Bayesian demography 250 years after Bayes

Exploiting the Bayesian approach to derive counts of married women of reproductive age across Cameroon for healthcare planning, 2000-2030

Estimates of married women of reproductive age (MWRA) are needed for policy decisions to enhance reproductive health. Given the unavailability in Cameroon, this study aimed to derive MWRA counts by regions and divisions from 2000 to 2030. Data included 1976, 1987, and 2005 censuses with 606,542 women, five Demographic and Health Surveys from 1991 to 2018 with 48,981 women, and United Nations World Population Prospects from 1976 to 2030. Bayesian models were used in estimating fertility rates, net-migration, and finally, MWRA counts. The total MWRA population in Cameroon was estimated to increase from 2,260,665 (2,198,569-2,352,934) to 6,124,480 (5,862,854-6,482,921), reflecting a 5.7 (5.2-6.2) percentage points (%p) annual rise from 2000-2030. The Centre and Far North regions host the largest numbers, projected to reach 1,264,514 (1,099,373-1,470,021) and 1,069,814 (985,315-1,185,523), respectively, in 2030. The highest divisional-level increases are expected in Mfoundi [14.6%p (11.2-18.8)] and Bénoué [14.9%p (11.1-20.09). This study's findings, showing varied regional- and divisional-level estimates of and trends in MWRA counts should set a baseline for determining the demand for programmes such as family planning, and the scaling of relevant resources sub-nationally.

Bayesian variable selection using Knockoffs with applications to genomics

Given the costliness of HIV drug therapy research, it is important not only to maximize true positive rate (TPR) by identifying which genetic markers are related to drug resistance, but also to minimize false discovery rate (FDR) by reducing the number of incorrect markers unrelated to drug resistance. In this study, we propose a multiple testing procedure that unifies key concepts in computational statistics, namely Model-free Knockoffs, Bayesian variable selection, and the local false discovery rate. We develop an algorithm that utilizes the augmented data-Knockoff matrix and implement Bayesian Lasso. We then identify signals using test statistics based on Markov Chain Monte Carlo outputs and local false discovery rate. We test our proposed methods against non-bayesian methods such as Benjamini-Hochberg (BHq) and Lasso regression in terms TPR and FDR. Using numerical studies, we show the proposed method yields lower FDR compared to BHq and Lasso for certain cases, such as for low and equi-dimensional cases. We also discuss an application to an HIV-1 data set, which aims to be applied analyzing genetic markers linked to drug resistant HIV in the Philippines in future work.

Modelling Fish Growth with Imperfect Data: The Case of Trachurus picturatus

Growth modelling is essential to inform fisheries management but is often hampered by sampling biases and imperfect data. Additional methods such as interpolating data through back-calculation may be used to account for sampling bias but are often complex and time-consuming. Here, we present an approach to improve plausibility in growth estimates when small individuals are under-sampled, based on Bayesian fitting growth models using Markov Chain Monte Carlo (MCMC) with informative priors on growth parameters. Focusing on the blue jack mackerel, Trachurus picturatus, which is an important commercial fish in the southern northeast Atlantic, this Bayesian approach was evaluated in relation to standard growth model fitting methods, using both direct readings and back-calculation data. Matched growth parameter estimates were obtained with the von Bertalanffy growth function applied to back-calculated length at age and the Bayesian fitting, using MCMC to direct age readings, with both outperforming all other methods assessed. These results indicate that Bayesian inference may be a powerful addition in growth modelling using imperfect data and should be considered further in age and growth studies, provided relevant biological information can be gathered and included in the analyses.

Has demography witnessed a data revolution? Promises and pitfalls of a changing data ecosystem

Over the past 25 years, technological improvements that have made the collection, transmission, storage, and analysis of data significantly easier and more cost efficient have ushered in what has been described as the 'big data' era or the 'data revolution'. In the social sciences context, the data revolution has often been characterized in terms of increased volume and variety of data, and much excitement has focused on the growing opportunity to repurpose data that are the by-products of the digitalization of social life for research. However, many features of the data revolution are not new for demographers, who have long used large-scale population data and been accustomed to repurposing imperfect data not originally collected for research. Nevertheless, I argue that demography, too, has been affected by the data revolution, and the data ecosystem for demographic research has been significantly enriched. These developments have occurred across two dimensions. The first involves the augmented granularity, variety, and opportunities for linkage that have bolstered the capabilities of 'old' big population data sources, such as censuses, administrative data, and surveys. The second involves the growing interest in and use of 'new' big data sources, such as 'digital traces' generated through internet and mobile phone use, and related to this, the emergence of 'digital demography'. These developments have enabled new opportunities and offer much promise moving forward, but they also raise important ethical, technical, and conceptual challenges for the field.

Bayesian estimation of pharmacokinetic parameters: an important component to include in the teaching of clinical pharmacokinetics and therapeutic drug monitoring

Bayesian estimation of pharmacokinetic parameters (PKP), as discussed in this review, provides a powerful approach towards the individualization of dosing regimens. The method was first described by Lewis Sheiner and colleagues and it is well suited in clinical environs where few blood fluid measures of drugs are available in the clinic. This makes it a valuable tool in the effective implementation of therapeutic drug monitoring. The principle behind the method is Bayes theorem, which incorporates elements of variability in a priori-known population estimates and variability in the pharmacokinetic parameters, and known errors intrinsic to the assay method used to estimate the blood fluid drug concentrations. This manuscript reviews the Bayesian method. The literature was scanned using Pubmed to provide background into the Bayesian method. An Add-in for Excel program was used to show the ability of the method to estimate PKP using sparse blood fluid concentration vs time data. Using a computer program, the method was able to find reasonable estimates of individual pharmacokinetic parameters, assessed by comparing the estimated data to the true PKP. Education of students in clinical pharmacokinetics is incomplete without some mention and instruction of the Bayesian forecasting method. For a complete understanding, a computer program is needed to demonstrate its utility.

Hierarchical Models for International Comparisons: Smoking, Disability, and Social Inequality in 21 European Countries

BACKGROUND

International comparisons of social inequalities in health outcomes and behaviors are challenging. Due to the level of disaggregation often required, data can be sparse and methods to make adequately powered comparisons are lacking. We aimed to illustrate the value of a hierarchical Bayesian approach that partially pools country-level estimates, reducing the influence of sampling variation and increasing the stability of estimates. We also illustrate a new way of simultaneously displaying the uncertainty of both relative and absolute inequality estimates.

METHODS

We used the 2014 European Social Survey to estimate smoking prevalence, absolute, and relative inequalities for men and women with and without disabilities in 21 European countries. We simultaneously display smoking prevalence for people without disabilities (x-axis), absolute (y-axis), and relative inequalities (contour lines), capturing the uncertainty of these estimates by plotting a 2-D normal approximation of the posterior distribution from the full probability (Bayesian) analysis.

RESULTS

Our study confirms that across Europe smoking prevalence is generally higher for people with disabilities than for those without. Our model shifts more extreme prevalence estimates that are based on fewer observations, toward the European mean.

CONCLUSIONS

We demonstrate the utility of partial pooling to make adequately powered estimates of inequality, allowing estimates from countries with smaller sample sizes to benefit from the increased precision of the European average. Including uncertainty on our inequality plot provides a useful tool for evaluating both the geographical patterns of variation in, and strength of evidence for, differences in social inequalities in health.

Modernising fish and shark growth curves with Bayesian length-at-age models

Growth modelling is a fundamental component of fisheries assessments but is often hindered by poor quality data from biased sampling. Several methods have attempted to account for sample bias in growth analyses. However, in many cases this bias is not overcome, especially when large individuals are under-sampled. In growth models, two key parameters have a direct biological interpretation: L0, which should correspond to length-at-birth and L∞, which should approximate the average length of full-grown individuals. Here, we present an approach of fitting Bayesian growth models using Markov Chain Monte Carlo (MCMC), with informative priors on these parameters to improve the biological plausibility of growth estimates. A generalised framework is provided in an R package 'BayesGrowth', which removes the hurdle of programming an MCMC model for new users. Four case studies representing different sampling scenarios as well as three simulations with different selectivity functions were used to compare this Bayesian framework to standard frequentist growth models. The Bayesian models either outperformed or matched the results of frequentist growth models in all examples, demonstrating the broad benefits offered by this approach. This study highlights the impact that Bayesian models could provide in age and growth studies if applied more routinely rather than being limited to only complex or sophisticated applications.

Challenges of Modeling Outcomes for Surgical Infections: A Word of Caution

Background: We developed a novel analytic tool for colorectal deep organ/space surgical site infections (C-OSI) prediction utilizing both institutional and extra-institutional American College of Surgeons-National Surgical Quality Improvement Program (ACS-NSQIP) data. Methods: Elective colorectal resections (2006-2014) were included. The primary end point was C-OSI rate. A Bayesian-Probit regression model with multiple imputation (BPMI) via Dirichlet process handled missing data. The baseline model for comparison was a multivariable logistic regression model (generalized linear model; GLM) with indicator parameters for missing data and stepwise variable selection. Out-of-sample performance was evaluated with receiver operating characteristic (ROC) analysis of 10-fold cross-validated samples. Results: Among 2,376 resections, C-OSI rate was 4.6% (n = 108). The BPMI model identified (n = 57; 56% sensitivity) of these patients, when set at a threshold leading to 80% specificity (approximately a 20% false alarm rate). The BPMI model produced an area under the curve (AUC) = 0.78 via 10-fold cross- validation demonstrating high predictive accuracy. In contrast, the traditional GLM approach produced an AUC = 0.71 and a corresponding sensitivity of 0.47 at 80% specificity, both of which were statstically significant differences. In addition, when the model was built utilizing extra-institutional data via inclusion of all (non-Mayo Clinic) patients in ACS-NSQIP, C-OSI prediction was less accurate with AUC = 0.74 and sensitivity of 0.47 (i.e., a 19% relative performance decrease) when applied to patients at our institution. Conclusions: Although the statistical methodology associated with the BPMI model provides advantages over conventional handling of missing data, the tool should be built with data specific to the individual institution to optimize performance.

Choosing the choice: Reflections on modelling decisions and behaviour in demographic agent-based models

This paper investigates the issues associated with choosing appropriate models of choice for demographic agent-based models. In particular, we discuss the importance of context, time preference, and dealing with uncertainty in decision modelling, as well as the heterogeneity between agents in their decision-making strategies. The paper concludes by advocating empirically driven, modular, and multi-model approaches to designing simulations of human decision-making, given the lack of an agreed strategy for dealing with any of these issues. Furthermore, we suggest that an iterative process of data collection and simulation experiments, with the latter informing future empirical data collection, should form the basis of such an endeavour. The discussion is illustrated with reference to selected demographic agent-based models, with a focus on migration.

Bayesian population reconstruction of female populations for less developed and more developed countries

We show that Bayesian population reconstruction, a recent method for estimating past populations by age, works for data of widely varying quality. Bayesian reconstruction simultaneously estimates age-specific population counts, fertility rates, mortality rates, and net international migration flows from fragmentary data, while formally accounting for measurement error. As inputs, Bayesian reconstruction uses initial bias-reduced estimates of standard demographic variables. We reconstruct the female populations of three countries: Laos, a country with little vital registration data where population estimation depends largely on surveys; Sri Lanka, a country with some vital registration data; and New Zealand, a country with a highly developed statistical system and good quality vital registration data. In addition, we extend the method to countries without censuses at regular intervals. We also use it to assess the consistency of results between model life tables and available census data, and hence to compare different model life table systems.