Perception and reality: The mismatch between absolute and relative physical activity intensity during pregnancy and postpartum in United States women.

OBJECTIVE: To explore whether a mismatch between absolute physical activity intensity (PAI) and relative self-reported PAI exists during pregnancy and postpartum. METHODS: Women from the PIN3/Postpartum study completed physical activity questionnaires during pregnancy (n = 770; Trimester 2: T2, Trimester 3: T3) and postpartum (n = 181; 3 months: PP3, 12 months PP12) (2001-2005). Activities women engaged in were assigned Metabolic Equivalent (MET) values for absolute intensity; women self-reported perceived exertion (using the Borg scale) for each activity to provide relative intensity. Hierarchical regression models were used to determine whether a mismatch between absolute and relative PAI (for moderate or vigorous physical activity (MPA; VPA)) differed during pregnancy and postpartum. Models were adjusted for socio-demographic factors. RESULTS: Women commonly overestimated the amount of MPA and VPA they engaged in [T2 MPA mean 60.5 min/week (49.1, 72.0), VPA 3.7 (-1.4, 8.8); T3: MPA 47.7 (38.9, 56.4), 2.9 (-1.7, 7.4); PP3: MPA 69.5 (43.9, 95.1), VPA 15.8 (1.8, 29.7); PP12: MPA 42.20 (26.8, 57.6), VPA 2.75 (-7.8, 12.9)]. Women overestimated both MPA and VPA to a lesser extent at T3 compared to T2 (MPA: ß for difference:-12.6 [95%CI: -26.0, -0.9]; VPA: -0.9 [-6.4, 4.6]). Women continued to overestimate their MPA at PP3 and PP12. CONCLUSIONS: Compared to absolute PAI, perceived PAI was greater for MPA compared to VPA and differences persisted from pregnancy through postpartum. Future research should focus on how perceptions relate to women's actual physiological capacity and whether this mismatch influences the amount of physical activity women engage in during the transition to motherhood.

Bias analyses to investigate the impact of differential participation: Application to a birth defects case-control study.

BACKGROUND: Certain associations observed in the National Birth Defects Prevention Study (NBDPS) contrasted with other research or were from areas with mixed findings, including no decrease in odds of spina bifida with periconceptional folic acid supplementation, moderately increased cleft palate odds with ondansetron use and reduced hypospadias odds with maternal smoking. OBJECTIVES: To investigate the plausibility and extent of differential participation to produce effect estimates observed in NBDPS. METHODS: We searched the literature for factors related to these exposures and participation and conducted deterministic quantitative bias analyses. We estimated case-control participation and expected exposure prevalence based on internal and external reports, respectively. For the folic acid-spina bifida and ondansetron-cleft palate analyses, we hypothesized the true odds ratio (OR) based on prior studies and quantified the degree of exposure over- (or under-) representation to produce the crude OR (cOR) in NBDPS. For the smoking-hypospadias analysis, we estimated the extent of selection bias needed to nullify the association as well as the maximum potential harmful OR. RESULTS: Under our assumptions (participation, exposure prevalence, true OR), there was overrepresentation of folic acid use and underrepresentation of ondansetron use and smoking among participants. Folic acid-exposed spina bifida cases would need to have been ≥1.2× more likely to participate than exposed controls to yield the observed null cOR. Ondansetron-exposed cleft palate cases would need to have been 1.6× more likely to participate than exposed controls if the true OR is null. Smoking-exposed hypospadias cases would need to have been ≥1.2 times less likely to participate than exposed controls for the association to falsely appear protective (upper bound of selection bias adjusted smoking-hypospadias OR = 2.02). CONCLUSIONS: Differential participation could partly explain certain associations observed in NBDPS, but questions remain about why. Potential impacts of other systematic errors (e.g. exposure misclassification) could be informed by additional research.

Association of educational attainment with cancer mortality in a national cohort study of black and white adults: A mediation analysis.

Background: Low educational attainment is associated with excess cancer mortality. However, the mechanisms driving this association remain unknown. Methods: Using data from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, we evaluated the associations of participant and parental/caregiver education with cancer mortality using Cox proportional hazards models, adjusting for socio-demographic characteristics and health conditions. We used principal components analysis to generate indices of measures representing the social determinants of health (SDOH) and health behaviors. We used structural equation modeling to determine if the association between educational attainment and cancer mortality was mediated by these domains. Results: Among 30,177 REGARDS participants included in this analysis, 3798 (12.6%) had less than a high school degree. In fully adjusted models, those without a high school education experienced about 50% greater risk of death than high school graduates and higher (White participants HR: 1.47; 95% CI: 1.23, 1.76 and Black HR: 1.54; 95% CI: 1.33, 1.79). There was evidence of a modest mediation effect for the association between education and cancer mortality by the SDOH domain score (White total effect HR: 1.25; 95% CI: 1.18, 1.33, indirect effect HR: 1.04; 95% CI: 1.03, 1.05, direct effect HR: 1.21; 95% CI: 1.14, 1.28 and Black total effect HR: 1.24; 95% CI: 1.18, 1.29, indirect effect HR: 1.04; 95% CI: 1.03, 1.05, direct effect HR: 1.19; 95% CI: 1.14, 1.24). There was no evidence of mediation by the health behaviors score. No significant associations were found for female caregiver/mother's or male caregiver/father's education (N = 13,209). Conclusions: In conclusion, participant education was strongly associated with cancer mortality, and this association was partially mediated by the SDOH domain score.

Bayesian matrix completion for hypothesis testing.

We aim to infer bioactivity of each chemical by assay endpoint combination, addressing sparsity of toxicology data. We propose a Bayesian hierarchical framework which borrows information across different chemicals and assay endpoints, facilitates out-of-sample prediction of activity for chemicals not yet assayed, quantifies uncertainty of predicted activity, and adjusts for multiplicity in hypothesis testing. Furthermore, this paper makes a novel attempt in toxicology to simultaneously model heteroscedastic errors and a nonparametric mean function, leading to a broader definition of activity whose need has been suggested by toxicologists. Real application identifies chemicals most likely active for neurodevelopmental disorders and obesity.

Association Between Change in Physical Activity During Pregnancy and Infant Birth Weight.

OBJECTIVE: We assessed whether total, recreational, and non-recreational physical activity (PA) assessed twice during pregnancy, and its change, were associated with infant birth weight and small for gestational age (SGA). METHODS: We included 1467 Pregnancy, Infection, and Nutrition 3 Study participants who self-reported PA at time 1 (T1: 17-22 weeks' gestation) and time 2 (T2: 27-30 weeks' gestation). We assessed last week absolute intensities of PA (moderate: 4.7-7.1 METs; and vigorous: > 7.1 METs) and perceived intensities. Change in hours/week of PA was assessed continuously or categorically (increase or decrease ≥ 1 hour, and no change). Associations of continuous PA hours/week at T1, T2, and its change, with sex-specific z-scores of birth weight, were assessed using multivariable linear robust regressions. We used logistic regressions to assess categorical PA measures with SGA. Models were adjusted for adequacy of maternal weight gain, general health, maternal age, parity, race/ethnicity, and smoking. RESULTS: Hours/week of total and recreational absolute intensities of PA at T1, T2, and its change were generally not associated with birth weight, although two measures of non-recreational PA at T2 and its change were associated with increased birth weight. Perceived intensities of PA (at T1, T2, and its change) were largely not associated with sex-specific z-scores of infant birth weight. Absolute and perceived intensity PA were not associated with SGA. CONCLUSIONS FOR PRACTICE: In this observational cohort, increases and decreases in PA during pregnancy were not associated with differential changes in birthweight or SGA.

The association of prenatal phthalates, organophosphorous pesticides, and organophosphate esters with early child language ability in Norway.

BACKGROUND: Prenatal exposure to phthalates, organophosphate esters, and organophosphorous pesticides have been associated with neurodevelopmental deficits including language ability, however, few studies consider the effect of exposure mixtures and the potential longitudinal detriments over time. OBJECTIVE: This study examines the influence of prenatal exposure to phthalates, organophosphate esters, and organophosphorous pesticides, on children's language ability from toddlerhood to the preschool period. METHODS: This study includes 299 mother-child dyads from Norway in the Norwegian Mother, Father and Child Cohort Study (MoBa). Prenatal exposure to chemicals were assessed at 17 weeks' gestation, and child language skills were assessed at 18 months using the Ages and Stages Questionnaire communication subscale and at preschool age using the Child Development Inventory. We ran two structural equation models to examine the simultaneous influences of chemical exposures on parent-reported and teacher-reported child language ability. RESULTS: Prenatal organophosphorous pesticides were negatively associated with preschool language ability through language ability at 18 months. Additionally, there was a negative association between low molecular weight phthalates and teacher-reported preschool language ability. There was no effect of prenatal organophosphate esters on child language ability at either 18 months or preschool age. CONCLUSIONS: This study adds to the literature on prenatal exposure to chemicals and neurodevelopment and highlights the importance of developmental pathways in early childhood.

Prenatal Exposure to Organophosphorus Pesticides and Preschool ADHD in the Norwegian Mother, Father and Child Cohort Study.

Prenatal organophosphorus pesticide (OPP) exposure has been associated with child attention-deficit/hyperactivity disorder (ADHD) in agricultural communities and those that are exposed to residentially applied insecticides. To examine this association in populations that are exposed primarily through diet, we estimate the associations between prenatal OPP exposure and preschool ADHD in the Norwegian Mother, Father and Child Cohort Study (MoBa), and describe modification by paraoxonase 1 (PON1) gene variants. We used participants from the MoBa Preschool ADHD Sub-study (n = 259 cases) and a random sample of MoBa sub-cohort participants (n = 547) with birth years from 2004 to 2008. Prenatal urinary dialkylphosphate (DAP) metabolites (total diethylphosphate [∑DEP] and total dimethylphosphate [∑DMP]) were measured by an ultra-performance liquid chromatography-time-of-flight system and summed by molar concentration. Maternal DNA was genotyped for coding variants of PON1 (Q192R and L55M). We used a multivariable logistic regression to calculate the odds ratios (OR) and 95% confidence intervals, adjusted for maternal education, parity, income dependency, age, marital status, ADHD-like symptoms, pesticide use, produce consumption, and season. We found no associations between DAP metabolite concentrations and preschool ADHD. The adjusted ORs for exposure quartiles 2-4 relative to 1 were slightly inverse. No monotonic trends were observed, and the estimates lacked precision, likely due to the small sample size and variation in the population. We found no evidence of modification by PON1 SNP variation or child sex. Maternal urinary DAP concentrations were not associated with preschool ADHD.

Maternal Physical Activity at Term and Spontaneous Labor: A Case-Crossover Study.

BACKGROUND: This study assessed associations between antenatal physical activity and the onset of spontaneous labor (SL). METHODS: Data were taken from 541 participants in the third pregnancy, infection, and nutrition study who had no contraindications to antenatal physical activity. An interviewer-administered questionnaire assessed labor triggers, gestational age at birth, and physical activity within the week (24 h to 7 d) and the 24-hour period (0-24 h) prior to SL. A case-crossover design examined the association between physical activity (recreational, occupational, or any) and the risk of onset of SL within the subsequent 24 hours. RESULTS: Overall, 21% (any), 26% (recreational), and 14% (occupational) of participants reported physical activity during the week; whereas 5% (any), 7% (recreational), and 3% (occupational) reported physical activity during the 24-hour period, prior to SL onset. Participants who reported any or occupational physical activity during the 24-hour period had a decreased likelihood of SL within the subsequent 24 hours, while participants who reported at least 30 minutes of recreational physical activity had an increased likelihood. Results remained consistent among early, full, or postterm participants. CONCLUSION: Recreational, but not occupational, physical activity at term may increase the likelihood of SL; however, the authors cannot rule out reverse causality.

Meal patterning and the onset of spontaneous labor.

BACKGROUND: There is a lack of consensus in the literature about the association between meal patterning during pregnancy and birth outcomes. This study examined whether maternal meal patterning in the week before birth was associated with an increased likelihood of imminent spontaneous labor. METHODS: Data came from 607 participants in the third phase of the Pregnancy, Infection, and Nutrition Study (PIN3). Data were collected through an interviewer-administered questionnaire after birth, before hospital discharge. Questions included the typical number of meals and snacks consumed daily, during both the week before labor onset and the 24-hour period before labor onset. A self-matched, case-crossover study design examined the association between skipping one or more meals and the likelihood of spontaneous labor onset within the subsequent 24 hours. RESULTS: Among women who experienced spontaneous labor, 87.0% reported routinely eating three daily meals (breakfast, lunch, and dinner) during the week before their labor began, but only 71.2% reported eating three meals during the 24-hour period before their labor began. Compared with the week before their labor, the odds of imminent spontaneous labor were 5.43 times as high (95% CI: 3.41-8.65) within 24 hours of skipping 1 or more meals. The association between skipping 1 or more meals and the onset of spontaneous labor remained elevated for both pregnant individuals who birthed early (37-<39 weeks) and full-term (≥39 weeks). CONCLUSIONS: Skipping meals later in pregnancy was associated with an increased likelihood of imminent spontaneous labor, though we are unable to rule out reverse causality.

Socioeconomic gradients in the Westernization of diet in China over 20 years.

INTRODUCTION: In low-middle income countries, urbanization leads to changes from traditional to Western diet, which are often accompanied by reductions in cardiometabolic health. Whether socioeconomic status buffers these urbanization-related diet changes over time is unknown. OBJECTIVE: To examine whether the association between urbanization and a key indicator of Western diet, percent of calories from animal-source foods (1) varies depending on income and (2) whether this association changes over time. MATERIALS AND METHODS: We used data from nine waves of the longitudinal, population-based China Health and Nutrition Survey [n = 22,360 Chinese adults (1991-2015)], followed across 24 years, including diet data from 3 repeated 24-h dietary recalls. We used simultaneous year-stratified linear regression models to examine whether changes in the association between urbanization level and percent of calories from animal-source foods differed by income. Models allowed for variation in associations across the 24 years of urbanization, accounting for within-individual correlation over time and controlling for age, sex, region, physical activity, and caloric intake. RESULTS: In 1991, on average 15% of calories for Chinese adults came from animal source foods and by 2015, this percentage rose to approximately one quarter of total calories. Over the 24 years of follow-up, urbanization and income were each strongly related to percent of calories from animal-source foods with differential association across income levels (p < 0.0001). We also found evidence that this association changed over time (p < 0.0001). Income gradients in animal source food consumption were smallest in the most urban areas in early years with some temporal variation, but over time income gradients narrowed in some later years in low and moderately urbanized areas. However, by 2015 there were few income differences in animal source food consumption across urbanization levels. CONCLUSIONS: Throughout 24-years of urbanization, income seemed to buffer the transition from traditional to Western diet. However, the degree to which income buffered these urbanization-related changes depended on the level and history of community urbanization. At later stages of urbanization when Western diet behaviors were more widespread, urban-rural differences in Western diet behaviors varied little by income.

Bayesian hierarchical factor regression models to infer cause of death from verbal autopsy data.

In low-resource settings where vital registration of death is not routine it is often of critical interest to determine and study the cause of death (COD) for individuals and the cause-specific mortality fraction (CSMF) for populations. Post-mortem autopsies, considered the gold standard for COD assignment, are often difficult or impossible to implement due to deaths occurring outside the hospital, expense, and/or cultural norms. For this reason, Verbal Autopsies (VAs) are commonly conducted, consisting of a questionnaire administered to next of kin recording demographic information, known medical conditions, symptoms, and other factors for the decedent. This article proposes a novel class of hierarchical factor regression models that avoid restrictive assumptions of standard methods, allow both the mean and covariance to vary with COD category, and can include covariate information on the decedent, region, or events surrounding death. Taking a Bayesian approach to inference, this work develops an MCMC algorithm and validates the FActor Regression for Verbal Autopsy (FARVA) model in simulation experiments. An application of FARVA to real VA data shows improved goodness-of-fit and better predictive performance in inferring COD and CSMF over competing methods. Code and a user manual are made available at https://github.com/kelrenmor/farva.

Gestational Phthalate Exposure and Preschool Attention Deficit Hyperactivity Disorder in Norway.

Prenatal phthalate exposure has been linked to altered neurobehavioral development in both animal models and epidemiologic studies, but whether or not these associations translate to increased risk of neurodevelopmental disorders is unclear. We used a nested case-cohort study design to assess whether maternal urinary concentrations of 12 phthalate metabolites at 17 weeks gestation were associated with criteria for Attention Deficit Hyperactivity Disorder (ADHD) classified among 3-year-old children in the Norwegian Mother, Father and Child Cohort Study (MoBa). Between 2007 and 2011, 260 children in this substudy were classified with ADHD using a standardized, on-site clinical assessment; they were compared with 549 population-based controls. We modeled phthalate levels both linearly and by quintiles in logistic regression models adjusted for relevant covariates and tested for interaction by child sex. Children of mothers in the highest quintile of di-iso-nonyl phthalate (∑DiNP) metabolite levels had 1.70 times the odds of being classified with ADHD compared with those in the lowest quintile (95% confidence interval [CI] = 1.03 to 2.82). In linear models, there was a trend with the sum of di-2-ethylhexyl phthalate metabolites (∑DEHP); each natural log-unit increase in concentration was associated with 1.22 times the odds of ADHD (95% CI = 0.99 to 1.52). In boys, but not girls, mono-n-butyl phthalate exposure was associated with increased odds of ADHD (odds ratio [OR] 1.42; 95% CI = 1.07 to 1.88). Additional adjustment for correlated phthalate metabolites attenuated estimates. These results suggest gestational phthalate exposure may impact the behavior of children as young as 3 years.

Pregnancy exposure to common-detect organophosphate esters and phthalates and maternal thyroid function.

BACKGROUND: Contemporary human populations are exposed to elevated concentrations of organophosphate esters (OPEs) and phthalates. Some metabolites have been linked with altered thyroid function, however, inconsistencies exist across thyroid function biomarkers. Research on OPEs is sparse, particularly during pregnancy, when maintaining normal thyroid function is critical to maternal and fetal health. In this paper, we aimed to characterize relationships between OPEs and phthalates exposure and maternal thyroid function during pregnancy, using a cross-sectional investigation of pregnant women nested within the Norwegian Mother, Father, and Child Cohort (MoBa). METHODS: We included 473 pregnant women, who were euthyroid and provided bio-samples at 17 weeks' gestation (2004-2008). Four OPE and six phthalate metabolites were measured from urine; six thyroid function biomarkers were estimated from blood. Relationships between thyroid function biomarkers and log-transformed concentrations of OPE and phthalate metabolites were characterized using two approaches that both accounted for confounding by co-exposures: co-pollutant adjusted general linear model (GLM) and Bayesian Kernal Machine Regression (BKMR). RESULTS: We restricted our analysis to common-detect OPE and phthalate metabolites (>94%): diphenyl phosphate (DPHP), di-n-butyl phosphate (DNBP), and all phthalate metabolites. In GLM, pregnant women with summed di-isononyl phthalate metabolites (∑DiNP) concentrations in the 75th percentile had a 0.37 ng/µg lower total triiodothyronine (TT3): total thyroxine (TT4) ratio (95% credible interval: [-0.59, -0.15]) as compared to those in the 25th percentile, possibly due to small but diverging influences on TT3 (-1.99 ng/dL [-4.52, 0.53]) and TT4 (0.13 µg/dL [-0.01, 0.26]). Similar trends were observed for DNBP and inverse associations were observed for DPHP, monoethyl phthalate, mono-isobutyl phthalate, and mono-n-butyl phthalate. Most associations observed in co-pollutants adjusted GLMs were attenuated towards the null in BKMR, except for the case of ∑DiNP and TT3:TT4 ratio (-0.48 [-0.96, 0.003]). CONCLUSIONS: Maternal thyroid function varied modestly with ∑DiNP, whereas results for DPHP varied by the type of statistical models.

Prenatal phthalate exposures and executive function in preschool children.

BACKGROUND: Prenatal phthalate exposure has been linked with altered neurodevelopment, including externalizing behaviors and attention-deficit hyperactivity disorder (ADHD). However, the implicated metabolite, neurobehavioral endpoint, and child sex have not always been consistent across studies, possibly due to heterogeneity in neurodevelopmental instruments. The complex set of findings may be synthesized using executive function (EF), a construct of complex cognitive processes that facilitate ongoing goal-directed behaviors. Impaired EF can be presented with various phenotypes of poor neurodevelopment, differently across structured conditions, home/community, or preschool/school. We evaluated the relationship between prenatal phthalate exposure and comprehensive assessment of preschool EF. METHODS: Our study comprised 262 children with clinically significant/subthreshold ADHD symptoms and 78 typically developing children who were born between 2003 and 2008 and participated in the Preschool ADHD Substudy, which is nested within a population-based prospective cohort study, the Norwegian Mother, Father, and Child Cohort (MoBa). Twelve phthalate metabolites were measured from urine samples that their mothers had provided during pregnancy, at 17 weeks' gestation. All children, at approximately 3.5-years, took part in a detailed clinical assessment that included parent-and teacher-rated inventories and administered tests. We used instruments that measured constructs related to EF, which include a parent-and teacher-reported Behavior Rating Inventory of Executive Function-Preschool (BRIEF-P) and three performance-based tests: A Developmental NEuroPSYchological Assessment (NEPSY), Stanford-Binet intelligence test V (SB5), and the cookie delay task (CDT). The standard deviation change in test score per interquartile range (IQR) increase in phthalate metabolite was estimated with multivariable linear regression. We applied weighting in all models to account for the oversampling of children with clinically significant or subthreshold symptoms of ADHD. Additionally, we assessed modification by child sex and potential co-pollutant confounding. RESULTS: Elevated exposure to mono-benzyl phthalate (MBzP) during pregnancy was associated with poorer EF, across all domains and instruments, in both sex. For example, an IQR increase in MBzP was associated with poorer working memory rated by parent (1.23 [95% CI: 0.20, 2.26]) and teacher (1.13 [0.14, 2.13]) using BRIEF-P, and administered tests such as SB5 (no-verbal: 0.19 [0.09, 0.28]; verbal: 0.13 [0.01, 0.25]). Adverse associations were also observed for mono-n-butyl phthalate (MnBP) and mono-iso-butyl phthalate (MiBP), although results varied by instruments. EF domains reported by parents using BRIEF-P were most apparently implicated, with stronger associations among boys (e.g., MnBP and inhibition: 2.74 [1.77, 3.72]; MiBP and inhibition: 1.88 [0.84, 2.92]) than among girls (e.g., MnBP and inhibition: -0.63 [-2.08, 0.83], interaction p-value: 0.04; MiBP and inhibition: -0.15 [-1.04, 0.74], interaction p-value: 0.3). Differences by sex, however, were not found for the teacher-rated BRIEF-P or administered tests including NEPSY, SB5, and CDT. CONCLUSION AND RELEVANCE: Elevated mid-pregnancy MBzP, MiBP, and MnBP were associated with more adverse profiles of EF among preschool-aged children across a range of instruments and raters, with some associations found only among boys. Given our findings and accumulating evidence of the prenatal period as a critical window for phthalate exposure, there is a timely need to expand the current phthalate regulations focused on baby products to include pregnancy exposures.

PERTURBED FACTOR ANALYSIS: ACCOUNTING FOR GROUP DIFFERENCES IN EXPOSURE PROFILES.

In this article we investigate group differences in phthalate exposure profiles using NHANES data. Phthalates are a family of industrial chemicals used in plastics and as solvents. There is increasing evidence of adverse health effects of exposure to phthalates on reproduction and neurodevelopment and concern about racial disparities in exposure. We would like to identify a single set of low-dimensional factors summarizing exposure to different chemicals, while allowing differences across groups. Improving on current multigroup additive factor models, we propose a class of Perturbed Factor Analysis (PFA) models that assume a common factor structure after perturbing the data via multiplication by a group-specific matrix. Bayesian inference algorithms are defined using a matrix normal hierarchical model for the perturbation matrices. The resulting model is just as flexible as current approaches in allowing arbitrarily large differences across groups but has substantial advantages that we illustrate in simulation studies. Applying PFA to NHANES data, we learn common factors summarizing exposures to phthalates, while showing clear differences across groups.

BAYESIAN JOINT MODELING OF CHEMICAL STRUCTURE AND DOSE RESPONSE CURVES.

Today there are approximately 85,000 chemicals regulated under the Toxic Substances Control Act, with around 2,000 new chemicals introduced each year. It is impossible to screen all of these chemicals for potential toxic effects, either via full organism in vivo studies or in vitro high-throughput screening (HTS) programs. Toxicologists face the challenge of choosing which chemicals to screen, and predicting the toxicity of as yet unscreened chemicals. Our goal is to describe how variation in chemical structure relates to variation in toxicological response to enable in silico toxicity characterization designed to meet both of these challenges. With our Bayesian partially Supervised Sparse and Smooth Factor Analysis (BS3FA) model, we learn a distance between chemicals targeted to toxicity, rather than one based on molecular structure alone. Our model also enables the prediction of chemical dose-response profiles based on chemical structure (i.e., without in vivo or in vitro testing) by taking advantage of a large database of chemicals that have already been tested for toxicity in HTS programs. We show superior simulation performance in distance learning and modest to large gains in predictive ability compared to existing methods. Results from the high-throughput screening data application elucidate the relationship between chemical structure and a toxicity-relevant high-throughput assay. An R package for BS3FA is available online at https://github.com/kelrenmor/bs3fa.

Centered Partition Processes: Informative Priors for Clustering (with Discussion).

There is a very rich literature proposing Bayesian approaches for clustering starting with a prior probability distribution on partitions. Most approaches assume exchangeability, leading to simple representations in terms of Exchangeable Partition Probability Functions (EPPF). Gibbs-type priors encompass a broad class of such cases, including Dirichlet and Pitman-Yor processes. Even though there have been some proposals to relax the exchangeability assumption, allowing covariate-dependence and partial exchangeability, limited consideration has been given on how to include concrete prior knowledge on the partition. For example, we are motivated by an epidemiological application, in which we wish to cluster birth defects into groups and we have prior knowledge of an initial clustering provided by experts. As a general approach for including such prior knowledge, we propose a Centered Partition (CP) process that modifies the EPPF to favor partitions close to an initial one. Some properties of the CP prior are described, a general algorithm for posterior computation is developed, and we illustrate the methodology through simulation examples and an application to the motivating epidemiology study of birth defects.