Hierarchical Bayesian Methods Research Articles

Flexible and fast estimation of binary merger population distributions with an adaptive kernel density estimator

The LIGO Scientific, Virgo and KAGRA Collaborations recently released the third gravitational wave transient catalog or GWTC-3, significantly expanding the number of gravitational wave signals. To address the---still uncertain---formation channels of the source compact binaries, their population properties must be characterized. The computational cost of the Bayesian hierarchical methods employed thus far scales with the size of the event catalogs, and such methods have until recently assumed fixed functional forms for the source distribution. Here we propose a fast and flexible method to reconstruct the population of LIGO-Virgo merging black hole (BH) binaries without such assumptions. For sufficiently high event statistics and sufficiently low individual event measurement error (relative to the scale of population features) a kernel density estimator (KDE) reconstruction of the event distribution will be accurate. We improve the accuracy and flexibility of KDE for finite event statistics using an adaptive bandwidth KDE (awKDE). We apply awKDE to publicly released parameter estimates for 44 significant (69) BH binary mergers in GWTC-2 (GWTC-3), in combination with a fast polynomial fit of search sensitivity, to obtain a nonparametric estimate of the mass distribution, and compare to Bayesian hierarchical methods. We also demonstrate a robust peak detection algorithm based on awKDE and use it to calculate the significance of the apparent peak in the BH mass distribution around $35\text{ }\text{ }{\mathrm{M}}_{\ensuremath{\bigodot}}$. We find such a peak is very unlikely to have occurred if the true distribution is a featureless power-law (significance of $3.6\ensuremath{\sigma}$ for confident GWTC-2 BBH events, $3.0\ensuremath{\sigma}$ for confident GWTC-3 BBH events).

Behavioral welfare economics and risk preferences: a Bayesian approach

We propose the use of Bayesian estimation of risk preferences of individuals for applications of behavioral welfare economics to evaluate observed choices that involve risk. Bayesian estimation provides more systematic control of the use of informative priors over inferences about risk preferences for each individual in a sample. We demonstrate that these methods make a difference to the rigorous normative evaluation of decisions in a case study of insurance purchases. We also show that hierarchical Bayesian methods can be used to infer welfare reliably and efficiently even with significantly reduced demands on the number of choices that each subject has to make. Finally, we illustrate the natural use of Bayesian methods in the adaptive evaluation of welfare.

Hierarchical Bayesian classification methods to identify topics by journal quartile with an application in biological sciences

This manuscript builds on a novel, automatic, freely-available Bayesian approach to extract information in abstracts and titles to classify research topics by quartile. This approach is demonstrated for all N= 149,129 ISI-indexed publications in biological sciences journals during 2017. A Bayesian multinomial inverse regression approach is used to extract rankings of topics without the need of a pre-defined dictionary. Bigrams are used for extraction of research topics across manuscripts, and rankings of research topics are constructed by quartile. Worldwide and local results (e.g., comparison between two peer/aspirational research institutions in Colombia) are provided, and differences are explored both at the global and local levels. Some topics persist across quartiles, while the relevance of others is quartile-specific. Challenges in sustainable development appear as more prevalent in top quartile journals across institutions, while the two Colombian institutions favour plant and microorganism research. This approach can reduce information inequities, by allowing young/incipient researchers in biological sciences, especially within lower income countries or universities with limited resources, to freely assess the state of the literature and the relative likelihood of publication in higher impact journals by research topic. This can also serve institutions of higher education to identify missing research topics and areas of competitive advantage.

Modelling approaches for histology-independent cancer drugs to inform NICE appraisals: a systematic review and decision-framework.

The first histology-independent marketing authorisation in Europe was granted in 2019. This was the first time that a cancer treatment was approved based on a common biomarker rather than the location in the body at which the tumour originated. This research aims to explore the implications for National Institute for Health and Care Excellence appraisals. Targeted reviews were undertaken to determine the type of evidence that is likely to be available at the point of marketing authorisation and the analyses required to support National Institute for Health and Care Excellence appraisals. Several challenges were identified concerning the design and conduct of trials for histology-independent products, the greater levels of heterogeneity within the licensed population and the use of surrogate end points. We identified approaches to address these challenges by reviewing key statistical literature that focuses on the design and analysis of histology-independent trials and by undertaking a systematic review to evaluate the use of response end points as surrogate outcomes for survival end points. We developed a decision framework to help to inform approval and research policies for histology-independent products. The framework explored the uncertainties and risks associated with different approval policies, including the role of further data collection, pricing schemes and stratified decision-making. We found that the potential for heterogeneity in treatment effects, across tumour types or other characteristics, is likely to be a central issue for National Institute for Health and Care Excellence appraisals. Bayesian hierarchical methods may serve as a useful vehicle to assess the level of heterogeneity across tumours and to estimate the pooled treatment effects for each tumour, which can inform whether or not the assumption of homogeneity is reasonable. Our review suggests that response end points may not be reliable surrogates for survival end points. However, a surrogate-based modelling approach, which captures all relevant uncertainty, may be preferable to the use of immature survival data. Several additional sources of heterogeneity were identified as presenting potential challenges to National Institute for Health and Care Excellence appraisal, including the cost of testing, baseline risk, quality of life and routine management costs. We concluded that a range of alternative approaches will be required to address different sources of heterogeneity to support National Institute for Health and Care Excellence appraisals. An exemplar case study was developed to illustrate the nature of the assessments that may be required. Adequately designed and analysed basket studies that assess the homogeneity of outcomes and allow borrowing of information across baskets, where appropriate, are recommended. Where there is evidence of heterogeneity in treatment effects and estimates of cost-effectiveness, consideration should be given to optimised recommendations. Routine presentation of the scale of the consequences of heterogeneity and decision uncertainty may provide an important additional approach to the assessments specified in the current National Institute for Health and Care Excellence methods guide. Further exploration of Bayesian hierarchical methods could help to inform decision-makers on whether or not there is sufficient evidence of homogeneity to support pooled analyses. Further research is also required to determine the appropriate basis for apportioning genomic testing costs where there are multiple targets and to address the challenges of uncontrolled Phase II studies, including the role and use of surrogate end points. This project was funded by the National Institute for Health Research (NIHR) Evidence Synthesis programme and will be published in full in Health Technology Assessment; Vol. 25, No. 76. See the NIHR Journals Library website for further project information.

Partial Bayes Estimation of Two Parameter Gamma Distribution Under Non-Informative Prior

In Bayesian analysis, empirical and hierarchical methods are two main approaches for the estimation of the parameter(s) involved in the prior distribution of one parameter. But in the multi-parameter model, e.g., Gamma(α, p), where both the parameters are unknown, idea of the ‘Partial Bayes (PB) Estimation’ is introduced. When we do no have proper belief regarding the joint parameters of the distribution of the variable and when we are estimating one parameter in presence of others, such method may be used. Partial Bayes estimation of the scale parameter p is done by putting the estimate of the another parameter α obtained by some other classical method in case of two parameter Gamma distribution. Using non-informative prior and computing the risk, it is found that the Partial Bayes estimator has less risk than the Bayes estimator. For this, simulation studies for some choices of shape parameter values have been done. In case of the shape parameter, posterior mean and posterior variance are evaluated through simulations to obtain the risk values for estimator of α with known scale parameter. Finally after fifitting this distribution, two real datasets are illustrated to see the performance of the Partial Bayes estimator.

Abstract 13715: Efficacy of Implantable Cardioverter-Defibrillators/Cardiac Resynchronization Therapy Defibrillator Remote Monitoring in Patients With Heart Failure: A Bayesian Meta-Analysis of Randomized Controlled Trials

Introduction: Remote monitoring using implantable cardioverter-defibrillators (ICDs) or cardiac resynchronization therapy defibrillator (CRT-D) can be used as an adjunctive diagnostic tool to traditional hospital and clinic visits to monitor the volume status of patients with heart failure (HF). Methods: We searched PubMed, Cochrane Library, and ClinicalTrials.gov (inception through February 4th, 2021) for randomized controlled trials (RCTs) evaluating ICDs/CRT-D remote monitoring versus standard of care in patients with HF. Of 642 articles retrieved after removing duplicate items, we included 11 RCTs encompassing 5,965 patients. Absolute risk difference (ARD) with 95% credible interval (CrI) was estimated. Pooled (posterior) risk difference was computed using Bayesian hierarchical methods, random effects model and weakly informative prior. Results: The ARD for mortality was centered at -0.01 (95%CrI: -0.03; 0.01, Tau: 0.02), with 82% probability of ARD of ICD/CRT-D remote monitoring with respect to control being less than 0. The ARD for cardiovascular mortality was centered at -0.03 (95%CrI: -0.11; 0.05, Tau: 0.10), with 84% probability of ARD of ICD/CRT-D remote monitoring with respect to control being less than 0. The ARD for HF admissions was centered at -0.00 (95%CrI: -0.07; 0.06, Tau: 0.08), with 57% probability of ARD of ICD/CRT-D remote monitoring with respect to control being less than 0 [Figure]. Conclusions: This meta-analysis suggests that ICD/CRT-D remote monitoring in patients with HF is associated with a higher probability of reduced all-cause mortality, and cardiovascular mortality compared with standard care. However, ICD-CRT/D remote monitoring was associated with minimal probability of difference in HF hospitalizations.

1074The role of epidemiology in improving the health of people with disability

Abstract Focus and outcomes for participants The symposium will focus on the role of epidemiologists in building an evidence base to improve the health of the 15% of the world’s population with disability who currently experience vast health inequalities. Participants will be introduced to new ways of conceptualising disability in epidemiology; state of the art approaches to monitoring disability-related socio-economic and health inequalities; methodological challenges and solutions to address the biases due to misclassification, confounding and reverse causation; and the application of causal mediation analysis and natural experiments in identifying potential policy solutions. Participants will gain a greater understanding of how epidemiological methods can be applied to improve the health of people with disability, as well as insights and ideas for their research. A network of epidemiologists interested in this topic will be generated to foster ongoing communication and collaborative opportunities. Rationale for the symposium, including for its inclusion in the Congress The health of disabled people has largely been ignored by epidemiologists. This is despite emerging evidence that people with disability experience poorer health because of factors unrelated to their impairment, including socio-economic disadvantage, discrimination, and violence. However, turning epidemiologists’ efforts to the health of people with disability presents conceptual and methodological challenges, some of which are unique to the content area. Participants will be shown a suite of approaches that can be deployed to address these problems. Participatory methods and innovative graphical and statistical methods for analysing disability-related health inequalities, approaches rarely used in epidemiology, will be covered. The symposium will also concentrate on the application of methods to optimise causal inference in the presence of multiple potential biases, and methods that simulate randomised controlled trial conditions to model policy interventions. Presentation program The presentations are from researchers from the CRE-DH, funded through Australia’s National Health and Medical Research Council organised four themes. Theme 1: Conceptualisation of disability We will present findings from a scoping review of original articles in epidemiology journals and will argue that, while, disability is usually conceptualised in epidemiology as an outcome, reconceiving of disability as an exposure, mediator and/or effect modifier can provide important insights on the determinants of health of people with disability. Theme 2: Monitoring disability-related inequalities We will demonstrate how the CRE-DH has used participatory methods, where people with disability are ‘experts through lived experience’, to develop indicators to monitor disability-related inequalities and design a National Community Attitudes survey. We will demonstrate innovative ways to graphically illustrate prevalence, absolute and relative inequalities simultaneously, and discuss how hierarchical Bayesian methods can be used to overcome inadequate power due to disaggregation and assess inequalities under uncertainty. Theme 3: Approaches to minimising bias We will talk about how biases can affect estimates of disability prevalence and disability-outcome associations, including reverse causation, confounding and misclassification. We will discuss a range of approaches we have used to address these challenges including modelling incident (rather than prevalent) disability, using fixed effects models and propensity score approaches, and approaches to addressing misclassification bias drawing on examples from our program of research. Theme 4: Identification of policy interventions We will discuss methods that can be used to model the impact of policies on the health of people with disability using examples from our research. We will present the results of a causal mediation analysis modelling the impact of different employment policy interventions on mental health outcomes. We will illustrate the value of natural policy experiments for estimating effects of policy changes on employment and health of people with disability using two examples – the 2014 reassessment of Disability Support Pensioners under stricter impairment tables and the introduction of Australia’s National Disability Insurance Scheme. The symposium will conclude with a facilitated discussion focussed on how epidemiologists can come together internationally to grasp the opportunities and address the challenges in research focussed on the health of people with disabilities. Names of presenters Professor Anne Kavanagh, PhD Dr Nicola Fortune, PhD Dr George Disney, PhD Dr Zoe Aitken Dr Samia Badji, PhD

Vitamin D status in Mainland of China: A systematic review and meta-analysis.

BackgroundLow vitamin D (VitD) status is becoming a global health issue. Previous heterogenous results are urging a meta-analysis to delineate a panorama of VitD conditions in the general population in Mainland of China.MethodsWe performed a systematic review and meta-analysis by searching PubMed, Web of Science, EMBASE, China National Knowledge Infrastructure, WanFang, and VIP databases up to June 4, 2021. The inclusion criteria were as follows: (1) original articles or dissertations focused on VitD status of people in Mainland of China; and (2) studies were population-based, cross-sectional, or longitudinal cohort with baseline data. The outcomes were serum 25(OH)D concentration and the prevalence of low VitD status. Low VitD status included VitD deficiency (< 30 nmol/L) and VitD inadequacy (< 50 nmol/L). Data were estimated by Hierarchical Bayesian methods. All included studies were cross-sectional or longitudinal cohort studies about VitD status of people in Mainland of China. (Registration: PROSPERO CRD42021226130).FindingsA total of 105 eligible studies including 234,519 subjects were included. In adults, the overall mean 25(OH)D concentration was 44.3 nmol/L (95% Credible Interval [CrI]: 39.8–48.7). The pooled prevalence of VitD deficiency and inadequacy was 20.7% (95% CrI: 11.9–32.9) and 63.2% (95% CrI: 53.5–72.3), respectively. In children and adolescents, the overall mean 25(OH)D concentration was 52.2 nmol/L (95% CrI: 46.7–57.5). The pooled prevalence of VitD deficiency and inadequacy was 23.0% (95% CrI: 8.9–44.3) and 46.8% (95% CrI: 37.2–56.6), respectively. Specially, we identified that the prevalence of VitD inadequacy increased with age in populations with age ≤ 18 years and ≥ 60 years.InterpretationLow VitD status is prominent in general population of Mainland of China, especially for adults.FundingNational Major Scientific and Technological Special Project for “Significant New Drugs Development” (2017ZX09304022).

Modeling Inoculum Availability of Plurivorosphaerella nawae in Persimmon Leaf Litter with Bayesian Beta Regression.

Circular leaf spot (CLS), caused by Plurivorosphaerella nawae, is a serious disease affecting persimmon (Diospyros kaki) that is characterized by necrotic lesions on leaves, defoliation, and fruit drop. Under Mediterranean conditions, P. nawae forms pseudothecia in the leaf litter in winter, and ascospores are released in spring, infecting susceptible leaves. Persimmon growers are advised to apply fungicides for CLS control during the period of inoculum availability, which was previously defined based on ascospore counts under the microscope. A model of inoculum availability of P. nawae was developed and evaluated as an alternative to ascospore counts. Leaf litter samples were collected weekly in L'Alcúdia (Spain) from 2010 to 2015. Leaves were soaked and placed in a wind tunnel, and the released ascospores of P. nawae were counted. Hierarchical Bayesian beta regression methods were used to model the dynamics of ascospore production in the leaf litter. The selected model included accumulated degree-days (ADDs) and ADDs taking into account the vapor pressure deficit (ADDvpd) as fixed effects and year as random effect. This model had a mean absolute error of 0.042 and a root mean square error of 0.062. The beta regression model was evaluated in four orchards from 2010 to 2015. Higher accuracy was obtained at the beginning and the end of the ascospore production period, which are the events of interest to schedule fungicide sprays for CLS control in Spain. This same modeling framework can be extended to other fungal plant pathogens whose inoculum dynamics are expressed as proportion data.[Formula: see text] Copyright © 2021 The Author(s). This is an open access article distributed under the CC BY 4.0 International license.

Modeling changes in probabilistic reinforcement learning during adolescence.

In the real world, many relationships between events are uncertain and probabilistic. Uncertainty is also likely to be a more common feature of daily experience for youth because they have less experience to draw from than adults. Some studies suggest probabilistic learning may be inefficient in youths compared to adults, while others suggest it may be more efficient in youths in mid adolescence. Here we used a probabilistic reinforcement learning task to test how youth age 8-17 (N = 187) and adults age 18-30 (N = 110) learn about stable probabilistic contingencies. Performance increased with age through early-twenties, then stabilized. Using hierarchical Bayesian methods to fit computational reinforcement learning models, we show that all participants’ performance was better explained by models in which negative outcomes had minimal to no impact on learning. The performance increase over age was driven by 1) an increase in learning rate (i.e. decrease in integration time scale); 2) a decrease in noisy/exploratory choices. In mid-adolescence age 13-15, salivary testosterone and learning rate were positively related. We discuss our findings in the context of other studies and hypotheses about adolescent brain development.

Evaluating the Benefits of Bayesian Hierarchical Methods for Analyzing Heterogeneous Environmental Datasets: A Case Study of Marine Organic Carbon Fluxes

Large compilations of heterogeneous environmental observations are increasingly available as public databases, allowing researchers to test hypotheses across datasets. Statistical complexities arise when analyzing compiled data due to unbalanced spatial sampling, variable environmental context, mixed measurement techniques, and other reasons. Hierarchical Bayesian modeling is increasingly used in environmental science to describe these complexities, however few studies explicitly compare the utility of hierarchical Bayesian models to simpler and more commonly applied methods. Here we demonstrate the utility of the hierarchical Bayesian approach with application to a large compiled environmental dataset consisting of 5,741 marine vertical organic carbon flux observations from 407 sampling locations spanning eight biomes across the global ocean. We fit a global scale Bayesian hierarchical model that describes the vertical profile of organic carbon flux with depth. Profile parameters within a particular biome are assumed to share a common deviation from the global mean profile. Individual station-level parameters are then modeled as deviations from the common biome-level profile. The hierarchical approach is shown to have several benefits over simpler and more common data aggregation methods. First, the hierarchical approach avoids statistical complexities introduced due to unbalanced sampling and allows for flexible incorporation of spatial heterogeneitites in model parameters. Second, the hierarchical approach uses the whole dataset simultaneously to fit the model parameters which shares information across datasets and reduces the uncertainty up to 95% in individual profiles. Third, the Bayesian approach incorporates prior scientific information about model parameters; for example, the non-negativity of chemical concentrations or mass-balance, which we apply here. We explicitly quantify each of these properties in turn. We emphasize the generality of the hierarchical Bayesian approach for diverse environmental applications and its increasing feasibility for large datasets due to recent developments in Markov Chain Monte Carlo algorithms and easy-to-use high-level software implementations.

Length-weight relationship of bigeye scad, Selar crumenophthalmus (Bloch, 1793): Regular vs hierarchical bayesian methods

Fisheries management or conservation requires information on length-weight relationship (LWR) for the fishing regulation and biomass estimation. This study aims to assess LWR estimation using two methods, regular and Bayesian hierarchical approached for big-eye Scad (Selar crumenophthalmus). Samples of big eye Scad were collected at several fish landings around Ambon Island from March to August 2020. Length-weight relationship measurement to obtain the parameters of W = a*Lb was tested using generalized linear model and t-test. The parameter b for monthly sampling was not significantly different (F = 0.77, df = 70, P = 0.89) and showed isometric growth b=3 (t = -1.13, df = 4, P = 0.32). Regular measurement resulted parameter log10(a) = -1.99 (±SD = 1.06) dan b = 3.06 (±SD = 0.084). Bayesian method produced parameter log10(a) = -2.07 (±SD = 0.2365) dan parameter b = 3.21 (±SD = 0.1497). Weight measurement from HB approach was significantly higher than the regular method (t = 1.65; df = 405; P &lt;0.0001), and might produce over-estimated of weight from length data. Discrepancy of these methods was overcome by combining all information of LWR to obtain the best estimation on LWR parameters.

Evaluating multiple diagnostic tests: An application to cervical cancer

Systematic Reviews of diagnostic test accuracy (DTA) studies are increasingly comparing the accuracy of multiple tests to facilitate selection of the best performing test(s). Common approaches to compare multiple tests include multiple meta-analyses or meta-regression with the test type as a covariate. Within-study correlation between tests are typically not considered in these approaches. Several DTA network meta-analysis (DTA-NMA) models have been suggested to compare the accuracy of multiple index tests in a single model. Our aim was to identify all DTA-NMA methods for comparing the accuracy of multiple diagnostic tests. We conducted a methodological review of the DTA-NMA models. We searched PubMed, Web of Science, and Scopus from inception until the end of July 2019. Studies of any design published in English were eligible for inclusion. We also reviewed relevant unpublished material. The methods were applied in a network of 37 studies comparing human papillomavirus (HPV) DNA, mRNA, and cytology (ASCUS+/ LSIL+ threshold) for the diagnosis of invasive cervical cancer (CIN2+). We included 10 relevant studies, and identified four Bayesian hierarchical DTA-NMA methods including the 2×2 data table for each index test. Using CIN2+ as a case study, we applied the DTA-NMA methods to determine the most promising test, in terms of sensitivity and specificity. All models showed the mRNA test as the most accurate test followed by HPV DNA: relative sensitivity compared to the cytology test 1.36-1.39 and 1.33-1.35, respectively. However, both tests had similar or worse specificity than cytology (relative specificity range in mRNA 0.96-0.98 and in HPV-DNA 0.94-0.95). Both sensitivity and specificity of mRNA were associated with the highest uncertainty across all models (widest 95% credible intervals 0.68-0.97 and 0.74-0.94, respectively). Precision and estimation of between-study and within-study variability vary across models, which might be due to the differences in the key properties of the models. Different DTA-NMA methods may lead to different results. The choice of a DTA-NMA method for the comparison of multiple diagnostic tests may depend on the available data, e.g., threshold data, as well as on clinically-related factors.

A Monte Carlo method to estimate cell population heterogeneity from cell snapshot data

Variation is characteristic of all living systems. Laboratory techniques such as flow cytometry can probe individual cells, and, after decades of experimentation, it is clear that even members of genetically identical cell populations can exhibit differences. To understand whether variation is biologically meaningful, it is essential to discern its source. Mathematical models of biological systems are tools that can be used to investigate causes of cell-to-cell variation. From mathematical analysis and simulation of these models, biological hypotheses can be posed and investigated, then parameter inference can determine which of these is compatible with experimental data. Data from laboratory experiments often consist of “snapshots” representing distributions of cellular properties at different points in time, rather than individual cell trajectories. These data are not straightforward to fit using hierarchical Bayesian methods, which require the number of cell population clusters to be chosen a priori. Nor are they amenable to standard nonlinear mixed effect methods, since a single observation per cell is typically too few to estimate parameter variability. Here, we introduce a computational sampling method named “Contour Monte Carlo” (CMC) for estimating mathematical model parameters from snapshot distributions, which is straightforward to implement and does not require that cells be assigned to predefined categories. The CMC algorithm fits to snapshot probability distributions rather than raw data, which means its computational burden does not, like existing approaches, increase with the number of cells observed. Our method is appropriate for underdetermined systems, where there are fewer distinct types of observations than parameters to be determined, and where observed variation is mostly due to variability in cellular processes rather than experimental measurement error. This may be the case for many systems due to continued improvements in resolution of laboratory techniques. In this paper, we apply our method to quantify cellular variation for three biological systems of interest and provide Julia code enabling others to use this method.

Nutritional screening tools for adult cancer patients: A hierarchical Bayesian latent-class meta-analysis

Cancer treatment requires attentiveness to its broader effect on the body. Cancer's effect on appetite, strength, and body composition is contained in the summary term malnutrition. The tools used to detect malnutrition are a critical part of effective cancer care. In clinical care, selection of any specific tool is random. The relative validity of these tools have not been systematically compared. Using hierarchical Bayesian latent-class meta-analysis methods, this report compares three tools used for adult cancer patients - the Mini Nutritional Assessment (MNA), the Nutritional Risk Screening 2002 (NRS-2002) and the Patient Generated Subjective Global Assessment (PG-SGA). Drawing from English and Chinese language databases, a broad pool of eligible studies were identified for further selection and assessment. Using the hierarchical summary receiver operating characteristic (HSROC) model, pooled sensitivity, specificity, and other measurements the accuracy of the three tools were compared. A total of 37 eligible studies involving the MNA, NRS-2002 and PG-SGA were included in this meta-analysis. The pooled sensitivity was 0.910 (95% CI: 0.763 to 0.970) for MNA, 0.747 (95% CI: 0.680 to 0.804) for NRS-2002, and 0.964 (95% CI: 0.913 to 0.986) for PG-SGA. The pooled specificity was 0.720 (95% CI: 0.623 to 0.800) for MNA, 0.854 (95% CI: 0.808 to 0.891) for NRS-2002, 0.905 (95% CI: 0.807 to 0.956) for PG-SGA, respectively. The back-calculated likelihood ratio (LR) showed that MNA had a low negative likelihood ratio (LR-), NRS-2002 corresponded to a high positive likelihood ratio (LR+) and PG-SGA represented the best LR+ and LR-. While there is no standard approach to assessment of malnutrition, the PG-SGA has the best diagnostic performance with cancer patients. Further work is needed to refine the utility of these tools in larger clinical samples.

Response time models separate single- and dual-process accounts of memory-based decisions

Human decisions often deviate from economic rationality and are influenced by cognitive biases. One such bias is the memory bias according to which people prefer choice options they have a better memory of—even when the options’ utilities are comparatively low. Although this phenomenon is well supported empirically, its cognitive foundation remains elusive. Here we test two conceivable computational accounts of the memory bias against each other. On the one hand, a single-process account explains the memory bias by assuming a single biased evidence-accumulation process in favor of remembered options. On the contrary, a dual-process account posits that some decisions are driven by a purely memory-driven process and others by a utility-maximizing one. We show that both accounts are indistinguishable based on choices alone as they make similar predictions with respect to the memory bias. However, they make qualitatively different predictions about response times. We tested the qualitative and quantitative predictions of both accounts on behavioral data from a memory-based decision-making task. Our results show that a single-process account provides a better account of the data, both qualitatively and quantitatively. In addition to deepening our understanding of memory-based decision-making, our study provides an example of how to rigorously compare single- versus dual-process models using empirical data and hierarchical Bayesian parameter estimation methods.

Predicting spatiotemporal abundance of breeding waterfowl across Canada: A Bayesian hierarchical modelling approach

AbstractAimOur aim was to develop predictive statistical models for mapping the abundance of 18 waterfowl species at a pan‐Canadian level. We refined the previous generation of national waterfowl models by (a) developing new, more interpretable statistical models that (b) explicitly account for spatiotemporal variations in waterfowl abundance, while (c) testing for associations with an updated suite of habitat covariates.LocationAll of Canada, excluding the Northern Arctic ecozone.MethodsOur response variables were annual species counts on 2,227 aerial‐survey segments over a period of 25 years (1990–2015). Combining machine‐learning and hierarchical regression modelling, we devised an innovative covariate selection strategy to select for each species the best subset of a panel of 232 candidate habitat covariates. With the selected covariates, we implemented hierarchical generalized linear models in a Bayesian framework, using the integrated nested Laplace approximation and stochastic partial differential equation approaches.ResultsOn average, our models explained 47% of the observed variance for spatiotemporal predictions and 74% for temporally averaged spatial predictions. The 18 species models included 94 significant waterfowl‐habitat associations involving 42 distinct habitat covariates, with an average of 5.3 covariates per model. Covariates for forest attributes were the most represented in our models. The proportional biomass of Populus tremuloides was the most frequently selected covariate (10/94 associations in 10/18 species). Model predictions generated spatial and spatiotemporal maps of species abundances over almost all of Canada.Main conclusionsWe showed that it is possible to efficiently combine machine‐learning, variable selection and hierarchical Bayesian methods that exploit high‐dimensional covariate spaces. Our approach yielded powerful and easily interpretable species distribution models with very few covariates, while accounting for residual autocorrelation. Possible applications of the resulting models and maps include the development of biodiversity indicators, the evaluation and execution of conservation planning strategies, and ecosystem services monitoring.

Reconsidering electrophysiological markers of response inhibition in light of trigger failures in the stop-signal task.

This study investigates the neural correlates underpinning response inhibition using a parametric ex-Gaussian model of stop-signal task performance, fit with hierarchical Bayesian methods, in a large healthy sample (N=156). The parametric model accounted for both stop-signal reaction time (SSRT) and trigger failure (i.e., failures to initiate the inhibition process). The returned SSRT estimate (SSRTEXG3 ) was attenuated by ≈65ms compared to traditional nonparametric SSRT estimates (SSRTint ). The amplitude and latency of the N1 and P3 event-related potential components were derived for both stop-success and stop-failure trials and compared to behavioral estimates derived from traditional (SSRTint ) and parametric (SSRTEXG3 , trigger failure) models. Both the fronto-central N1 and P3 peaked earlier and were larger for stop-success than stop-failure trials. For stop-failure trials only, N1 peak latency correlated with both SSRT estimates as well as trigger failure and temporally coincided with SSRTEXG3 , but not SSRTint . In contrast, P3 peak and onset latency were not associated with any behavioral estimates of inhibition for either trial type. While the N1 peaked earlier for stop-success than stop-failure trials, this effect was not found in poor task performers (i.e., high trigger failure/slow SSRT). These findings are consistent with attentional modulation of both the speed and reliability of the inhibition process, but not for poor performers. Together with the absence of any P3 onset latency effect, our findings suggest that attentional mechanisms are important in supporting speeded and reliable inhibition processes required in the stop-signal task.

Statistical inference of Type-I progressively censored step-stress accelerated life test with dependent competing risks

This paper considers a step-stress accelerated dependent competing risks model under progressively Type-I censoring schemes. The dependence structure between competing risks is modeled by a general bivariate function, the cumulative exposure model is assumed and the accelerated model is described by the power rule model. The point and interval estimation of the model parameters and the reliability under normal usage level at mission time are obtained by using the maximum likelihood method and the asymptotic normal theory. We also consider the Bayesian estimators and the highest posterior density credible intervals based on conjugate priors, E-Bayesian, hierarchical Bayesian and empirical Bayesian methods. To illustrate the proposed methodology, the Marshall-Olkin bivariate exponential distribution is used to model the dependence structure between competing risks. A Monte Carlo simulation study and a real data analysis are presented to study the performance of different estimation methods.

Hierarchies improve individual assessment of temporal discounting behavior.

Delay discounting behavior has proven useful in assessing impulsivity across a wide range of populations. As such, accurate estimation of the shape of each individual's temporal discounting profile is paramount when drawing conclusions about how impulsivity relates to clinical and health outcomes such as gambling, addiction, and obesity. Here, we identify an estimation problem with current methods of assessing temporal discounting behavior, and propose a simple solution. First, through a simulation study we identify types of temporal discounting profiles that cannot reliably be estimated. Second, we show how imposing constraints through hierarchical modeling ameliorates these recovery problems. Finally, we apply our solution to a large data set from a temporal discounting task, and illustrate the importance of reliable estimation within patient populations. We conclude with a brief discussion on how hierarchical Bayesian methods can aid in model estimation, compensate for small samples, and improve predictions of externalizing psychopathology.