing methods to detect variance effects across a large range of simulations drawn from scenarios common to the analysis of quantitative traits.

Results

We find that BTH and a double generalized linear model (dglm) outperform classical tests used for detecting variance effects in recent genomic studies. We show BTH and dglm are less likely to generate spurious discoveries through simulations and application to identifying methylation variance QTs and expression variance QTs. We identify four variance effects of sex in the Cardiovascular and Pharmacogenetics study. Our work is the first to offer a comprehensive view of variance identifying methodology. We identify shortcomings in previously used methodology and provide a more conservative and robust alternative. We extend variance effect analysis to a wide array of covariates that enables a new statistical dimension in the study of sex and age specific quantitative trait effects.

Availability and implementation

https://github.com/b2du/bth.

Supplementary information

Supplementary data are available at Bioinformatics online.

The Geography of Child Maltreatment: A Spatiotemporal Analysis Using Bayesian Hierarchical Analysis With Integrated Nested Laplace Approximation

This study quantifies the spatiotemporal risk of child abuse and neglect in Los Angeles at the census tract level over a recent 4-year period, identifies areas of increased risk, and evaluates the role of structural disadvantage in substantiated child maltreatment referrals. Child maltreatment data on 83,379 child maltreatment cases in 1,678 census tracts spanning 2006-2009 were obtained from the Los Angeles County Department of Children and Family Services. Substantiated referral counts were analyzed across census tracts with Bayesian hierarchical spatial models using integrated nested Laplace approximations. Results showed that the unadjusted yearly rate of child abuse and neglect held fairly steady over the study period decreasing by only 2.57%. However, the temporal term in the spatiotemporal model reflected a downward trend beginning in 2007. High rates of abuse and neglect were predicted by several neighborhood-level measures of structural burden. Every 1-unit decrease in the social vulnerability index reduced the risk of child abuse and neglect by 98.3% (95% CrI = 1.869-2.1042) while every 1-unit increase in the Black-White dissimilarity index decreased child abuse and neglect risk by 70.6%. The interaction of these variables demonstrated the protective effect of racial heterogeneity in socially vulnerable neighborhoods. No such effect was found in neighborhoods characterized by low levels of vulnerability. Population-based child abuse and neglect prevention and intervention efforts should be aided by the characteristics of neighborhoods that demonstrate strong spatial patterns even after accounting for the role of race and place.

Estimating the number of people living with HIV and the undiagnosed fraction in Spain in 2013

OBJECTIVE

We aim to estimate the number of people living with HIV and the undiagnosed fraction in Spain, where coverage of the HIV surveillance system has only recently become complete.

METHODS

The reconstruction of all HIV diagnoses and infections was obtained by combining HIV and AIDS surveillance data. The imputation of the diagnoses and back-calculation of the infection incidence are integrated in a Bayesian framework to take into account the uncertainty associated with unavailable data.

RESULTS

An estimated 141 000 [95% credible interval (CI) 128 000-155 000] persons were living with HIV by the end of 2013, in Spain and 18% (95% CI 14.3-22.1%) were unaware of it. A similar fraction of undiagnosed infections was obtained in men who have sex with men and heterosexuals (18.8 and 20.1%, respectively), but for injection drug users, this fraction was 3.5%.

CONCLUSION

This study provides the first estimates of the number of people living with HIV and the undiagnosed fraction in Spain, using routine surveillance data. The proposed method could be useful for countries where the geographical coverage of the HIV surveillance system is partial or was completed only recently.

Evaluation of spatio-temporal Bayesian models for the spread of infectious diseases in oil palm

In the field of epidemiology, studies are often focused on mapping diseases in relation to time and space. Hierarchical modeling is a common flexible and effective tool for modeling problems related to disease spread. In the context of oil palm plantations infected by the fungal pathogen Ganoderma boninense, we propose and compare two spatio-temporal hierarchical Bayesian models addressing the lack of information on propagation modes and transmission vectors. We investigate two alternative process models to study the unobserved mechanism driving the infection process. The models help gain insight into the spatio-temporal dynamic of the infection by identifying a genetic component in the disease spread and by highlighting a spatial component acting at the end of the experiment. In this challenging context, we propose models that provide assumptions on the unobserved mechanism driving the infection process while making short-term predictions using ready-to-use software.

Monitoring schistosomiasis risk in East China over space and time using a Bayesian hierarchical modeling approach

Schistosomiasis remains a major public health problem and causes substantial economic impact in east China, particularly along the Yangtze River Basin. Disease forecasting and surveillance can assist in the development and implementation of more effective intervention measures to control disease. In this study, we applied a Bayesian hierarchical spatio-temporal model to describe trends in schistosomiasis risk in Anhui Province, China, using annual parasitological and environmental data for the period 1997–2010. A computationally efficient approach–Integrated Nested Laplace Approximation–was used for model inference. A zero-inflated, negative binomial model best described the spatio-temporal dynamics of schistosomiasis risk. It predicted that the disease risk would generally be low and stable except for some specific, local areas during the period 2011–2014. High-risk counties were identified in the forecasting maps: three in which the risk remained high, and two in which risk would become high. The results indicated that schistosomiasis risk has been reduced to consistently low levels throughout much of this region of China; however, some counties were identified in which progress in schistosomiasis control was less than satisfactory. Whilst maintaining overall control, specific interventions in the future should focus on these refractive counties as part of a strategy to eliminate schistosomiasis from this region.

Joint estimation over multiple individuals improves behavioural state inference from animal movement data

State-space models provide a powerful way to scale up inference of movement behaviours from individuals to populations when the inference is made across multiple individuals. Here, I show how a joint estimation approach that assumes individuals share identical movement parameters can lead to improved inference of behavioural states associated with different movement processes. I use simulated movement paths with known behavioural states to compare estimation error between nonhierarchical and joint estimation formulations of an otherwise identical state-space model. Behavioural state estimation error was strongly affected by the degree of similarity between movement patterns characterising the behavioural states, with less error when movements were strongly dissimilar between states. The joint estimation model improved behavioural state estimation relative to the nonhierarchical model for simulated data with heavy-tailed Argos location errors. When applied to Argos telemetry datasets from 10 Weddell seals, the nonhierarchical model estimated highly uncertain behavioural state switching probabilities for most individuals whereas the joint estimation model yielded substantially less uncertainty. The joint estimation model better resolved the behavioural state sequences across all seals. Hierarchical or joint estimation models should be the preferred choice for estimating behavioural states from animal movement data, especially when location data are error-prone.

Using spatiotemporal species distribution models to identify temporally evolving hotspots of species co-occurrence

Identifying spatiotemporal hotspots is important for understanding basic ecological processes, but is particularly important for species at risk. A number of terrestrial and aquatic species are indirectly affected by anthropogenic impacts, simply because they tend to be associated with species that are targeted for removals. Using newly developed statistical models that allow for the inclusion of time-varying spatial effects, we examine how the co-occurrence of a targeted and nontargeted species can be modeled as a function of environmental covariates (temperature, depth) and interannual variability. The nontarget species in our case study (eulachon) is listed under the U.S. Endangered Species Act, and is encountered by fisheries off the U.S. West Coast that target pink shrimp. Results from our spatiotemporal model indicated that eulachon bycatch risk decreases with depth and has a convex relationship with sea surface temperature. Additionally, we found that over the 2007-2012 period, there was support for an increase in eulachon density from both a fishery data set (+40%) and a fishery-independent data set (+55%). Eulachon bycatch has increased in recent years, but the agreement between these two data sets implies that increases in bycatch are not due to an increase in incidental targeting of eulachon by fishing vessels, but because of an increasing population size of eulachon. Based on our results, the application of spatiotemporal models to species that are of conservation concern appears promising in identifying the spatial distribution of environmental and anthropogenic risks to the population.

Combined effects of climate, predation, and density dependence on Greater and Lesser Scaup population dynamics

An understanding of species relationships is critical in the management and conservation of populations facing climate change, yet few studies address how climate alters species interactions and other population drivers. We use a long-term, broad-scale data set of relative abundance to examine the influence of climate, predators, and density dependence on the population dynamics of declining scaup (Aythya) species within the core of their breeding range. The state-space modeling approach we use applies to a wide range of wildlife species, especially populations monitored over broad spatiotemporal extents. Using this approach, we found that immediate snow cover extent in the preceding winter and spring had the strongest effects, with increases in mean snow cover extent having a positive effect on the local surveyed abundance of scaup. The direct effects of mesopredator abundance on scaup population dynamics were weaker, but the results still indicated a potentil interactive process between climate and food web dynamics (mesopredators, alternative prey, and scaup). By considering climate variables and other potential effects on population dynamics, and using a rigorous estimation framework, we provide insight into complex ecological processes for guiding. conservation and policy actions aimed at mitigating and reversing the decline of scaup.

Dynamic compositional modeling of pedestrian crash counts on urban roads in Connecticut

Uncovering the temporal trend in crash counts provides a good understanding of the context for pedestrian safety. With a rareness of pedestrian crashes it is impossible to investigate monthly temporal effects with an individual segment/intersection level data, thus the time dependence should be derived from the aggregated level data. Most previous studies have used annual data to investigate the differences in pedestrian crashes between different regions or countries in a given year, and/or to look at time trends of fatal pedestrian injuries annually. Use of annual data unfortunately does not provide sufficient information on patterns in time trends or seasonal effects. This paper describes statistical methods uncovering patterns in monthly pedestrian crashes aggregated on urban roads in Connecticut from January 1995 to December 2009. We investigate the temporal behavior of injury severity levels, including fatal (K), severe injury (A), evident minor injury (B), and non-evident possible injury and property damage only (C and O), as proportions of all pedestrian crashes in each month, taking into consideration effects of time trend, seasonal variations and VMT (vehicle miles traveled). This type of dependent multivariate data is characterized by positive components which sum to one, and occurs in several applications in science and engineering. We describe a dynamic framework with vector autoregressions (VAR) for modeling and predicting compositional time series. Combining these predictions with predictions from a univariate statistical model for total crash counts will then enable us to predict pedestrian crash counts with the different injury severity levels. We compare these predictions with those obtained from fitting separate univariate models to time series of crash counts at each injury severity level. We also show that the dynamic models perform better than the corresponding static models. We implement the Integrated Nested Laplace Approximation (INLA) approach to enable fast Bayesian posterior computation. Taking CO injury severity level as a baseline for the compositional analysis, we conclude that there was a noticeable shift in the proportion of pedestrian crashes from injury severity A to B, while the increase for injury severity K was extremely small over time. This shift to the less severe injury level (from A to B) suggests that the overall safety on urban roads in Connecticut is improving. In January and February, there was some increase in the proportions for levels A and B over the baseline, indicating a seasonal effect. We found evidence that an increase in VMT would result in a decrease of proportions over the baseline for all injury severity levels. Our dynamic model uncovered a decreasing trend in all pedestrian crash counts before April 2005, followed by a noticeable increase and a flattening out until the end of the fitting period. This appears to be largely due to the behavior of injury severity level A pedestrian crashes.