Sample Weight Research Articles

Association between serum vitamin D and depression among non-alcoholic fatty liver disease.

While previous population-based studies have suggested a link between serum vitamin D levels and depression in individuals with non-alcoholic fatty liver disease (NAFLD), the exact correlation between serum vitamin D and depression among NAFLD patients remains controversial and disputed. Thus, we conducted this study to evaluate the relationship between serum vitamin D and depression in NAFLD participants diagnosed via transient elastography. This cross-sectional study was extracted from the latest NHANES 2017-2018 dataset. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9) score of ≥10. NAFLD phenotype was identified by vibration-controlled transient elastography (VCTE) examination based on diagnostic criteria. Binary logistic regression models were applied to estimate the impact of increased serum vitamin D on the reduced risk of depression based on sample weights. A total of 1339 participants with NAFLD were included in this investigation, of which 127 (8.58%) were diagnosed with depression according to PHQ-9 scores. Binary logistic regression analysis presented that high serum vitamin D level was a protective factor for depression in NAFLD (OR=0.61, 95% CI: 0.37-0.99, p=0.048) after adjusting for all confounding factors. In subgroup analyses, these associations were more pronounced among men (OR=0.32, 95% CI: 0.13-0.81, p=0.024) and obese population (OR=0.53, 95% CI: 0.33-0.86, p=0.019). Increased serum vitamin D was negatively associated to the prevalence of depression in males and obese individuals with NAFLD diagnosed by VCTE.

The impact of protein-energy malnutrition on outcomes in patients with cholangiocarcinoma admitted with sepsis: Insights from a national database.

535 Background: Patients with cholangiocarcinoma (CCA) are prone to infection and related complications. Protein-energy malnutrition (PEM) is common among these patients, arising from either the malignancy or its treatments. PEM is linked to worse medical and surgical outcomes and longer hospital stays across various cancers. The impact of PEM on CCA patients with sepsis, however, remains underexplored. This study aims to fill this gap by assessing how PEM affects in-hospital outcomes in CCA patients admitted with sepsis. Methods: We queried the Healthcare Cost and Utilization Project National Inpatient Sample and identified adult patients with CCA admitted with the principal diagnosis of sepsis from 2018 to 2021, using the ICD-10 code. We categorized the cohort based on the presence or absence of PEM and applied complex sampling weights for national representativeness. We compared socio-demographic characteristics and comorbidities between CCA patients with sepsis who have PEM and those who do not. The primary outcome studied was all-cause in-hospital mortality. Secondary outcomes examined were acute kidney injury (AKI), renal replacement therapy (RRT), respiratory failure (RF), invasive ventilation (IV), septic shock, vasopressor requirement (VP), mean length of hospital stay (LOS), and mean total hospitalization charges (THC). Multivariate regression models assessed outcome disparities between groups, adjusting for patient characteristics and comorbidities, with statistical significance set at p < 0.05. Results: We identified a total of 22,680 adults with CCA who were admitted with the principal diagnosis of sepsis from 2018 to 2021. Among these, 7,245 patients (31.9%) had a concurrent diagnosis of PEM. The PEM group had a higher proportion of female (47.3% vs 44.1%, p=0.04) and non-white (41.7% vs 37.2%, p<0.001) patients and patients with Charlson Comorbidity Index > 3 (86.5% vs 83.3%, p<0.001). Patients with PEM had higher odds of all-cause in-hospital mortality with an adjusted odds ratio (aOR) of 1.26 (95% confidence interval (CI) 1.05-1.51). They also had higher odds of AKI with an aOR of 1.32 (95% CI 1.16-1.50), RF with an aOR of 1.30 (95% CI 1.09-1.51), and septic shock with an aOR of 1.22 (95% CI 1.06-1.40). Patients with PEM also had a higher LOS (9.0 vs 6.3 days) with an adjusted incidence rate ratio (aIRR) of 1.41 (95% CI 1.33-1.49) and THC ($115,971 vs $83,176) with aIRR of 1.35 (95% CI 1.25-1.45). Conclusions: Our study revealed that CCA patients with PEM admitted with sepsis had significantly higher odds of in-hospital mortality, AKI, RF, and septic shock compared to those without PEM. They also experienced longer hospital stays and higher hospitalization charges. These findings highlight the urgent need to address PEM in CCA patients and tailor inpatient management to reduce complications in this high-risk group.

The proper application of logistic regression model in complex survey data: a systematic review

BackgroundLogistic regression is a useful statistical technique commonly used in many fields like healthcare, marketing, or finance to generate insights from binary outcomes (e.g., sick vs. not sick). However, when applying logistic regression to complex survey data, which includes complex sampling designs, specific methodological issues are often overlooked.MethodsThe systematic review extensively searched the PubMed and ScienceDirect databases from January 2015 to December 2021, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, focusing primarily on the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS). 810 articles met the inclusion criteria and were included in the analysis. When discussing logistic regression, the review considered multiple methodological problems such as the model adequacy assessment, handling dependence of observations, utilization of complex survey design, dealing with missing values, outliers, and more.ResultsAmong the selected articles, the DHS database was used the most (96%), with MICS accounting for only 3%, and both DHS and MICS accounting for 1%. Of these, it was found that only 19.7% of the studies employed multilevel mixed-effects logistic regression to account for data dependencies. Model validation techniques were not reported in 94.8% of the studies with limited uses of the bootstrap, jackknife, and other resampling methods. Moreover, sample weights, PSUs, and strata variables were used together in 40.4% of the articles, and 41.7% of the studies did not use any of these variables, which could have produced biased results. Goodness-of-fit assessments were not mentioned in 75.3% of the articles, and the Hosmer–Lemeshow and likelihood ratio test were the most common among those reported. Furthermore, 95.8% of studies did not mention outliers, and only 41.0% of studies corrected for missing information, while only 2.7% applied imputation techniques.ConclusionsThis systematic review highlights important gaps in the use of logistic regression with complex survey data, such as overlooking data dependencies, survey design, and proper validation techniques, along with neglecting outliers, missing data, and goodness-of-fit assessments, all of which point to the need for clearer methodological standards and more thorough reporting to improve the reliability of results. Future research should focus on consistently following these standards to ensure stronger and more dependable findings.

P0049 Bile acid profiling of ileal fluid in Crohn’s disease using metabolomics

Abstract Background An ileal resection is known to cause bile acid malabsorption in Crohn’s disease (CD) 1, but its impact on the bile acid landscape remains understudied. Therefore, we studied bile acid composition in ileal fluid of CD and control patients using targeted metabolomics, and assessed the impact of sample preparation. Methods Ileal fluid was collected from CD patients (n=10) and non-IBD controls (n=10) having an ileostomy (patient characteristics in Table 1). Targeted metabolomics (LC-MS/MS) was applied on full ileal fluid samples as well as on their supernatants (ie. after centrifugation of a full fluid sample aliquot). Bile acid raw abundances were normalised to sample weight (g), and only bile acids detected in at least 40% of all samples (per sample type) were included for further analysis. A comparative analysis between full ileal fluid and its supernatants across all samples was assessed by Spearman correlations, while the bile acid abundances between CD and controls was compared using Mann-Whitney U tests. A nominal p value < 0.05 was considered significant. Results For the CD group, the median [min-max] length of resected ileum was 15.0 [0.0-33.5] cm and 0.0 [0.0-15.0] cm for the control group (Table 1). Ten primary, 6 secondary and 2 modified bile acids were present in full ileal fluid as well as in the corresponding supernatants. While bile acid abundances were 1.2 to 6.0 times higher in full ileal fluid compared to paired supernatants (Figure 1A), the overall bile acid profiles remained highly consistent. In particular, 13 out of 18 bile acids showed a strong correlation between both sample types (r>0.8, p<0.05), and the remaining 5 showed moderate but significant correlations (r=0.5-0.7, p<0.05). Lastly, while the abundances between the full CD and control groups were similar, a subanalysis comparing CD patients with an ileal resection (n=5) to controls with a total mesorectal excision (n=6), revealed an increase in primary bile acids glycocholic acid (GCA) (p=0.052) and glycochenodeoxycholic acid (GCDCA) (p=0.056) in the CD subgroup (Figure 1B). Conclusion In this pilot metabolomics study, we detected both primary and secondary bile acids in the ileal fluid of CD patients and controls. Bile acid composition in full ileal fluid aligns well with its derived supernatant, suggesting supernatant as a viable alternative for measurements, simplifying sample preparation. GCA and GCDCA, both primary bile acids typically reabsorbed in the ileum, tended to be increased after ileal resection in CD patients. A broader screening of bile acid and microbiota metabolites in ileal fluid on a larger scale may unravel novel targets contributing to disease recurrence after resection. References (1)Vitek L. Bile acid malabsorption in inflammatory bowel disease. Inflamm Bowel Dis. 2015;21(2):476-83

Predictors of insecticide-treated nets utilization among children under five years in refugee settlements in Uganda: analysis of the 2018–2019 Uganda Malaria Indicator Survey

BackgroundDespite significant distribution of insecticide-treated net (ITNs) by the Government of Uganda to refugees, malaria is major cause of mortality and morbidity among children under five years in refugee settlements. This highlights the persistent challenges and complexities surrounding malaria control and prevention efforts in these settings. Studies that focus on the determinants of ITN utilization among children under five years in refugee settlements in Uganda are not available. Using the 2018–2019 Uganda’s Malaria Indicator Survey (UMIS) data, analysis of the individual and household factors associated with utilization of ITN among children under five in refugee settlements of Uganda was conducted.MethodsThis study focused on 589 children under five staying in refugee settlements located in Uganda. The extracted variables from the UMIS included social-economic factors associated with ITN utilization. Descriptive analysis was performed to generate summarized statistics, while inferential statistics by way of bivariate analysis were performed to assess the association between the outcome and the independent variables using the chi-square test, and multivariable logistic regression modelling to assess the magnitude of the associations after controlling for other covariates. All analyses considered the survey sampling design and sampling weights, and are conducted in Stata version 18.ResultsThe odds of children sleeping under ITN were higher if their mothers had secondary and higher education (8.1 times) as well as primary education (1.5 times). The odds of children sleeping under ITN reduced by 50% if their mothers were pregnant. Interestingly, the odds of children sleeping under ITN were 70% lower if their mothers knew that ‘not sleeping in nets’ caused malaria. Mothers who were exposed to malaria messages had lower odds of their children sleeping under ITNs.ConclusionsThe results highlight areas of intervention that can increase ITN use in refugee settlements of Uganda. Improving access to education for mothers, providing targeted health education on the importance of ITN, dispelling misconceptions about malaria transmission, facilitating the proper installation of ITNs among others, can all contribute to increased ITN utilization among children under five.

Prevalence of epilepsy across 25 sub-counties in three districts of northern Uganda.

Epilepsy is one of the most common neurological conditions worldwide, with large variation in prevalence across Sub-Saharan Africa countries. Northern Uganda is one of the poorest areas of the country and has seen high density of pigs and prevalence of Taenia solium, a zoonotic tapeworm transmitted which cause neurocysticercosis in humans. The objective of this study was to estimate the population-level prevalence of active epilepsy in 25 sub-counties of northern Uganda. This cross-sectional study was conducted in 2010-2011 in 25 sub-counties of Moyo, Adjumani and Gulu districts, northern Uganda. Participants were sampled using a multistage cluster sampling strategy including sub-counties, parishes, villages, and households as sampling levels. Eligible individuals were interviewed using a previously validated screening questionnaire for epilepsy. Screen positive individuals were further examined by a team of neurologists for confirmation of active epilepsy. Sampling weights and post-stratification to account for sex distribution in each of the 25 sub-counties sampled based on projected 2010 population sizes were applied. A total of 38,303 individuals were sampled across 299 villages from 25 sub-counties. The overall weighted and post-stratified prevalence estimate of active epilepsy was 3.7% (95% Confidence Interval: 3.4%-3.9%). However, there were large variation across sex (4.6% (95%CI: 4.2%-5.0%) in men and 2.7% (95%CI: 2.4%-3.0%) in women) and across sub-counties ranging from 1.7% in Pece Division (Gulu District) and Moyo Town Council (Moyo District) to 6.6% in Awach (Gulu District). People aged between 10 and 19 were the most affected. In northern Uganda, active epilepsy was very prevalent but varied largely across sub-counties. Males were a lot more affected than women, making the use of weighted and post-stratified methods to estimate the prevalence of epilepsy essential. Implementing programs and interventions targeting the control of local risk factors of epilepsy such as neurocysticercosis and improving population health care access could help reduce the rather high prevalence of epilepsy in this area of the country.

A descriptive study of smoking, socioeconomic position, and health-related behaviors in urban Colombia

IntroductionData about recent trends in smoking in Latin America are scarce. We estimated the prevalence of smoking, and explored its association with socioeconomic position, health-related behaviors, perceptions, and trusted information sources in Colombian cities.MethodsWe collected a population-based, multi-stage, stratified probabilistic sample of individuals aged 14–80 years, between November and December 2022. We compared the prevalence of health-related behaviors and perceptions in smokers versus non-smokers employing sample weights according to the sampling design.ResultsThe study included 1904 participants (53.9% female, age 47 ± 19 years). The estimated prevalence of smoking was 22.0% (29.0% in males, 15.7% in females), underage smoking was 4.8%. Among males, the prevalence of smoking was highest in the lowest socioeconomic stratum (31.0%), while among females it was lowest in this stratum (8.3%). Educational level was negatively associated with smoking among men, going from 38.8% for those with only elementary school to 25.5% for college graduates, while among women the trend was the complete opposite. Perceptions about healthy foods and lifestyle habits incorporated in the last year were largely similar between smokers and non-smokers. Smokers and non-smokers appealed essentially to the same health information sources, especially healthcare professionals.ConclusionThe worldwide trend towards a reduction in smoking seems to have been reversed in Colombia, smoking is increasing fast in disadvantaged population segments. Strong tobacco control policies should focus on men with less income and lower education. The same communication avenues can be used to convey health information and motivational campaigns to smokers and non-smokers.

Prevalence of prescription medication use that can exacerbate heart failure among US adults with heart failure.

Heart failure (HF) affects more than 6 million adults in the United States, contributing to substantial morbidity, mortality, and health care costs. Despite advances in medical care, many medications can exacerbate HF, yet their prevalence of use remains unknown. This study examined the national use of prescription medications that could exacerbate HF in adults with self-reported HF. We analyzed data from US adults with self-reported HF in the National Health and Nutrition Examination Survey (NHANES) from 2011 to March 2020. Medications known to exacerbate HF, identified from HF guidelines, were documented through pill bottle reviews. Weighted estimates were used to calculate prevalence overall and by sex, race and ethnicity, and level of evidence for avoidance. Multivariable logistic regression models calculated adjusted odds ratios (aORs) and 95% confidence intervals (95% CIs) for the use of these high-risk medications by sex and race and ethnicity. A total of 687 participants, representing 5.2 million U.S. adults with HF after applying sampling weights, were included (mean age, 66.1 [95% CI 64.9, 67.4] years; 50.4% female [95% CI 45.9%, 55.0%]). Overall, 14.5% (95% CI 10.4%, 19.5%; n = 92) of adults with HF were prescribed at least one medication known to exacerbate HF, with the most common being diltiazem, meloxicam, and ibuprofen. Use of these medications was not significantly different by sex nor by race and ethnicity. Of these medications, 21.7% (95% CI 10.7%, 38.8%) had level A evidence warning against use, and 78.3% (95% CI 61.2%, 89.3%) had B level evidence. Over one-seventh of U.S. adults with HF were likely to have been prescribed medications that could exacerbate the condition, underscoring the need to optimize care. Reducing high-risk medication use may mitigate HF exacerbations and improve outcomes in this vulnerable population.

Association between weight-adjusted-waist index and long-term prognostic outcomes in cardiovascular disease patients: results from the NHANES 1999–2018 study

BackgroundAs cardiovascular disease (CVD) morbidity and mortality increase yearly, this study aimed to explore the potential of the weight-adjusted-waist index (WWI) and its relation to long-term mortality in patients with CVD.MethodsThe diagnosis of CVD was based on standardized medical condition questionnaires that incorporated participants’ self-reported physician diagnoses. WWI (cm/√kg) is a continuous variable and calculated as waist circumference (WC, cm) divided by square root of body weight (kg). For analysis purposes, the participants were divided into four groups based on the quartiles (Q1 – Q4) of the WWI. The study’s primary outcome was all-cause mortality in patients with CVD, with cardiovascular mortality as the secondary outcome, and sample weights and complex survey designs were used to ensure reliable, accurate results.ResultsThe final analysis included 4,445 study participants. In the fully adjusted model, the highest quartile (WWI > 12.05 cm/√ kg) showed a higher all-cause mortality rate compared with the lowest quartile (WWI < 11.03 cm/√ kg) (HR = 1.37, 95% CI: 1.03, 1.82, P < 0.05). The risk of all-cause mortality increased with WWI and showed a linear association in patients with congestive heart failure, heart attack (P-overall < 0.05, P − nonlinear > 0.05); WWI was nonlinearly associated with the risk of all-cause mortality in patients with coronary heart disease and angina (P-overall < 0.05, P − nonlinear < 0.05). Survival curve analysis further showed that all cause and cardiovascular mortality were higher in the high WWI group (Q4) (P < 0.001). The time-dependent receiver operating characteristic (ROC) curve showed that WWI’s area under the curves (AUC) for 5- and 10-year survival rates were 0.76 and 0.792 for all-cause mortality and 0.734 and 0.757 for CVD mortality. WWI’s AUC were higher than those of body mass index (BMI) and WC (all P < 0.01).ConclusionOur findings indicate that a high WWI is positively associated with an increased risk of all-cause mortality. Additionally, the high AUC values for WWI strengthen its potential as a meaningful prognostic marker, underscoring its utility in clinical practice for assessing long-term survival risk in patients with CVD.

DGMSCL: A dynamic graph mixed supervised contrastive learning approach for class imbalanced multivariate time series classification.

In the Imbalanced Multivariate Time Series Classification (ImMTSC) task, minority-class instances typically correspond to critical events, such as system faults in power grids or abnormal health occurrences in medical monitoring. Despite being rare and random, these events are highly significant. The dynamic spatial-temporal relationships between minority-class instances and other instances make them more prone to interference from neighboring instances during classification. Increasing the number of minority-class samples during training often results in overfitting to a single pattern of the minority class. Contrastive learning ensures that majority-class instances learn similar features in the representation space. However, it does not effectively aggregate features from neighboring minority-class instances, hindering its ability to properly represent these instances in the ImMTS dataset. Therefor, we propose a dynamic graph-based mixed supervised contrastive learning method (DGMSCL) that effectively fits minority-class features without increasing their number, while also separating them from other instances in the representation space. First, it reconstructs the input sequence into dynamic graphs and employs a hierarchical attention graph neural network (HAGNN) to generate a discriminative embedding representation between instances. Based on this, we introduce a novel mixed contrast loss, which includes weight-augmented inter-graph supervised contrast (WAIGC) and context-based minority class-aware contrast (MCAC). It adjusts the sample weights based on their quantity and intrinsic characteristics, placing greater emphasis on minority-class loss to produce more effective gradient gains during training. Additionally, it separates minority-class instances from adjacent transitional instances in the representation space, enhancing their representational capacity. Extensive experiments across various scenarios and datasets with differing degrees of imbalance demonstrate that DGMSCL consistently outperforms existing baseline models. Specifically, DGMSCL achieves higher overall classification accuracy, as evidenced by significantly improved average F1-score, G-mean, and kappa coefficient across multiple datasets. Moreover, classification results on a real-world power data show that DGMSCL generalizes well to real-world application.

ADMNet: adaptive deformable convolution large model combining multi-level progressive fusion for Building Change Detection

ABSTRACT Building Change Detection (BCD) based on high-resolution Remote Sensing Images (RSI) simplifies urban surface monitoring. Nevertheless, the mainstream detection methods utilizing traditional convolution and attention mechanisms are often prone to errors due to the loss of edge detail information and underutilization of global context information. To address these issues, this paper presents a large model, namely ADMNet, which is built on adaptive deformable convolution and is designed to handles various types of building change information. First, we propose a Siamese neural network based on adaptive deformable convolution (ADC) modules. The ADC module incorporates spatial offset parameters into convolutional kernel sampling and mapping weights to capture irregularly varying edge features for local adaptive receptive fields. Second, we utilize a large model semantically driven to enhance model context awareness and construct long-range feature dependencies from multi-scale edge information, which are then integrated with locally adaptive edge structure features to achieve accurate edge localization. Furthermore, we design a Multi-Level Progressive Feature Fusion (MLPFF) module that enhances feature characterization capabilities to ensure internal integrity and improves model detection performance by integrating a priori knowledge from large-model transfer learning. To evaluate the effectiveness and generalizability of ADMNet, we conduct comparative experiments with current mainstream methods on two building datasets, LEVIR-CD and WHU-CD, and a land cover dataset, SYSU-CD. The results show that ADMNet outperforms all comparative methods. The source code is available at https://github.com/spaceYu180/ADMNet.

Understanding the private-public school performance gap in PISA: evidence from Portugal

We analyse the PISA-reported convergence in the performance of private and public schools in Portugal. When PISA sampling weights are used, the number of students enrolled in those types of schools and specific grades/tracks of study differs significantly from official population figures. To account for those differences, we apply a post-stratification adjustment; however, sample sizes are small, resulting in estimates with low precision for several subgroups. We propose recommendations for improving the handling of these issues in future PISA cycles. In an additional analysis, we also account for changes in the composition of the student population. When all factors are considered, the convergence in scores is far less impressive than reported. For instance, in Science, after adjusting the sampling weights and removing population composition effects, the reported convergence of 46 points between private and public schools from 2015 to 2018 amounts to only 9 points. The decomposition and sample adjustment methods used in this paper can be easily adapted to other contexts.

Urban-rural disparities in cervical cancer screening among Indian women between 30–49 years: a geospatial and decomposition analysis using a nationally representative survey

IntroductionExisting evidence suggests a lower uptake of cervical cancer screening among Indian women. Coverage is lower in rural than urban women, but such disparities are less explored. So, the present study was conducted to explore the self-reported coverage of cervical cancer screening in urban and rural areas stratified by socio-demographic characteristics, determine the spatial patterns and identify any regional variations, ascertain the factors contributing to urban-rural disparities and those influencing the likelihood of screening among women aged 30–49 years factors residing in urban, rural, and overall Indian settings.MethodsWe did a secondary analysis of the fifth round of the National Family Health Survey in India (2019-21) data with a sample size of 3,48,882 women. The coverage of cervical cancer screening was estimated using sampling weights. Urban-rural differences were compared using the chi-square test. Spatial patterns were analysed using aggregated district-level data, and the contribution of different independent variables to the urban-rural disparities was estimated using multivariate decomposition analysis. Multivariable logistic regression was conducted using STATA 17 to obtain the significant factors of reported screening in urban and rural areas.ResultsThe nationwide coverage of cervical cancer screening was 2.0% (95% CI: 1.9-2.0). The urban (2.4%; 2.3–2.5) participants had higher screening coverage than their rural (1.8%; 1.7–1.8) counterparts. Moran’s I statistic confirmed the presence of spatial dependence and geographical gradient. Decomposition analysis depicted small urban-rural differences in the screening coverage of 0.60% (0.4–0.8). Endowment and coefficient contributed to 88.15% and 11.85% of the disparities. Compositional changes were contributed majorly by regional differences, low education, scheduled tribes, and having living children > 2. Higher odds of having screening were associated with older age (AOR 1.45, 95% CI: 1.03–1.28), higher education (1.32; 1.13–1.55), higher age of first intercourse (1.60; 1.43–1.79), married (1.25; 1.08–1.45) and diabetic (1.39; 1.17–1.65) women, and those from South India (6.76; 5.90–7.75). The odds were lower among Muslims, scheduled tribes and participants using hormonal contraceptives.ConclusionThere are significant urban-rural disparities in cervical cancer screening uptake that can be attributed to regional variation, educational inequalities, tribal groups, socio-economic inequalities and parity, necessitating the need to comprehensively design tailor-made advocacy initiatives and simultaneously address the broader determinants of health.

Machine learning in public health informatics: Evidence that complex sampling structures may not be needed for prediction models with imbalanced outcomes.

The objective of this study is to investigate the predictive ability of machine learning models for imbalanced outcomes from national survey data without the use of sampling weights. We evaluated the predictive performance of machine learning models on imbalanced outcomes from the US National Health and Nutrition Examination Survey (USNHANES) without using sampling weights. Four machine learning models (support vector machine, random forest, least absolute shrinkage and selection operator regression, and deep neural network) were compared with a logistic model that incorporates the survey's complex sampling design. Three resampling methods (oversampling, undersampling, and combined) were used to address class imbalance during the model training process. For all models, the balanced accuracy was similar (ranging from 0.72 to 0.76) and the specificity was smaller than sensitivity except for random forest. The support vector machine and neural networks performed best with sensitivity (ranging from 0.79 to 0.83), while the random forest had the largest specificity (ranging from 0.86 to 0.96), with one exception. PR-AUC scores and Brier scores were low ranging from 0.2529 to 0.3313 (lower scores worse) and 0.1005-0.3245 (lower scores better), respectively CONCLUSIONS: The machine learning models had overall similar predictive capacity to the recommended methods which integrate the complex sampling design for the prediction of osteoarthritis occurrence with USNHANES.

Unpacking the Mood States of Children and Youth in Saskatchewan, Canada, in the Context of the COVID-19 Pandemic: Insights from the "See Us, Hear Us 2.0" Study.

The COVID-19 pandemic created a growing need for insights into the mental health of children and youth and their use of coping mechanisms during this period. We assessed mood symptoms and related factors among children and youth in Saskatchewan. We examined if coping abilities mediated the relationship between risk factors and mood states. "See Us, Hear Us 2.0", a cross-sectional study of 563 child-parent dyads, provided the data. The dependent variable, current mood state, was measured by the CoRonavIruS health Impact Survey (CRISIS) scale. Independent variables included sociodemographics, behaviors, household conditions, and coping ability. Multiple linear regression and mediation analyses were conducted, ensuring sample representativeness with sampling weights and addressing missing data through multiple imputations. The participants reported mood symptoms ("moderate" to "extreme") ranging from 23% to 38% on the CRISIS scale. Factors such as older children, hybrid learning, disrupted activities, and increased screen time worsened moods. The ethnic minority groups (BIPOC) living in mid-sized cities/towns experienced more negative moods compared to Whites residing in cities. Coping ability mediated the relationship between extracurricular activities and mood states. Our results underscore the importance of tailored interventions, recognizing the diverse needs of specific age groups, gender identities, and ethnicities and addressing the adverse effects of the pandemic-related disruptions on the mental health and well-being of school children in Saskatchewan. Our study also suggests prioritizing the diverse needs of children and youth during the planning and implementation of mental health services in the province.

Geographic factors associated with SARS-CoV-2 prevalence during the first wave − 6 districts in Zambia, July 2020

BackgroundGeographical factors can affect infectious disease transmission, including SARS-CoV-2, a virus that is spread through respiratory secretions. Prioritization of surveillance and response activities during a pandemic can be informed by a pathogen’s geographical transmission patterns. We assessed the relationship between geographical factors and SARS-CoV-2 prevalence in Zambia.MethodsWe did a cross-sectional study of SARS-CoV-2 prevalence in six districts in July 2020, which was during the upslope of the first wave in Zambia. In each district, 16 Standard Enumeration Areas (SEAs) were randomly selected and 20 households from each SEA were sampled. The SEA PCR prevalence was calculated as the number of persons testing PCR positive for SARS-CoV-2 in the SEA times the individual sampling weight for the SEA divided by the SEA population. We analysed SEA geographical data for population density, socioeconomic status (SES) (with lower scores indicating reduced vulnerability), literacy, access to water, and sanitation, and hygiene (WASH) factors. Gaussian conditional autoregressive (CAR) models and Generalised estimating equations (GEE) were used to measure adjusted prevalence Ratios (aPRs) and 95% confidence intervals (CIs) for SARS-CoV-2 prevalence with geographical factors, after adjusting for clustering by district, in R.ResultsOverall, the median SARS-CoV-2 prevalence in the 96 SEAs was 41.7 (Interquartile range (IQR), 0.0-96.2) infections per 1000 persons. In the multivariable CAR analysis, increasing SES vulnerability was associated with lower SARS-CoV-2 prevalence (aPR) = 0.85, 95% CI: 0.78–0.94). Conversely, urban SEAs and poor access to WASH were associated with a higher SARS-CoV-2 prevalence (aPR = 1.73, 95% CI: 1.46–2.03, No soap: aPR = 1.47, 95% CI: 1.05–2.05, households without piped water: aPR = 1.32, 95% CI: 1.05–1.65, 30 min to fetch water: aPR = 23.39, 95% CI: 8.89–61.52). Findings were similar in the multivariable GEE analysis.ConclusionsSARS-CoV-2 prevalence was higher in wealthier, urban EAs, with poor access to WASH. As this study was conducted early in the first wave could have impacted our findings. Additional analyses from subsequent waves could confirm if these findings persist. During the beginning of a COVID-19 wave in Zambia, surveillance and response activities should be focused on urban population centres and improving access to WASH.

Degradation Potential of Metakaolin-Based Geopolymer Composites Immersed in Real and Simulated Acidic Environments

This study investigates the degradation potential of metakaolin-based geopolymer (GP) composites when exposed to real and simulated acidic environments. Traditional OPC concrete, commonly used in wastewater treatment facilities, faces considerable deterioration due to the destructive chemical composition of municipal wastewater. This extensive investigation aims to assess the performance of GP composites as a sustainable alternative to such materials. The metakaolin-based GP mortar samples were prepared and subjected to immersion in a primary clarifier unit at a local wastewater treatment plant (real) and a laboratory-made acetic acid solution (simulation) for up to four weeks after curing. The analysis included measurements of % strength and % weight loss, as well as characterization techniques such as isothermal calorimetry, microstructure (SEM), and mercury porosimetry (MIP). The outcomes signified a cumulative heat generation of 534 J/g at three days, with an average compressive strength of 79.4 MPa past 28 days. Exposure to acetic acid led to a 13% decline in compressive strength and a 3.90% loss in sample weight, while exposure to real wastewater resulted in an 18% strength and a 5.60% weight loss. Observations from SEM revealed microstructural changes, including the formation of biofilms and air voids, indicating multifaceted interactions between the GP matrix and its surrounding environment. This research effectively highlights the potential of metakaolin-based geopolymer composites to improve durability against acidic conditions, suggesting future applications in the construction of infrastructure exposed to such harsh chemical environments.

The protective effect of frequent religious service attendance on pregnancy loss

BackgroundAccumulating evidence suggests that spirituality and religiosity may be associated with improved health outcomes. However, few studies have examined maternal religiosity as a protective factor for perinatal outcomes. We explored the association between maternal religious attendance and pregnancy loss.MethodsData were drawn from the Future Families & Child Wellbeing Study’s first and second waves and medical records (n = 1874). Religious attendance was a self-reported response to the question “About how often do you attend religious services?” Pregnancy loss was measured from responses to the second wave survey question, “Since focal child’s birth, have you had any miscarriages/abortions/stillbirths?” Logistic regression estimated odds ratios (OR) and 95% confidence intervals (CI) for the association between maternal religious attendance frequency and pregnancy loss. Models were adjusted for sampling weights, religious preference, socioeconomic and behavioral factors.Results8% (n = 164) of mothers reported having a pregnancy loss. Of those with a pregnancy loss, 28% (n = 46) attended services hardly ever and 20% (n = 20) attended services once a week or more. Women who attended services more frequently had 58% increased odds of not experiencing a pregnancy loss (OR:1.58;95%CI:1.01,2.48) after adjusting for potential confounding. A post hoc analysis found no difference in pregnancy loss type or subsequent reproductive history based on attendance level.ConclusionsResults suggest that higher maternal religious attendance frequency may be a protective factor for pregnancy loss. Further research is needed to understand the association between maternal religious attendance and mechanisms for pregnancy loss.

Differences in US Adult Dietary Patterns by Cardiovascular Health and Socioeconomic Vulnerability.

Naturally occurring dietary patterns, a major contributor to health, are not well described among those with cardiovascular disease (CVD) - particularly in light of socioeconomic vulnerability. We sought to identify major dietary patterns in the US and their distribution by CVD, social risk factors, and Supplemental Nutrition Assistance Program (SNAP) participation. This was a cross-sectional study among 32,498 noninstitutionalized adults from the National Health and Nutrition Examination Survey (2009-2020). We used principal component analysis to identify common dietary patterns. Individuals were assigned to the pattern for which they had the highest component score. Using multinomial logit regression, we estimated the percentage whose diets aligned with each pattern in population subgroups stratified by CVD, social risk factors, and SNAP. Analyses were adjusted for age, gender, race and ethnicity, total energy intake, and year, with sampling weights to provide nationally representative estimates. Four dietary patterns were identified among US adults: American (33.7%; high in solid fats, added sugars, and refined grains), Prudent (22.6%; high in vegetables, nuts/seeds, oils, seafood, and poultry), Legume (15.8%), and Fruit/Whole Grain/Dairy (27.9%), that together explained 29.2% of dietary variance. More adults with prevalent CVD (37.1%) than without (33.3%, p=0.005) aligned with the American Pattern, with no differences among other patterns. Each additional social risk factor associated with more adults aligned with American (2.5% absolute increase) and Legume (1.3%), and fewer aligned with Prudent (-1.9%) and Fruit/Whole Grain/Dairy (-1.9%) patterns (p<0.001 each). Analysis of dietary patterns across SNAP participation showed higher proportion of SNAP participants and income-eligible SNAP non-participants compared to non-eligible adults for the American (40.2% [38.1, 42.3%], 35.1% [32.7, 37.5%], 31.9% [31.0, 32.8%], respectively) and Legume patterns (17.2% [15.6, 18.8%], 17.8% [16.1, 19.5%]), 15.4% [14.6,16.1%], respectively) and less for Prudent (17.0% [15.5, 18.6%], 20.2% [18.2, 22.3%], 24.2% [23.3, 25.1%], respectively) and Fruit/Whole Grain/Dairy Patterns (25.6% [23.8%, 27.3%], 26.9%[27.6%,29.5%], 28.6% [27.6%, 29.5%], respectively). Empirical dietary patterns vary by CVD and socioeconomic vulnerability. Initiatives to improve nutrition in at-risk individuals should consider these naturally occurring dietary patterns and their variation in key subgroups.

Spectral Graph Sample Weighting for Interpretable Sub-cohort Analysis in Predictive Models for Neuroimaging.

Recent advancements in medicine have confirmed that brain disorders often comprise multiple subtypes of mechanisms, developmental trajectories, or severity levels. Such heterogeneity is often associated with demographic aspects (e.g., sex) or disease-related contributors (e.g., genetics). Thus, the predictive power of machine learning models used for symptom prediction varies across subjects based on such factors. To model this heterogeneity, one can assign each training sample a factor-dependent weight, which modulates the subject's contribution to the overall objective loss function. To this end, we propose to model the subject weights as a linear combination of the eigenbases of a spectral population graph that captures the similarity of factors across subjects. In doing so, the learned weights smoothly vary across the graph, highlighting sub-cohorts with high and low predictability. Our proposed sample weighting scheme is evaluated on two tasks. First, we predict initiation of heavy alcohol drinking in young adulthood from imaging and neuropsychological measures from the National Consortium on Alcohol and NeuroDevelopment in Adolescence (NCANDA). Next, we detect Dementia vs. Mild Cognitive Impairment (MCI) using imaging and demographic measurements in subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Compared to existing sample weighting schemes, our sample weights improve interpretability and highlight sub-cohorts with distinct characteristics and varying model accuracy.