Being Bayesian in the 2020s: opportunities and challenges in the practice of modern applied Bayesian statistics.

Building on a strong foundation of philosophy, theory, methods and computation over the past three decades, Bayesian approaches are now an integral part of the toolkit for most statisticians and data scientists. Whether they are dedicated Bayesians or opportunistic users, applied professionals can now reap many of the benefits afforded by the Bayesian paradigm. In this paper, we touch on six modern opportunities and challenges in applied Bayesian statistics: intelligent data collection, new data sources, federated analysis, inference for implicit models, model transfer and purposeful software products. This article is part of the theme issue 'Bayesian inference: challenges, perspectives, and prospects'.

Managing surgical waiting lists through dynamic priority scoring.

Prioritising elective surgery patients under the Australian three-category system is inherently subjective due to variability in clinician decision making and the potential for extraneous factors to influence category assignment. As a result, waiting time inequities can exist which may lead to adverse health outcomes and increased morbidity, especially for patients deemed to be low priority. This study investigated the use of a dynamic priority scoring (DPS) system to rank elective surgery patients more equitably, based on a combination of waiting time and clinical factors. Such a system enables patients to progress on the waiting list in a more objective and transparent manner, at a rate relative to their clinical need. Simulation results comparing the two systems indicate that the DPS system has potential to assist in managing waiting lists by standardising waiting times relative to urgency category, in addition to improving waiting time consistency for patients of similar clinical need. In clinical practice, this system is likely to reduce subjectivity, increase transparency, and improve overall efficiency of waiting list management by providing an objective metric to prioritise patients. Such a system is also likely to increase public trust and confidence in the systems used to manage waiting lists.

Bayesian design for minimizing prediction uncertainty in bivariate spatial responses with applications to air quality monitoring.

Model-based geostatistical design involves the selection of locations to collect data to minimize an expected loss function over a set of all possible locations. The loss function is specified to reflect the aim of data collection, which, for geostatistical studies, could be to minimize the prediction uncertainty at unobserved locations. In this paper, we propose a new approach to design such studies via a loss function derived through considering the entropy about the model predictions and the parameters of the model. The approach includes a multivariate extension to generalized linear spatial models, and thus can be used to design experiments with more than one response. Unfortunately, evaluating our proposed loss function is computationally expensive so we provide an approximation such that our approach can be adopted to design realistically sized geostatistical studies. This is demonstrated through a simulated study and through designing an air quality monitoring program in Queensland, Australia. The results show that our designs remain highly efficient in achieving each experimental objective individually, providing an ideal compromise between the two objectives. Accordingly, we advocate that our approach could be adopted more generally in model-based geostatistical design.

Pattern-based assessment of the influence of rainfall characteristics on urban stormwater quality.

Urbanisation increases pollutant generation within catchments and their transport to receiving waters. Changes to rainfall patterns, particularly in the age of climate change, make pollution mitigation a challenging task. Understanding how rainfall characteristics could influence the changes to stormwater pollutant runoff is important for designing effective mitigation strategies. This study employed a pattern-based assessment of relationships between rainfall characteristics and stormwater quality in urban catchments to develop this understanding. The research outcomes showed that rainfall events could be distinctly clustered based on intensity and duration, and each cluster of events would produce different stormwater quality responses. The high-intensity bursts occurring in the latter part of long-duration events were found to produce uniform and low concentrations of suspended solids. One the contrary, high intensity bursts occurring in the initial part of short-duration events triggered the first-flush effect, thus producing high concentrations of suspended solids. Furthermore, the first-flush effect was likely to present when the high intensity bursts occurred in the mid portion of rainfall events and produced variable concentrations of suspended solids. It was also found that the average rainfall intensity plays a key role in mobilising and transporting pollutants accumulated on urban surfaces.

New conceptualisation of first flush phenomena in urban catchments.

First flush is an important phenomenon commonlyused in stormwater treatment system design where only the highly concentrated initial part of the runoff hydrograph is subject to treatment. Despite the existing methods for estimating the first flush, a robust quantitative definition is difficult to find. This paper discusses a novel approach, where a new parameter is introduced to analyse the variability in the discharge of pollutants at different times throughout a runoff event and thereby enable the identification of first flush. It was found that due to variability in rainfall, the first flush runoff volume varies from event to event. Therefore, a static estimate of the first flush is not applicable for a runoff event. The Monte Carlo simulation undertaken strengthened the analysis by providing credible limits to the outcomes. Accordingly, an interval estimation was obtained in which the first flush runoff can vary, and it was found that most commonly, the first flush can exist through the initial 30%-50% of the runoff. Therefore, in order to treat the stormwater runoff with minimum risk of discharging high loads of pollutants to the receiving water environment, at least the initial 30% of the runoff should be subject to treatment. This understanding provides a fundamental basis for the design of robust stormwater treatment systems.

Catchment based estimation of pollutant event mean concentration (EMC) and implications for first flush assessment.

The Event Mean Concentration (EMC) is considered as a key analytical parameter for assessing the quality of stormwater. The conventional estimation methods to determine EMC do not necessarily address the variability associated with the hydrologic characteristics. Accordingly, this study was conducted to identify the potential hydrologic variables that can influence EMC and thereby to create a mathematical model to determine EMC using the hydrologic variables while incorporating the catchment as an influential factor. This paper introduces an innovative approach to estimate EMC of a runoff event using a stepwise multiple linear regression model. The model incorporates hydrologic variables together with their two-way interaction terms. The catchment was included in the model as a dummy variable. This allows identifying the variability of EMC between catchments. Model can reasonably predict the EMC with an overall prediction error of 0.811. The regression coefficients of the model specify that, maximum rainfall intensity is the most influential variable having a coefficient of 1.008, followed by the average intensity with a coefficient -0.586. The interaction term of rainfall depth and the antecedent dry period indicates that for a relatively small rainfall event (<5 mm), an optimum value of antecedent dry period exists that maximises the EMC. Subsequently, EMC was employed to define the first flush runoff as an alternative approach to the conventional approaches for determining the first flush. The dynamic mean concentration (DMCt), was introduced as a parameter for estimating the first flush using EMC. The maximum accumulated runoff volume such that, DMCt≥EMC was defined as the first flush runoff. It was found that residential catchments generate more intense first flush compared to catchments with totally impervious surface areas and thereby a significant pollutant load is transported within a small initial fraction of the runoff.f.

Bayesian adaptive N-of-1 trials for estimating population and individual treatment effects.

This article proposes a novel adaptive design algorithm that can be used to find optimal treatment allocations in N-of-1 clinical trials. This new methodology uses two Laplace approximations to provide a computationally efficient estimate of population and individual random effects within a repeated measures, adaptive design framework. Given the efficiency of this approach, it is also adopted for treatment selection to target the collection of data for the precise estimation of treatment effects. To evaluate this approach, we consider both a simulated and motivating N-of-1 clinical trial from the literature. For each trial, our methods were compared with the multiarmed bandit approach and a randomized N-of-1 trial design in terms of identifying the best treatment for each patient and the information gained about the model parameters. The results show that our new approach selects designs that are highly efficient in achieving each of these objectives. As such, we propose our Laplace-based algorithm as an efficient approach for designing adaptive N-of-1 trials.

Relative rate of change in cognitive score network dynamics via Bayesian hierarchical models reveal spatial patterns of neurodegeneration.

The degeneration of the human brain is a complex process, which often affects certain brain regions due to healthy aging or disease. This degeneration can be evaluated on regions of interest (ROI) in the brain through probabilistic networks and morphological estimates. Current approaches for finding such networks are limited to analyses at discrete neuropsychological stages, which cannot appropriately account for connectivity dynamics over the onset of cognitive deterioration, and morphological changes are seldom unified with connectivity networks, despite known dependencies. To overcome these limitations, a probabilistic wombling model is proposed to simultaneously estimate ROI cortical thickness and covariance networks contingent on rates of change in cognitive decline. This proposed model was applied to analyze longitudinal data from healthy control (HC) and Alzheimer's disease (AD) groups and found connection differences pertaining to regions, which play a crucial role in lasting cognitive impairment, such as the entorhinal area and temporal regions. Moreover, HC cortical thickness estimates were significantly higher than those in the AD group across all ROIs. The analyses presented in this work will help practitioners jointly analyze brain tissue atrophy at the ROI-level conditional on neuropsychological networks, which could potentially allow for more targeted therapeutic interventions.

Principles of Experimental Design for Big Data Analysis.

Big Datasets are endemic, but are often notoriously difficult to analyse because of their size, heterogeneity and quality. The purpose of this paper is to open a discourse on the potential for modern decision theoretic optimal experimental design methods, which by their very nature have traditionally been applied prospectively, to improve the analysis of Big Data through retrospective designed sampling in order to answer particular questions of interest. By appealing to a range of examples, it is suggested that this perspective on Big Data modelling and analysis has the potential for wide generality and advantageous inferential and computational properties. We highlight current hurdles and open research questions surrounding efficient computational optimisation in using retrospective designs, and in part this paper is a call to the optimisation and experimental design communities to work together in the field of Big Data analysis.

Catchment scale assessment of risk posed by traffic generated heavy metals and polycyclic aromatic hydrocarbons.

Heavy metals (HMs) and polycyclic aromatic hydrocarbons (PAHs) are among the most toxic chemical pollutants present in urban stormwater. Consequently, urban stormwater reuse is constrained due to the human health risk posed by these pollutants. This study developed a scientifically robust approach to assess the risk to human health posed by HMs and PAHs in urban stormwater in order to enhance its reuse. Accordingly, an innovative methodology was created consisting of four stages: quantification of traffic and land use parameters; estimation of pollutant concentrations for model development; risk assessment, and risk map presentation. This methodology will contribute to catchment scale assessment of the risk associated with urban stormwater and for risk mitigation. The risk map developed provides a simple and efficient approach to identify the critical areas within a large catchment. The study also found that heavy molecular weight PAHs (PAHs with 5-6 benzene rings) in urban stormwater pose higher risk to human health compared to light molecular PAHs (PAHs with 2-4 benzene rings). These outcomes will facilitate the development of practical approaches for applying appropriate mitigation measures for the safe management of urban stormwater pollution and for the identification of enhanced reuse opportunities.

Application of landscape epidemiology to assess potential public health risk due to poor sanitation.

Clear identification of areas vulnerable to waterborne diseases is essential for protecting community health. This is particularly important in developing countries where unsafe disposal of domestic wastewater and limited potable water supply pose potential public health risks. However, data paucity can be a compounding issue. Under these circumstances, landscape epidemiology can be applied as a resource efficient approach for mapping potential disease risk areas associated with poor sanitation. However, in order to realise the full potential offered by this approach, an in-depth understanding of the impact of different classes of an explanatory variable on a target disease and the validity of hotspot analysis using limited datasets is needed. Accordingly, this research study focused on typhoid and diarrhoea incidence with respect to different classes of elevation, flood inundation, land use, soil permeability, population density and rainfall as explanatory variables. An integrated methodology consisting of hot spot analysis and Poisson regression was employed to map potential disease risk areas. The study findings confirmed the significant differences in the influence exerted by the various classes of an explanatory variable in relation to a target disease. The results also confirmed the feasibility of the hotspot analysis for identifying areas vulnerable to the target diseases using a limited dataset. The study outcomes are expected to contribute to creating an in-depth understanding of the relationship between disease prevalence and associated landscape factors for the delineation of disease risk zones in the context of data paucity.

Bayesian adaptive design: improving the effectiveness of monitoring of the Great Barrier Reef.

Monitoring programs are essential for understanding patterns, trends, and threats in ecological and environmental systems. However, such programs are costly in terms of dollars, human resources, and technology, and complex in terms of balancing short- and long-term requirements. In this work, We develop new statistical methods for implementing cost-effective adaptive sampling and monitoring schemes for coral reef that can better utilize existing information and resources, and which can incorporate available prior information. Our research was motivated by developing efficient monitoring practices for Australia's Great Barrier Reef. We develop and implement two types of adaptive sampling schemes, static and sequential, and show that they can be more informative and cost-effective than an existing (nonadaptive) monitoring program. Our methods are developed in a Bayesian framework with a range of utility functions relevant to environmental monitoring. Our results demonstrate the considerable potential for adaptive design to support improved management outcomes in comparison to set-and-forget styles of surveillance monitoring.

Resistive Index Variability in Anterior Cerebral Artery Measurements During Daily Transcranial Duplex Sonography: A Predictor of Cerebrovascular Complications in Infants Undergoing Extracorporeal Membrane Oxygenation?

OBJECTIVES: The purpose of this study was to determine the value of resistive index (RI) variability in predicting cerebrovascular complications during extracorporeal membrane oxygenation (ECMO). METHODS: This retrospective study included 36 infants treated by ECMO. The RI was measured on daily transfontanellar duplex sonography, obtained first without fontanel compression and then after gentle compression with the transducer. The age at ECMO cannulation, sex, gestational age at birth, method of delivery, indication, and type and duration of ECMO were recorded. RESULTS: There was a statistically significant difference in RI variability in infants who developed cerebrovascular complications as opposed to those who did not (P = .002). Resistive index variability of 10% or greater on any day was associated with an increased risk for cerebrovascular complications (P = .0482; χ2 = 3.9). Variability in the first 5 days was significantly higher than on following days (P < .0001). The age at ECMO cannulation showed a significant difference, with mean ± SD values of 1.1 ± 0.9 days in the complications group and 2.7 ± 2.2 days in the no-complications group (P = .043). CONCLUSIONS: Resistive index variability of 10% or greater on any day had a statistically significant risk of cerebrovascular complication development. Extracorporeal membrane oxygenation cannulation at younger than 3 days conferred an increased risk of cerebrovascular complications.

Geographic variation in the intended choice of adjuvant treatments for women diagnosed with screen-detected breast cancer in Queensland.

BACKGROUND: Although early diagnosis and improved treatment can reduce breast cancer mortality, there still appears to be a geographic differential in patient outcomes. This study aims to determine and quantify spatial inequalities in intended adjuvant (radio-, chemo- and hormonal) therapy usage among women with screen-detected breast cancer in Queensland, Australia. METHODS: Linked population-based datasets from BreastScreen Queensland and the Queensland Cancer Registry during 1997-2008 for women aged 40-89 years were used. We adopted a Bayesian shared spatial component model to evaluate the relative intended use of each adjuvant therapy across 478 areas as well as common spatial patterns between treatments. RESULTS: Women living closer to a cancer treatment facility were more likely to intend to use adjuvant therapy. This was particularly marked for radiotherapy when travel time to the closest radiation facility was 4 + h (OR =0.41, 95 % CrI: [0.23, 0.74]) compared to <1 h. The shared spatial effect increased towards the centres with concentrations of radiotherapy facilities, in north-east (Townsville) and south-east (Brisbane) regions of Queensland. Moreover, the presence of residual shared spatial effects indicates that there are other unmeasured geographical barriers influencing women's treatment choices. CONCLUSIONS: This highlights the need to identify the additional barriers that impact on treatment intentions among women diagnosed with screen-detected breast cancer, particularly for those women living further away from cancer treatment centers.

Fetal alcohol spectrum disorder and attention deficit hyperactivity disorder stimulant trial in children: an N-of-1 pilot trial to compare stimulant to placebo (FASST): protocol.

INTRODUCTION: Fetal alcohol spectrum disorder (FASD) is a neurodevelopmental disorder caused by alcohol exposure during pregnancy. FASD is associated with neurodevelopmental deviations, and 50%-94% of children with FASD meet the Diagnostic and Statistical Manual of Mental Disorders-fifth edition diagnostic criteria for attention deficit hyperactivity disorder (ADHD). There is a paucity of evidence around medication efficacy for ADHD symptoms in children with FASD. This series of N-of-1 trials aims to provide pilot data on the feasibility of conducting N-of-1 trials in children with FASD and ADHD. METHODS AND ANALYSIS: A pilot N-of-1 randomised trial design with 20 cycles of stimulant and placebo (four cycles of 2-week duration) for each child will be conducted (n=20) in Melbourne, Australia.Feasibility and tolerability will be assessed using recruitment and retention rates, protocol adherence, adverse events and parent ratings of side effects. Each child's treatment effect will be determined by analysing teacher ADHD ratings across stimulant and placebo conditions (Wilcoxon rank). N-of-1 data will be aggregated to provide an estimate of the cohort treatment effect as well as individual-level treatment effects. We will assess the sample size and number of cycles required for a future trial. Potential mediating factors will be explored to identify variables that might be associated with treatment response variability. ETHICS AND DISSEMINATION: The study was approved by the Hospital and Health Service Human Research Ethics Committee (HREC/74678/MonH-2021-269029), Monash (protocol V6, 25 June 2023).Individual outcome data will be summarised and provided to participating carers and practitioners to enhance care. Group-level findings will be presented at a local workshop to engage stakeholders. Findings will be presented at national and international conferences and published in peer-reviewed journals. All results will be reported so that they can be used to inform prior information for future trials. TRIAL REGISTRATION NUMBER: NCT04968522.

Incorporating uncertainty in Indigenous sea Country monitoring with Bayesian statistics: Towards more informed decision-making.

Partnerships in marine monitoring combining Traditional Ecological Knowledge and western science are developing globally to improve our understanding of temporal changes in ecological communities that better inform coastal management practices. A fuller communication between scientists and Indigenous partners about the limitations of monitoring results to identify change is essential to the impact of monitoring datasets on decision-making. Here we present a 5-year co-developed case study from a fish monitoring partnership in northwest Australia showing how uncertainty estimated by Bayesian models can be incorporated into monitoring management indicators. Our simulation approach revealed there was high uncertainty in detecting immediate change over the following monitoring year when translated to health performance indicators. Incorporating credibility estimates into health assessments added substantial information to monitoring trends, provided a deeper understanding of monitoring limitations and highlighted the importance of carefully selecting the way we evaluate management performance indicators.

Does the registry speak your language? A case study of the Global Angelman Syndrome Registry.

Global disease registries are critical to capturing common patient related information on rare illnesses, allowing patients and their families to provide information about their condition in a safe, accessible, and engaging manner that enables researchers to undertake critical research aimed at improving outcomes. Typically, English is the default language of choice for these global digital health platforms. Unfortunately, language barriers can significantly inhibit participation from non-English speaking participants. In addition, there is potential for compromises in data quality and completeness. In contrast, multinational commercial entities provide access to their websites in the local language of the country they are operating in, and often provide multiple options reflecting ethnic diversity. This paper presents a case study of how the Global Angelman Syndrome Registry (GASR) has used a novel approach to enable multiple language translations for its website. Using a "semi-automated language translation" approach, the GASR, which was originally launched in English in September 2016, is now available in several other languages. In 2020, the GASR adopted a novel approach using crowd-sourcing and machine translation tools leading to the availability of the GASR in Spanish, Traditional Chinese, Italian, and Hindi. As a result, enrolments increased by 124% percent for Spain, 67% percent for Latin America, 46% percent for Asia, 24% for Italy, and 43% for India. We describe our approach here, which we believe presents an opportunity for cost-effective and timely translations responsive to changes to the registry and helps build and maintain engagement with global disease communities.

Comparison of robust criteria for D-optimal designs.

This study compared the performance of a local and three robust optimality criteria in terms of the standard error for a one-parameter and a two-parameter nonlinear model with uncertainty in the parameter values. The designs were also compared in conditions where there was misspecification in the prior parameter distribution. The impact of different correlation between parameters on the optimal design was examined in the two-parameter model. The designs and standard errors were solved analytically whenever possible and numerically otherwise.

A general method to determine sampling windows for nonlinear mixed effects models with an application to population pharmacokinetic studies.

Optimal design methods have been proposed to determine the best sampling times when sparse blood sampling is required in clinical pharmacokinetic studies. However, the optimal blood sampling time points may not be feasible in clinical practice. Sampling windows, a time interval for blood sample collection, have been proposed to provide flexibility in blood sampling times while preserving efficient parameter estimation. Because of the complexity of the population pharmacokinetic models, which are generally nonlinear mixed effects models, there is no analytical solution available to determine sampling windows. We propose a method for determination of sampling windows based on MCMC sampling techniques. The proposed method attains a stationary distribution rapidly and provides time-sensitive windows around the optimal design points. The proposed method is applicable to determine sampling windows for any nonlinear mixed effects model although our work focuses on an application to population pharmacokinetic models.

Temperature Sensing and Honey Bee Colony Strength.

Strength auditing of European honey bee (Apis mellifera Linnaeus, 1758 [Hymenoptera: Apidae]) colonies is critical for apiarists to manage colony health and meet pollination contracts conditions. Colony strength assessments used during pollination servicing in Australia typically use a frame-top cluster-count (Number of Frames) inspection. Sensing technology has potential to improve auditing processes, and commercial temperature sensors are widely available. We evaluate the use and placement of temperature sensing technology in colony strength assessment and identify key parameters linking temperature to colony strength. Custom-built temperature sensors measured hive temperature across the top of hive brood boxes. A linear mixed-effect model including harmonic sine and cosine curves representing diurnal temperature fluctuations in hives was used to compare Number of Frames with temperature sensor data. There was a significant effect of presence of bees on hive temperature and range: hives without bees recorded a 5.5°C lower mean temperature and greater temperature ranges than hives containing live bees. Hives without bees reach peak temperature earlier than hives with bees, regardless of colony strength. Sensor placement across the width of the hive was identified as an important factor when linking sensor data with colony strength. Data from sensors nearest to the hive geometric center were found to be more closely linked to colony strength. Furthermore, a one unit increase in Number of Frames was significantly associated with a mean temperature increase of 0.36°C. This demonstrates that statistical models that account for diurnal temperature patterns could be used to predict colony strength from temperature sensor data.