Bayesian Hierarchical Model Research Articles

Adverse Events of Factor Xa Inhibitors in Pediatric Patients: A Meta-analysis and Pharmacovigilance Study.

This study aimed to provide a comprehensive review of adverse events (AEs) associated with factor Xa (FXa) inhibitors in pediatric patients. We searched PubMed, Embase, Cochrane Library, ClinicalTrials.gov, and the European Union Clinical Trials Register for English-language records from the establishment of the database up to October 17, 2023. Both randomized controlled trials and single-arm trials were included. AEs were analyzed using a Bayesian hierarchical model. For the pharmacovigilance study, data from the US Food and Drug Administration Adverse Event Reporting System from January 1, 2007, to December 31, 2023, were obtained. The proportional imbalance method and the Medicines and Healthcare products Regulatory Agency method were used to detect AE signals. Further characterization of patients presenting with AEs was performed. Of 451 records identified, 12 eligible studies were included. A total of 50.6% (95% Bayesian credible interval [CrI] 33.1-67.2, τ = 0.796) of patients experienced at least one AE, and 9.9% (95% CrI 3.9-19.5, τ = 0.552) developed at least one serious AE. Major and clinically relevant non-major bleeding occurred in 2.4% (95% CrI 0.8-4.8, τ = 1.61) of patients. The most common bleeding AEs were epistaxis (8.4% [95% CrI 3.9-14.9, τ = 1.96]), subcutaneous hematoma (6.4% [95% CrI 0.5-26.2, τ = 0.54]), and wound hemorrhage (3.7% [95% CrI 0.4-13.3, τ = 0.55]). Non-hemorrhagic AEs were pyrexia (9.2% [95% CrI 4.6-15.3, τ = 1.18]), vomiting (7.8% [95% CrI 4.0-12.3, τ = 0.08]), and abdominal pain (7.4% [95% CrI 1.5-19.4, τ = 0.84]). A total of 39 AE signals were detected in the pharmacovigilance study. The top three highest overall relative odds ratio (ROR) for AEs were observed for haemorrhoidal hemorrhage at 1211.82 (95% CI, 312.69-4696.29), thrombophlebitis at 134.64 (95% CI, 42.18-429.81), and deep vein thrombosis at 68.3 (95% CI, 42.53-109.68). Patients experiencing bleeding AEs had received a mean dosage of rivaroxaban 0.16 mg/kg and apixaban 0.08 mg/kg. Systematically quantified AEs of FXa inhibitors in clinical trials and real-world studies provide an important guide for clinicians. The use of FXa inhibitors in pediatric patients is associated with an acceptable rate of AEs. The most common bleeding AE was epistaxis. Pediatric patients treated with FXa inhibitors were more prone to hemorrhoidal hemorrhage. A safe approach may involve prior use of other anticoagulants followed by careful administration of FXa inhibitors, with a dosing regimen tailored to age and weight. Close monitoring is recommended for peri-procedural anticoagulation and vomiting.

Bayesian Spatio-Temporal Modeling of Hyperlipidemia Risk in IRAN; A Repeated Cross- Sectional Analysis

Introduction: The incidence of hyperlipidemia in Iran is on a consistent rise, potentially contributing significantly to increased susceptibility to cardiovascular diseases and other health complications linked to elevated blood lipid levels. This study employs hierarchical Bayesian model to assess the heightened lipid risk on a broader scale across Iran's provinces. Methods: This study included individuals diagnosed with hyperlipidemia from all provinces of Iran in 2019. The primary focus of the investigation included essential variables such as the mean age, gender distribution, and the documented incidence of hyperlipidemia cases in each province. Population data, stratified by province, age group, and gender, were sourced from the Iranian Statistics Center database. The analysis employed the Besag-York-Mollié (BYM) model, with parameter estimation executed through the Hamiltonian Monte Carlo method. Results: In this investigation, the prevalence and spatial distribution of hyperlipidemia were explored within a diverse population of 1,609,538 patients across various regions in Iran. The relative risk of hyperlipidemia surpassed 1 in 16% of Iranian provinces (Posterior probability [PP] > 0.8), with a calculated 95% Confidence interval (CI) of 0.304 to 0.879. The overall prevalence of hyperlipidemia was determined to be 0.815. Significant heterogeneity in hyperlipidemia was identified among different provinces, with Tehran exhibiting the highest relative risk (RR=1.701; 95% CrI: 1.69, 1.713). Notably, gender (RR=1.008; CI: 1.007, 1.009 for males and RR=1.005;CI: 1.003, 1.007 for females) and age were not found to have a statistically significant effect on the relative risk of the disease. Conclusion: In conclusion, the study sheds light on the spatial dynamics of hyperlipidemia in Iran. 16% of provinces displayed a heightened relative risk, emphasizing the need for targeted public health strategies.

Impact of particulate matter exposure on melanoma risk: A multicentre case-control study.

The relationship between particulate matter (PM) exposure and melanoma risk remains largely unexplored. This study aims to investigate the association between PM10 and PM2.5 long-term exposure and melanoma risk. Case-control study involving 2575 participants, comprising 1473 melanoma patients and 1102 healthy controls attending Departments of Dermatology of University Hospital in North and Central Italy. Demographic data, smoking status, history of sunburns and skin type were collected. PM10 and PM2.5 exposure levels were estimated for each participant's residential address using a Bayesian hierarchical model, providing daily concentrations at a 1 km2 spatial resolution from 2013 to 2021. Logistic regression analyses were performed to evaluate the association between PM exposure and melanoma risk, adjusting for potential confounders. Melanoma patients and controls were 52% males and had a mean age of 63.89 and 61.66 years, respectively. The majority of melanoma patients had Fitzpatrick phototypes 2 (59%) and 3 (36%). There were no significant differences in the geographical distribution of cases and controls based on ZIP codes (p = 0.894). The average melanoma Breslow thickness was 1.01 mm, with 68.15% of cases diagnosed at stage 0 and IA. The multivariate logistic regression revealed a protective effect for higher PM10 (OR = 0.89, 95%CI: 0.86-0.92, p < 0.001) and PM2.5 levels (OR = 0.72, 95%CI: 0.68-0.76, p < 0.001). Darker skin phototypes (Fitzpatrick 4) and cigarette smoking were also associated with a reduced risk of melanoma. Higher levels of PM10 and PM2.5 may have a protective effect against melanoma, potentially due to the reduction in ultraviolet radiation exposure. Further research to understand the complex interactions between environmental factors and melanoma risk are needed.

Frequency-specific changes in prefrontal activity associated with maladaptive belief updating in volatile environments in euthymic bipolar disorder

Bipolar disorder (BD) involves altered reward processing and decision-making, with inconsistencies across studies. Here, we integrated hierarchical Bayesian modelling with magnetoencephalography (MEG) to characterise maladaptive belief updating in this condition. First, we determined if previously reported increased learning rates in BD stem from a heightened expectation of environmental changes. Additionally, we examined if this increased expectation speeds up belief updating in decision-making, associated with modulation of rhythmic neural activity within the prefrontal, orbitofrontal, and anterior cingulate cortex (PFC, OFC, ACC). Twenty-two euthymic BD and 27 healthy control (HC) participants completed a reward-based motor decision-making task in a volatile setting. Hierarchical Bayesian modelling revealed BD participants anticipated greater environmental volatility, resulting in a more stochastic mapping from beliefs to actions and paralleled by lower win rates and a reduced tendency to repeat rewarded actions than HC. Despite this, BD individuals adjusted their expectations of action-outcome contingencies more slowly, but both groups invigorated their actions similarly. On a neural level, while healthy individuals exhibited an alpha-beta suppression and gamma increase during belief updating, BD participants showed dampened effects, extending across the PFC, OFC, and ACC regions. This was accompanied by an abnormally increased beta-band directed information flow in BD. Overall, the results suggest euthymic BD individuals anticipate environmental change without adequately learning from it, contributing to maladaptive belief updating. Alterations in frequency-domain amplitude and functional connectivity within the PFC, OFC, and ACC during belief updating underlie the computational effects and could serve as potential indicators for predicting relapse in future research.

Estimating correlations in low-reliability settings with constrained hierarchical models.

It is popular to study individual differences in cognition with experimental tasks, and the main goal of such approaches is to analyze the pattern of correlations across a battery of tasks and measures. One difficulty is that experimental tasks are often low in reliability as effects are small relative to trial-by-trial variability. Consequently, it remains difficult to accurately estimate correlations. One approach that seems attractive is hierarchical modeling where trial-by-trial variability and variability across conditions, tasks, and individuals are modeled separately. Here we show that hierarchical models may reduce the error in estimating correlations up to 43%, but only if substantive constraint is imposed. The approach here is Bayesian, and we develop novel Bayesian hierarchical factor models for experiments where trials are nested in conditions, tasks, and individuals. The prior on covariances across tasks can either be unconstrained, in which there is little error reduction, or constrained, in which there is substantial error reduction. The constraints are: (1) There is a low-dimension factor structure underlying the covariation across tasks, and (2) all loadings are non-negative leading to a positive manifold on correlations. We argue that both of these assumptions are reasonable in cognitive domains, and that with them, researchers may profitably use hierarchical models to estimate correlations across tasks in low-reliability settings.

Spatial analysis and risk mapping of Crimean-Congo hemorrhagic fever (CCHF) in Sub-saharan Africa

Crimean Congo hemorrhagic fever (CCHF) is a re-emerging tick-borne zoonosis that is caused by CCHF virus (CCHFV). The geographical distribution of the disease and factors that influence its occurrence are poorly known. We analysed historical records on its outbreaks in various countries across the sub-Saharan Africa (SSA) to identify hotspots and determine socioecological and demographicfactors associated with these outbreaks. We used data from historical outbreaks that were reported between 1981 and 2022 in various countries in SSA. To develop a common framework for merging the outbreak data and potential explanatory variables, we generated a common shapefile that combined Level 2 administrative units in all the countries. Several climatic, environmental, socioecological data were obtained from on-line GIS databases and extracted using the shapefile. The data were analysed using an approximate Bayesian hierarchical model using the R-INLA package. The outcome was a Boolean variable which indicated whether an administrative unit in the shapefile was affected in a given year or not. A neighborhood structure was also generated and used to account for spatial autocorrelation in the analysis. The final model that was obtained from the analysis was used to build a CCHF risk map. A total of 54 CCHF outbreaks were compiled across 414 districts in nine SSA countries. Factors that were positively associated with CCHF outbreaks included human population density, land area under grassland, bare soil cover and shrub cover. Conversely, high precipitation during wet months, elevated mean temperature and slope had negative effects. The risk map generated shows that CCHF occurrence risk is higher in arid and semi-arid land (ASAL) of West Africa, the Sahelian region, Central Africa, and the Eastern and Southern Africa region. The analysis identified ecological and demographic factors that are associated with CCHF outbreaks in SSA. This finding suggests the need to improve surveillance for the disease especially in the grasslands where the human population is increasing.

Relationship Between Intraocular Pressure and the True Rate of Functional and Structural Progression in the United Kingdom Glaucoma Treatment Study.

To investigate the effect of average intraocular pressure (IOP) on the true rate of glaucoma progression (RoP) in the United Kingdom Glaucoma Treatment Study (UKGTS). UKGTS participants were randomized to placebo or Latanoprost drops and monitored for up to two years with visual field tests (VF, 24-2 SITA standard), IOP measurements, and optic nerve imaging. We included eyes with at least three structural or functional assessments (VF with <15% false-positive errors). Structural tests measured rim area (RA) with Heidelberg retina tomography (HRT) and average peripapillary retinal nerve fiber layer (pRNFL) thickness with optical coherence tomography (OCT). One eye of 436 patients (222 on Latanoprost) was analyzed. A Bayesian hierarchical model estimated the true RoP of VF and structural metrics, and their correlations, using sign-reversed multivariable exponential distribution. RA and pRNFL measurements were converted to a dB scale, matching the VF metric (mean deviation [MD]). The effect of average IOP on the true RoPs was estimated. True RoP at the mean average IOP (17 mm Hg) was faster (P < 0.001) for VF-MD (-0.59 [-0.73, -0.48] dB/year) than HRT-RA (-0.05 [-0.07, -0.03] dB/year) and OCT-pRNFL (-0.08 [-0.11, -0.06] dB/year). The proportional acceleration of RoP per mm Hg increase was, however, not significantly different (smallest P = 0.15). Accounting for the structural floor-effect largely eliminated the differences in RoPs (smallest P = 0.25). VF appeared to deteriorate at a faster rate than structural measurements. However, this could be explained by the floor-effect from nonfunctional tissue. IOP induced a similar acceleration in RoP per mm Hg increase.

Economic Freedom and Presidential Elections: A Bayesian Spatial Probit Hierarchical Modeling Approach

ABSTRACTThere is a growing consensus that spatial effects can help to explain electoral outcomes. Additionally, the concept of economic freedom is also correlated with many different economic outcomes. This paper combines these two ideas using a Bayesian hierarchical spatial probit econometric model. Hierarchical models have a long history in the Bayesian literature and allow for the estimation of Level 1 (county) and Level 2 (state) variables. In this study, we use county‐level data on county‐level votes for Donald Trump in the 2020 presidential election and control for state‐level economic freedom using a newly developed Bayesian hierarchical spatial probit model. The results indicate that state‐level economic freedom contributed positively to Donald Trump's vote at the county level.

Efficient sparsity-promoting MAP estimation for Bayesian linear inverse problems

Abstract Bayesian hierarchical models can provide efficient algorithms for finding sparse solutions to ill-posed linear inverse problems. The models typically comprise a conditionally Gaussian prior model for the unknown augmented by a generalized gamma hyper-prior model for the variance hyper-parameters. This investigation generalizes such models and their efficient maximum a posterior estimation using the iterative alternating sequential algorithm in two ways: (1) general sparsifying transforms: diverging from conventional methods, our approach permits use of sparsifying transformations with nontrivial kernels; (2) unknown noise variances: the noise variance is treated as a random variable to be estimated during the inference procedure. This is important in applications where the noise estimate cannot be accurately estimated a priori. Remarkably, these augmentations neither significantly burden the computational expense of the algorithm nor compromise its efficacy. We include convexity and convergence analysis and demonstrate our method’s efficacy in several numerical experiments.

Demographic inequities and cumulative environmental burdens within communities near superfund sites on Long Island, New York.

Nassau and Suffolk Counties of Long Island, New York are densely populated and contain 34 federally-designated and 449 state-designated Superfund sites, potentially exposing communities to toxic releases. We conducted a distributive justice analysis assessing proximity to Superfund sites, community socio-demographics, and other environmental burdens. Socio-demographic and environmental variables for 665 census tracts were obtained from the United States Census and Environmental Protection Agency's Environmental Justice Screening and Mapping Tool. Hierarchical Bayesian spatial Poisson regression models evaluated the relationship between socio-demographic and environmental variables and counts of Superfund sites per census tract. Analyses were further stratified by county and site type (Federal versus State). A 10% increase in low-income residents was associated with a 47% increase in Superfund sites (Risk Ratio [RR]: 1.47; 95% credible interval (CI): 1.20-1.81). A 10% increase in Hispanic/Latino residents was associated with a 20% increase (RR: 1.20; 95%CI: 1.02-1.42). Higher PM2.5 concentrations (RR:1.64, 95% CI: 1.09-2.48), higher toxic air releases (RR: 1.27, 95%CI: 1.03-1.61), and greater proximity to underground gas storage tanks (RR: 1.27, 95%CI: 1.09-1.48) were associated with increases in Superfund counts. Stratified analyses revealed that low-income residents are concentrated near state not federal Superfund sites. County stratification found that only Suffolk County residents near Superfund sites have increased lead exposure potential, and Black residents in Suffolk (not Nassau) were more likely to live near Superfund sites. We observed localized distributive inequities in community demographics near Superfund sites on Long Island, and communities near Superfund sites are more likely to experience other environmental burdens.

Historical Redlining and Community-Reported Housing Quality: A Spatial Analysis.

Historical redlining, a racially discriminatory practice implemented by the USgovernment in the 1930s, has been associated with present-day environmental outcomes. However, there is limited research examining the relationship between historical redlining and contemporary housing quality. The objective of the present study was to investigate the relationship between historical redlining and contemporary housing quality in Atlanta, Georgia. Spatial patterns of housing code violation complaints from 2015 to 2019 were examined using point-pattern and spatial cluster analyses. We used Bayesian hierarchical models, accounting for spatial autocorrelation, to estimate associations between historical redlining and housing complaints, after adjusting for contemporary neighborhood characteristics, such as poverty, median structure age, vacant and renter-occupied properties, and residential racial segregation. A total of 48,626 housing code violation complaints were reported during the study period, including 6531 complaints deemed "hazardous." Historical redlining was a statistically significant predictor of housing complaints. We observed a 167% increased risk (IRR = 2.67, 95% confidence interval = 1.49, 4.77) of housing complaints for historically redlined neighborhoods compared to neighborhoods historically graded as "best" or "still desirable," after adjusting for neighborhood characteristics. Redlined neighborhoods also had an increased risk of "hazardous" housing complaints (IRR = 1.94, 95% confidence interval = 1.11, 3.40), after adjusting for contemporary neighborhood characteristics. Historically redlined neighborhoods exhibited disproportionately higher rates of housing code violation complaints. Spatial analysis of housing code violation complaints can provide insights into housing quality and inform interventions targeted at addressing the environmental legacy of structural racism.

Complementary benefits of multivariate and hierarchical models for identifying individual differences in cognitive control

Abstract Understanding individual differences in cognitive control is a central goal in psychology and neuroscience. Reliably measuring these differences, however, has proven extremely challenging, at least when using standard measures in cognitive neuroscience such as response times or task-based fMRI activity. While prior work has pinpointed the source of the issue — the vast amount of cross-trial variability within these measures — solutions remain elusive. Here, we do so with one potential way forward: an analytic framework that combines hierarchical Bayesian modeling with multivariate decoding of trial-level fMRI data. Using this framework and longitudinal data from the Dual Mechanisms of Cognitive Control project, we estimated individuals’ neural responses associated with cognitive control within a color-word Stroop task, then assessed the reliability of these individuals’ responses across a time interval of several months. We show that in many prefrontal and parietal brain regions, test–retest reliability was near maximal, and that only hierarchical models were able to reveal this state of affairs. Further, when compared to traditional univariate contrasts, multivariate decoding enabled individual-level correlations to be estimated with significantly greater precision. We specifically link these improvements in precision to the optimized suppression of cross-trial variability in decoding. Together, these findings not only indicate that cognitive control-related neural responses individuate people in a highly stable manner across time, but also suggest that integrating hierarchical and multivariate models provides a powerful approach for investigating individual differences in cognitive control, one that can effectively address the issue of high-variability measures.

Association of Blood Pressure and Retinal Nerve Fiber Layer Rates of Thinning in Patients with Moderate to Advanced Glaucoma.

Investigate the influence of baseline blood pressure (BP) on retinal nerve fiber layer (RNFL) rates of change (RoC) in glaucoma patients with central damage or moderate to severe disease. Prospective cohort study. 110 eyes with ≥4 RNFL optical coherence tomography scans and ≥2 years of follow-up. Global RNFL RoC were modeled with a Bayesian hierarchical model with subject- and sector-level random effects. Influence of baseline systolic and diastolic BP measures and their interactions with intraocular pressure (IOP) on global RNFL rates of change was investigated in prognostic models adjusting for relevant baseline demographic and clinical measures. Magnitude and direction of coefficients for BP, IOP, and their interaction for prediction of global RNFL RoC. One-sided Bayesian p-values denote posterior probability that a regression coefficient is greater than or less than zero with p <0.025 or >0.975 defining significance. Average (SD) 24-2 visual field mean deviation (MD) at baseline, follow-up time, and number of OCT scans were -8.8 (6.0) dB, 4.3 (0.5) years, and 8.3 (1.4), respectively. In multivariable analyses, female sex, Hispanic ethnicity (vs. White ethnicity), better baseline 24-2 MD, higher contrast sensitivity at 12 cycles per degree, presence of diabetes, and thicker central corneal predicted faster RNFL thinning. Adjusted for covariates, lower diastolic BP combined with higher IOP predicted faster RNFL rates of change. Parallel multivariable models incorporating systolic BP showed similar effects. Among various BP/IOP combinations, eyes with IOP at the 90th percentile and diastolic (systolic) BP at 10th percentile demonstrated the fastest RNFL thinning rates (-0.554 and -0.539 μm/year). Low BP and higher IOP at baseline predicted faster (worse) RNFL rates of change in glaucoma patients with central damage or moderate to advanced disease. While there may be potential benefits to BP management in glaucoma patients, the therapeutic value of BP manipulation in glaucoma patients is yet to be established given the proven benefits of tight BP control in reducing cardiovascular morbidity and mortality.

Enhancing robust acoustic DOA estimation against position errors via fast sparse Bayesian learning

Advanced methods for acoustical direction-of-arrival (DOA) estimation leverage sparse signal recovery techniques, which are valued for providing high-resolution results using a limited number of measurement vectors. Although these technologies are capable, they frequently overlook potential errors in sensor positioning, and their computational complexity is excessive, constraining their practical application. This paper introduces a new self-calibrating Bayesian method designed to address this issue. The approach begins by establishing a hierarchical Bayesian model that exploits the sparsity of both signal and position error vectors. Subsequently, a fast algorithm applies a basis selection strategy to efficiently update all variables. Furthermore, the study also explores off-grid estimation. The empirical results indicate that our method outperforms state-of-the-art methods in both accuracy and computational efficiency.

A Bayesian Approach to Modeling Variance of Intensive Longitudinal Biomarker Data as a Predictor of Health Outcomes.

Intensive longitudinal biomarker data are increasingly common in scientific studies that seek temporally granular understanding of the role of behavioral and physiological factors in relation to outcomes of interest. Intensive longitudinal biomarker data, such as those obtained from wearable devices, are often obtained at a high frequency typically resulting in several hundred to thousand observations per individual measured over minutes, hours, or days. Often in longitudinal studies, the primary focus is on relating the means of biomarker trajectories to an outcome, and the variances are treated as nuisance parameters, although they may also be informative for the outcomes. In this paper, we propose a Bayesian hierarchical model to jointly model a cross-sectional outcome and the intensive longitudinal biomarkers. To model the variability of biomarkers and deal with the high intensity of data, we develop subject-level cubic B-splines and allow the sharing of information across individuals for both the residual variability and the random effects variability. Then different levels of variability are extracted and incorporated into an outcome submodel for inferential and predictive purposes. We demonstrate the utility of the proposed model via an application involving bio-monitoring of hertz-level heart rate information from a study on social stress.

Higher Versus Lower Protein Delivery in Critically Ill Patients: A Systematic Review and Bayesian Meta-Analysis.

Recent multicenter trials suggest that higher protein delivery may result in worse outcomes in critically ill patients, but uncertainty remains. An updated Bayesian meta-analysis of recent evidence was conducted to estimate the probabilities of beneficial and harmful treatment effects. An updated systematic search was performed in three databases until September 4, 2024. The study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines and the protocol was preregistered in PROSPERO (CRD42024546387). Randomized controlled trials that studied adult critically ill patients comparing protein doses delivered enterally and/or parenterally with similar energy delivery between groups were included. Data extraction was performed by two authors independently, using a predefined worksheet. The primary outcome was mortality. Posterior probabilities of any benefit (relative risk [RR] < 1.00) or harm (RR > 1.00) and other important beneficial and harmful effect size thresholds were estimated. Risk of bias assessment was performed using the risk of bias 2.0 tool. All analyses were performed using a Bayesian hierarchical random-effects models, under vague priors. Twenty-two randomized trials (n = 4164 patients) were included. The mean protein delivery in the higher and lower protein groups was 1.5 ± 0.6 vs. 0.9 ± 0.4 g/kg/d. The median RR for mortality was 1.01 (95% credible interval, 0.84-1.16). The posterior probability of any mortality benefit from higher protein delivery was 43.6%, while the probability of any harm was 56.4%. The probabilities of a 1% (RR < 0.99) and 5% (RR < 0.95) mortality reduction by higher protein delivery were 38.7% and 22.9%, respectively. Conversely, the probabilities of a 1% (RR > 1.01) and 5% (RR > 1.05) mortality increase were 51.5% and 32.4%, respectively. There is a considerable probability of an increased mortality risk with higher protein delivery in critically ill patients, although a clinically beneficial effect cannot be completely eliminated based on the current data.

Intraoperative Parathyroid Hormone Monitoring Criteria in Primary Hyperparathyroidism

Intraoperative parathyroid hormone (IOPTH) monitoring is recommended by the American Association of Endocrine Surgeons for use during parathyroidectomy for patients with primary hyperparathyroidism (PHPT), but there is no clinician consensus regarding the IOPTH monitoring criteria that optimize diagnostic accuracy. To evaluate and rank the diagnostic properties of IOPTH monitoring criteria used during surgery for patients with PHPT. A bayesian diagnostic test accuracy network meta-analysis (DTA-NMA) was performed, in which peer-reviewed citations from January 1, 1990, to July 22, 2023, were searched for in MEDLINE, Embase, Web of Science, CENTRAL, and CINAHL. All full-text study designs that evaluated any IOPTH monitoring criteria as a diagnostic test were included in this meta-analysis. Any studies evaluating adult patients diagnosed with PHPT undergoing parathyroidectomy were also included. The reference standard used in this study was normalization of calcium and/or parathyroid hormone levels within 1 year of surgery. This DTA-NMA was reported in accordance with the applicable Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines. Two reviewers evaluated all abstracts and full-text articles using a piloted extraction form. A third author resolved any conflicts. There are no published Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) resources for DTA-NMA. The following conventional monitoring criteria were included: Halle, Miami, Rome, Vienna, and PTH normalization, and the following modified criteria were included: Miami and PTH normalization, modified Miami, and modified Vienna. A bayesian hierarchical DTA-NMA model with corresponding 95% credible intervals (CrIs) was used to describe the pooled diagnostic characteristics of the evaluated IOPTH monitoring criteria. Main outcomes included pooled diagnostic test properties, including sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio. A total of 72 studies, which included 19 072 patients, met the inclusion criteria. Sixty-nine studies (95.8%) investigated classic PHPT. In PHPT, the Miami criteria were investigated most often and had the best diagnostic properties (diagnostic odds ratio, 60.00 [95% CrI, 32.00-145.00]) when compared to other conventional criteria. Moreover, the modified Miami criteria, which measures a postexcision IOPTH level 15 minutes or more postexcision of all hyperfunctioning parathyroid tissue, were the overall best criteria (diagnostic odds ratio, 79.71 [95% CrI, 22.46-816.67]). There was a low risk of study bias and no publication bias. The results of this meta-analysis suggest that surgeons should use the modified Miami criteria when performing IOPTH-guided surgery for patients with PHPT because these criteria optimize intraoperative diagnostic accuracy by minimizing unnecessary neck exploration and revision surgery rates.

Hierarchical Bayesian models to mitigate systematic disparities in prediction with proxy outcomes

Abstract Label bias occurs when the outcome of interest is not directly observable and instead, modelling is performed with proxy labels. When the difference between the true outcome and the proxy label is correlated with predictors, this can yield systematic disparities in predictions for different groups of interest. We propose Bayesian hierarchical measurement models to address these issues. When strong prior information about the measurement process is available, our approach improves accuracy and helps with algorithmic fairness. If prior knowledge is limited, our approach allows assessment of the sensitivity of predictions to the unknown specifications of the measurement process. This can help practitioners gauge if enough substantive information is available to guarantee the desired accuracy and avoid disparate predictions when using proxy outcomes. We demonstrate our approach through practical examples.

GLP1Ra-based therapies and DXA-acquired musculoskeletal health outcomes: a focused meta-analysis of placebo-controlled trials.

The objective of this study was to evaluate the effect of glucagon-like peptide-1 receptor agonist (GLP1Ra)-based therapies on change in dual-energy x-ray absorptiometry (DXA)-acquired lean mass (LM) or bone mineral density (BMD). PubMed and Web of Science were searched from database inception through January 29, 2024, for randomized, placebo-controlled trials reporting on change in DXA-acquired LM or BMD measures associated with 12+ weeks of GLP1Ra-based treatment. Of 2618 articles, 9 trials met prespecified search criteria, with 7 reporting on change in total body LM and 2 reporting on change in BMD. For LM outcomes, a hierarchical Bayesian model was used to estimate treatment mean differences. BMD outcomes were described narratively. LM was reported in a total of 659 participants (GLP1Ra-based therapies: n = 419; placebo: n = 240), with follow-up times ranging from mean (SD) 12 to 72 (33.5) weeks. At baseline, participants were aged mean (SD) 41.7 (7.6) years, and 75% were female, with BMI values ranging from 30 to 43 kg/m2. Compared with placebo, GLP1Ra-based treatment was associated with significantly reduced total body weight (-6.9 kg; 95% credible interval [CI]: -10.7 to -3.0). GLP1Ra-based treatment was also associated with significantly reduced LM (-1.9 kg; 95% CI: -3.5 to -0.2). Approximately 30% of body weight lost with GLP1Ra-based therapy is LM. More data are needed assessing BMD outcomes.

Using hierarchical drift diffusion models to elucidate computational mechanisms of reduced reward sensitivity in adolescent major depressive disorder

BackgroundAnhedonia—a core symptom of major depressive disorder (MDD)—is closely related to diminished reward sensitivity. Nonetheless, the psychopathological and computational mechanism underlying anhedonia in young patients with MDD remains unclear. Therefore, this study aims to investigate reward sensitivity in adolescents and young adults with MDD using computational modelling.MethodsOverall, 70 patients with MDD and 54 age- and sex-matched healthy controls (HC) completed a probabilistic reward task (PRT) to assess their general behavioral inclination towards more frequently reinforced stimuli (i.e., “response bias”). Bayesian hierarchical drift diffusion modeling (HDDM) was employed to determine changes in reward sensitivity and computational process during decision-making.ResultsAdolescents with depression showed a trend toward reduced response bias compared to those in HC. HDDM analysis revealed wider decision thresholds in both adolescents and young adults with MDD group. Adolescents with MDD exhibited significantly lower drift rates and reduced starting point bias compared to those in HC. Higher anhedonia levels were linked to lower drift rates and wider decision thresholds. Additionally, increased discriminability correlated with higher drift rates, while higher response bias was linked to larger starting points.ConclusionsOur findings suggest that reduced reward sensitivity and slower evidence accumulation during reward learning may serve as potential indicators of anhedonia in adolescents with MDD. These findings provided crucial insights into the dysregulated positive affect model, underscoring a dysfunctional reward system as a key factor in anhedonia developmental psychopathology in depression.