David Welch

Approximate Bayesian computation scheme for parameter inference and model selection in dynamical systems

Approximate Bayesian computation (ABC) methods can be used to evaluate posterior distributions without having to calculate likelihoods. In this paper, we discuss and apply an ABC method based on sequential Monte Carlo (SMC) to estimate parameters of dynamical models. We show that ABC SMC provides information about the inferability of parameters and model sensitivity to changes in parameters, and tends to perform better than other ABC approaches. The algorithm is applied to several well-known biological systems, for which parameters and their credible intervals are inferred. Moreover, we develop ABC SMC as a tool for model selection; given a range of different mathematical descriptions, ABC SMC is able to choose the best model using the standard Bayesian model selection apparatus.

Evaluating the impact of wind turbine noise on health-related quality of life

We report a cross-sectional study comparing the health-related quality of life (HRQOL) of individuals residing in the proximity of a wind farm to those residing in a demographically matched area sufficiently displaced from wind turbines. The study employed a nonequivalent comparison group posttest-only design. Self-administered questionnaires, which included the brief version of the World Health Organization quality of life scale, were delivered to residents in two adjacent areas in semirural New Zealand. Participants were also asked to identify annoying noises, indicate their degree of noise sensitivity, and rate amenity. Statistically significant differences were noted in some HRQOL domain scores, with residents living within 2 km of a turbine installation reporting lower overall quality of life, physical quality of life, and environmental quality of life. Those exposed to turbine noise also reported significantly lower sleep quality, and rated their environment as less restful. Our data suggest that wind farm noise can negatively impact facets of HRQOL.

Effects of cannabis on lung function: a population-based cohort study

The effects of cannabis on lung function remain unclear and may be different from those of tobacco. We compared the associations between use of these substances and lung function in a population-based cohort (n = 1,037). Cannabis and tobacco use were reported at ages 18, 21, 26 and 32 yrs. Spirometry, plethysmography and carbon monoxide transfer factor were measured at 32 yrs. Associations between lung function and exposure to each substance were adjusted for exposure to the other substance. Cumulative cannabis use was associated with higher forced vital capacity, total lung capacity, functional residual capacity and residual volume. Cannabis was also associated with higher airway resistance but not with forced expiratory volume in 1 s, forced expiratory ratio or transfer factor. These findings were similar among those who did not smoke tobacco. In contrast, tobacco use was associated with lower forced expiratory volume in 1 s, lower forced expiratory ratio, lower transfer factor and higher static lung volumes, but not with airway resistance. Cannabis appears to have different effects on lung function from those of tobacco. Cannabis use was associated with higher lung volumes, suggesting hyperinflation and increased large-airways resistance, but there was little evidence for airflow obstruction or impairment of gas transfer.

Bayesian Inference of Sampled Ancestor Trees for Epidemiology and Fossil Calibration

Phylogenetic analyses which include fossils or molecular sequences that are sampled through time require models that allow one sample to be a direct ancestor of another sample. As previously available phylogenetic inference tools assume that all samples are tips, they do not allow for this possibility. We have developed and implemented a Bayesian Markov Chain Monte Carlo (MCMC) algorithm to infer what we call sampled ancestor trees, that is, trees in which sampled individuals can be direct ancestors of other sampled individuals. We use a family of birth-death models where individuals may remain in the tree process after sampling, in particular we extend the birth-death skyline model [Stadler et al., 2013] to sampled ancestor trees. This method allows the detection of sampled ancestors as well as estimation of the probability that an individual will be removed from the process when it is sampled. We show that even if sampled ancestors are not of specific interest in an analysis, failing to account for them leads to significant bias in parameter estimates. We also show that sampled ancestor birth-death models where every sample comes from a different time point are non-identifiable and thus require one parameter to be known in order to infer other parameters. We apply our phylogenetic inference accounting for sampled ancestors to epidemiological data, where the possibility of sampled ancestors enables us to identify individuals that infected other individuals after being sampled and to infer fundamental epidemiological parameters. We also apply the method to infer divergence times and diversification rates when fossils are included along with extant species samples, so that fossilisation events are modelled as a part of the tree branching process. Such modelling has many advantages as argued in the literature. The sampler is available as an open-source BEAST2 package (https://github.com/CompEvol/sampled-ancestors).

Exploring the Relationship between Noise Sensitivity, Annoyance and Health-Related Quality of Life in a Sample of Adults Exposed to Environmental Noise

The relationship between environmental noise and health is poorly understood but of fundamental importance to public health. This study estimated the relationship between noise sensitivity, noise annoyance and health-related quality of life in a sample of adults residing close to the Auckland International Airport, New Zealand. A small sample (n = 105) completed surveys measuring noise sensitivity, noise annoyance, and quality of life. Noise sensitivity was associated with health-related quality of life; annoyance and sleep disturbance mediated the effects of noise sensitivity on health.

Adolescent screen time and attachment to parents and peers.

OBJECTIVE To examine associations between screen time (television, video or DVD, gaming, and computer use) and attachment to parents and peers in 2 cohorts of adolescents 16 years apart. DESIGN Cross-sectional data regarding screen time and attachment to parents and peers were collected for 2 cohorts of adolescents, one in 1987-1988 (the Dunedin Multidisciplinary Health and Development Study [DMHDS] cohort) and the other in 2004 (the Youth Lifestyle Study [YLS] cohort). SETTING Members of the DMHDS cohort were interviewed as part of a full day of assessment, and members of the YLS cohort completed a self-report questionnaire in a supervised classroom setting. PARTICIPANTS The DMHDS cohort (n = 976) was aged 15 years in 1987-1988. The YLS cohort (n = 3043) was aged 14 to 15 years in 2004. MAIN OUTCOME MEASURES Screen time and low attachment to parents and peers as measured by the Inventory of Parent and Peer Attachment. RESULTS More time spent television viewing and less time spent reading and doing homework were associated with low attachment to parents for both cohorts. Among the YLS cohort, more time spent playing on a computer was also associated with low attachment to parents. Among the DMHDS cohort, more time spent television viewing was associated with low attachment to peers. CONCLUSIONS Screen time was associated with poor attachment to parents and peers in 2 cohorts of adolescents 16 years apart. Given the importance of attachment to parents and peers in adolescent health and development, concern about high levels of screen time among adolescents is warranted.

Bayesian Total-Evidence Dating Reveals the Recent Crown Radiation of Penguins

Abstract The total‐evidence approach to divergence time dating uses molecular and morphological data from extant and fossil species to infer phylogenetic relationships, species divergence times, and macroevolutionary parameters in a single coherent framework. Current model‐based implementations of this approach lack an appropriate model for the tree describing the diversification and fossilization process and can produce estimates that lead to erroneous conclusions. We address this shortcoming by providing a total‐evidence method implemented in a Bayesian framework. This approach uses a mechanistic tree prior to describe the underlying diversification process that generated the tree of extant and fossil taxa. Previous attempts to apply the total‐evidence approach have used tree priors that do not account for the possibility that fossil samples may be direct ancestors of other samples, that is, ancestors of fossil or extant species or of clades. The fossilized birth‐death (FBD) process explicitly models the diversification, fossilization, and sampling processes and naturally allows for sampled ancestors. This model was recently applied to estimate divergence times based on molecular data and fossil occurrence dates. We incorporate the FBD model and a model of morphological trait evolution into a Bayesian total‐evidence approach to dating species phylogenies. We apply this method to extant and fossil penguins and show that the modern penguins radiated much more recently than has been previously estimated, with the basal divergence in the crown clade occurring at ∼12.7 Ma and most splits leading to extant species occurring in the last 2 myr. Our results demonstrate that including stem‐fossil diversity can greatly improve the estimates of the divergence times of crown taxa. The method is available in BEAST2 (version 2.4) software www.beast2.org with packages SA (version at least 1.1.4) and morph‐models (version at least 1.0.4) installed.

Statistical inference to advance network models in epidemiology.

Contact networks are playing an increasingly important role in the study of epidemiology. Most of the existing work in this area has focused on considering the effect of underlying network structure on epidemic dynamics by using tools from probability theory and computer simulation. This work has provided much insight on the role that heterogeneity in host contact patterns plays on infectious disease dynamics. Despite the important understanding afforded by the probability and simulation paradigm, this approach does not directly address important questions about the structure of contact networks such as what is the best network model for a particular mode of disease transmission, how parameter values of a given model should be estimated, or how precisely the data allow us to estimate these parameter values. We argue that these questions are best answered within a statistical framework and discuss the role of statistical inference in estimating contact networks from epidemiological data.

Do quiet areas afford greater health-related quality of life than noisy areas?

People typically choose to live in quiet areas in order to safeguard their health and wellbeing. However, the benefits of living in quiet areas are relatively understudied compared to the burdens associated with living in noisy areas. Additionally, research is increasingly focusing on the relationship between the human response to noise and measures of health and wellbeing, complementing traditional dose-response approaches, and further elucidating the impact of noise and health by incorporating human factors as mediators and moderators. To further explore the benefits of living in quiet areas, we compared the results of health-related quality of life (HRQOL) questionnaire datasets collected from households in localities differentiated by their soundscapes and population density: noisy city, quiet city, quiet rural, and noisy rural. The dose-response relationships between noise annoyance and HRQOL measures indicated an inverse relationship between the two. Additionally, quiet areas were found to have higher mean HRQOL domain scores than noisy areas. This research further supports the protection of quiet locales and ongoing noise abatement in noisy areas.

Road traffic noise and health-related quality of life: A cross-sectional study

Evidence is emerging linking environmental noise to health problems. Noise can affect health directly and indirectly: For example, noise sensitivity moderates the effects of noise annoyance, which in turn mediates the effects of noise exposure. An alternative hypothesis is that noise sensitivity marks the presence of susceptibility to health problems in general, including annoyance from noise. Whether noise sensitivity causes poor health or whether it is a marker of susceptibility to health problems was addressed by comparing the results of a community-based survey of people with similar noise sensitivity profiles but different environmental noise exposures. A questionnaire was delivered to people living in two socio-economically-matched areas: One was within 50 m of a motorway and the other was away from any significant source of environmental noise. The questionnaire contained 58 questions comprised of the World Health Organization health-related quality of life questionnaire (WHOQOL), and questions about amenity, neighborhood issues, environmental annoyances, demographics, and noise sensitivity. Noise sensitivity did not vary with proximity to the motorway but annoyance with traffic noise and fumes was greater in those living close to the motorway than in those who were not. Scores on the four WHOQOL domains (physical, psychological, social, and environmental) were lower in those living close to the motorway, and the WHOQOL domain scores correlated negatively with noise sensitivity in those who lived near motorways but not in those who lived in the quieter areas. This suggests that noise sensitivity is related to poor health outcomes rather than being a trait marker of susceptibility to health problems in general.