Lowest Bayesian Information Criterion Research Articles

Conceptualizing the relationship between parental education and own cognition in the Kaiser Healthy Aging and Diverse Life Experiences (KHANDLE) study

AbstractBackgroundFew studies have considered how parental education influences late‐life cognition; those that have often focus on maternal and not paternal influences or as part of a broader SES index, which does not allow us to understand the functional form of the relationship between parental education and cognition. We assessed whether mother’s education dichotomized at 8th grade, a commonly used cutoff among similarly aged older adults to our sample, sufficiently captured effects of parental education by systematically evaluating conceptualizations of parental education.MethodsKaiser Healthy Aging and Diverse Life Experiences (KHANDLE) study baseline participants, aged 65 years plus (n=1,633), self‐reported own and parents’ education at baseline. Parental education was operationalized separately as (1) mother’s only, (2) father’s only, (3) mother and father together; each operationalization was considered as categorical (missing, <=8th grade, incomplete high school, high school diploma, postsecondary education), as >8th grade (with missingness indicator and with missing included in “low”), and as high school completion (with and without missingness indicator). Parental (4) concordance (both high, high mother, high father, both low), (5) highest parental level, and (6) lowest parental level were also evaluated. Verbal episodic memory (VEM), semantic memory (SEM), and executive functioning (EF) from the Spanish and English Neuropsychological Assessment Scales (SENAS) were considered separately in covariate‐adjusted linear models. The lowest Bayesian Information Criterion (BIC) model was considered to provide the best fit.ResultsAcross all parental education operationalizations, 8th grade cut‐offs performed best; categorical (5‐level) parent education performed the worst. Semantic memory was best predicted by the referent measure of mother’s education >8th (B̂: 0.33; 95% CI: 0.25, 0.42); however verbal memory was best predicted by mother’s education >8th grade (B̂: 0.14; 95% CI: 0.03, 0.24) plus missing indicator (B̂: ‐0.24; 95% CI: ‐0.37, ‐0.10) and executive function by both mother’s (B̂: 0.22; 95% CI: 0.12, 0.31) and father’s (B̂: 0.18; 95% CI: 0.09, 0.28) >8th together.ConclusionThe preferable performance of parental education as >8th grade likely reflects historical education trends when KHANDLE participants’ parents were educated; whether preference for dichotomized measures over full categories reflects uncertainties in recall of parental education remains unclear.

Improved Characterization of Diffusion in Normal and Cancerous Prostate Tissue Through Optimization of Multicompartmental Signal Models.

Multicompartmental modeling outperforms conventional diffusion-weighted imaging (DWI) in the assessment of prostate cancer. Optimized multicompartmental models could further improve the detection and characterization of prostate cancer. To optimize multicompartmental signal models and apply them to study diffusion in normal and cancerous prostate tissue in vivo. Retrospective. Forty-six patients who underwent MRI examination for suspected prostate cancer; 23 had prostate cancer and 23 had no detectable cancer. 3T multishell diffusion-weighted sequence. Multicompartmental models with 2-5 tissue compartments were fit to DWI data from the prostate to determine optimal compartmental apparent diffusion coefficients (ADCs). These ADCs were used to compute signal contributions from the different compartments. The Bayesian Information Criterion (BIC) and model-fitting residuals were calculated to quantify model complexity and goodness-of-fit. Tumor contrast-to-noise ratio (CNR) and tumor-to-background signal intensity ratio (SIR) were computed for conventional DWI and multicompartmental signal-contribution maps. Analysis of variance (ANOVA) and two-sample t-tests (α = 0.05) were used to compare fitting residuals between prostate regions and between multicompartmental models. T-tests (α = 0.05) were also used to assess differences in compartmental signal-fraction between tissue types and CNR/SIR between conventional DWI and multicompartmental models. The lowest BIC was observed from the 4-compartment model, with optimal ADCs of 5.2e-4, 1.9e-3, 3.0e-3, and >3.0e-2 mm2 /sec. Fitting residuals from multicompartmental models were significantly lower than from conventional ADC mapping (P < 0.05). Residuals were lowest in the peripheral zone and highest in tumors. Tumor tissue showed the largest reduction in fitting residual by increasing model order. Tumors had a greater proportion of signal from compartment 1 than normal tissue (P < 0.05). Tumor CNR and SIR were greater on compartment-1 signal maps than conventional DWI (P < 0.05) and increased with model order. The 4-compartment signal model best described diffusion in the prostate. Compartmental signal contributions revealed by this model may improve assessment of prostate cancer. Level of Evidence 3 Technical Efficacy Stage 3 J. MAGN. RESON. IMAGING 2021;53:628-639.

Computational analysis of SARS-CoV-2, SARS-CoV, and MERS-CoV genome using MEGA.

The novel coronavirus pandemic that has originated from China and spread throughout the world in three months. Genome of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) predecessor, severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV) play an important role in understanding the concept of genetic variation. In this paper, the genomic data accessed from National Center for Biotechnology Information (NCBI) through Molecular Evolutionary Genetic Analysis (MEGA) for statistical analysis. Firstly, the Bayesian information criterion (BIC) and Akaike information criterion (AICc) are used to evaluate the best substitution pattern. Secondly, the maximum likelihood method used to estimate of transition/transversions (R) through Kimura-2, Tamura-3, Hasegawa-Kishino-Yano, and Tamura-Nei nucleotide substitutions model. Thirdly and finally nucleotide frequencies computed based on genomic data of NCBI. The results indicate that general times reversible model has the lowest BIC and AICc score 347,394 and 347,287, respectively. The transition/transversions bias for nucleotide substitutions models varies from 0.56 to 0.59 in MEGA output. The average nitrogenous bases frequency of U, C, A, and G are 31.74, 19.48, 28.04, and 20.74, respectively in percentages. Overall the genomic data analysis of SARS-CoV-2, SARS-CoV, and MERS-CoV highlights the close genetic relationship.

Transmission of SARS-CoV-2 in South Asian countries: molecular evolutionary model based phylogenetic and mutation analysis.

The on-going coronavirus disease 19 (COVID-19) pandemic has caused a very high number of infections and deaths around the globe. The absence of vaccine/drugs to counter COVID-19 has scrambled scientific communities to repurpose available medicines/vaccines. As the virus is known to mutate, using the whole genome sequences, the transmission dynamics and molecular evolutionary models were evaluated for South Asian countries to determine the evolutionary rate of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Phylogenetic analyses were done using the data available on National Center for Biotechnology Information (NCBI). Different nucleotide substitution models and molecular evolutionary models were analyzed to see how SARS-CoV-2 was transmitted in the populations. Models for the viral ‘S’ and ‘N’ protein from selected strains were constructed, validated, and analyzed to determine the mutations and discover the potential therapeutics against this deadly viral disease. We found that the Hasegawa-Kishino-Yano (HKY) nucleotide substitution model was the best model with the lowest Bayesian information criterion (BIC) scores. Molecular clock RelTime analysis showed the evolutionary rate of SARS-CoV-2 substitutions in the genome was at 95% confidence interval, and heterogeneity was observed. Several mutations in the viral S-protein were found with one in the receptor-binding domain concerning SARS-CoV-2/Wuhan-1/S-Protein. Nucleocapsid protein also showed mutations in the strains from India and Sri Lanka. Our analysis suggests that SARS-CoV-2 is evolving at a diverse rate. The mutation leading to substitution in the nucleotide sequence occurred in the genome during the transmission of COVID-19 among individuals in the South Asian countries.Electronic supplementary materialThe online version of this article (10.1007/s42398-020-00123-z) contains supplementary material, which is available to authorized users.

Desorption isotherms and isosteric heat of araticum (Annona crassiflora Mart.) seeds

ABSTRACT The objective of this study was to determine the desorption isotherms and isosteric heat of Annona crassiflora Mart. seeds, using Akaike information criterion (AIC) and Schwarz’s Bayesian information criterion (BIC) to assist in the choice of the nonlinear regression model. The desorption isotherms were determined by indirect static method and water activity was obtained using the instrument HygroPalm; the product was put in the device in B.O.D. chamber set at 10, 20, 30 and 40 °C. Several nonlinear regression models were fitted to the experimental data by the Gauss-Newton method. The desorption isotherms of Annona crassiflora Mart. seeds can be represented by the models of Chung-Pfost, Copace, Modified GAB, Modified Henderson, Modified Oswin, Sabbah and Sigma Copace. However, the Sigma Copace model showed better fit to the experimental data, with lower AIC and BIC values, being chosen to represent the desorption isotherms of Annona crassiflora Mart seeds. Isosteric heat increased with decreasing moisture content, requiring a greater amount of energy to remove water from seeds, with values ranging from 2541.64 to 2481.56 kJ kg-1, for the moisture content range from 5.69 to 14.93% on a dry basis.

Obstructive Sleep Apnea Is Associated with Newly Diagnosed Gestational Diabetes Mellitus.

Rationale: Sleep-disordered breathing (SDB) is associated with increased risk of adverse pregnancy outcomes, including gestational diabetes mellitus (GDM). GDM is a significant cause of maternal and infant morbidities. Assessing these risk factors concurrently may facilitate both the identification of women at GDM risk and the initiation of GDM prevention strategies.Objectives: To investigate whether SDB events, including SDB in rapid eye movement (REM) sleep and other sleep parameters, are associated with increased risk of GDM and to evaluate the performance of the models investigating associations between breathing and sleep parameters and GDM risk.Methods: In this case-control study, 46 women with newly diagnosed GDM and 46 healthy control subjects, who were individually matched for age, gestational age, body mass index, race, and parity, completed overnight polysomnographic studies and sleep questionnaires after being screened for GDM during the late-second to mid-third trimesters. Conditional logistic regression analysis was used to identify models investigating associations between risk factors and GDM risk. The Bayesian information criterion (BIC) was employed to compare models; the model with the lowest BIC is preferred.Results: Obstructive sleep apnea (OSA; defined as an apnea-hypopnea index [AHI] >5 events/h) was present in 22% of subjects with GDM and 9% of control subjects (P < 0.001). Women with OSA had a higher GDM risk (odds ratio [OR], 4.71; 95% confidence interval [CI], 1.05-21.04). In individual models, GDM risk was also significantly higher among women with higher overall AHI (events/h OR, 1.81; 95% CI, 1.01-3.27), higher AHI in REM (events/h OR, 2.09; 95% CI, 1.02-4.31), higher oxygen desaturation index greater than or equal to 4% (ODI4; events/h OR, 2.21; 95% CI, 1.03-4.73), and higher Sleep Apnea Symptom Score (OR, 2.72; 95% CI, 1.11-6.69). The percentage of non-REM sleep was significantly associated with decreased risk of GDM (percentage of non-REM sleep OR, 0.88; 95% CI, 0.78-0.99). The BIC supports the conclusion that there is a strong association between AHI in REM and GDM risk compared with the other significant models.Conclusions: SDB events, including REM-related OSA, are linked to increased GDM risk. GDM risk is also influenced by intercorrelated sleep variables.

Evaluation of pathologic staging using number of nodes in p16-negative head and neck cancer.

The 8th edition AJCC staging guidelines for head and neck squamous cell carcinoma (HNSCC) recently introduced pathologic staging criteria for nodal disease among p16-positive patients. In this study we evaluate pathologic staging in p16-negative HNSCC. We compared pathologic staging to the 7th and 8th edition AJCC staging systems using a statewide population-based cohort. All M0 p16-negative surgical patients were included. The outcome was five-year overall survival. Of 304 patients identified, 113 were N0, 157 had 1-4 positive nodes, and 34 had ≥4 nodes. Survival was 71% (95% CI 61-78%) with no nodes, 48% (36%-60%) for 1-4 nodes, and 24% (11 - 39%) for>4 nodes. When compared to the AJCC systems, the pathologic staging yielded a larger total survival gradient, more montonic survival, better consistency across primary sites, and a slightly lower Bayesian information criterion (1510 vs 1538). After adjusting for disease characteristics, demographics, and tobacco use, hazard ratios for survival were similar using pathologic and AJCC criteria. In this cohort, pathological staging was more prognostic than AJCC staging. This is the first study to evaluate pathologic staging in p16-negative cancer; if these findings are verified, a universal nodal staging system could be introduced.

Is the incidence of breast cancer increasing? Using the NCDB and autoregressive integrated moving average models (ARIMA) to forecast breast cancer to 2030.

e13612 Background: The observed increase in the incidence rate of BC has been described due, in part, to higher, effective public health campaigns for screening, higher sensitivity among diagnostic testing modalities as well as an aging population. While the mortality rate associated with breast cancer has reported to fall by 1.8% every year since 2007, the disease burden itself has continued to grow and increased breast cancer rates will contribute to the strain already experienced within the United States healthcare system infrastructure. Therefore, the authors sought to use a large, national registry to develop a time series model that might help forecast the approximate breast cancer incidence rate, to the year 2030, as captured by the National Cancer Database (NCDB). Methods: In this time series forecast, autoregressive integrated moving average models (ARIMA) were constructed based on 2004-2016 historic breast cancer incidence rates, as reported by the NCDB. Multiple models were generated, using differing autoregressive parameters, and the most predictive model was chosen using the lowest Bayesian Information Criteria (BIC), and mean absolute percentage error (MAPE). Similar methodology has already been published to predict prostate cancer incidence. The best fit model was applied to forecast annual incidence in the NCDB to the year 2030. Statistics were performed using modeling systems in SPSS, version 26. Results: For this model, 12 years of NCDB breast cancer diagnoses were used, which included n = 1,924,425 cases, overall. Using ARIMA modeling, a best fit, stationary average was identified with autoregressive and difference terms of 1 (ARIMA (1,1,0), coefficient = 0.598; P = 0.028). Of the multiple models tested the model with the lowest BIC was chosen, with a MAPE of 4.71%. The best fit model forecasted n = 325,048 new breast cancer diagnoses to be captured annually by the NCDB, by 2030. Conclusions: In this analysis, the annual breast cancer incidence within the NCDB is predicted to increase by 21%, by 2030. This forecast, while slightly lower than previously reported by the National Cancer Institute, utilizes more recent historical data that reflects a period of leveling-off in disease incidence, during 2014-2016, as reported to the NCDB. This innovative model can be utilized to proactively plan public health strategies and allocate appropriate resources focused on reducing the burden of cancer.

Predictors of physical activity in older adults early in an emergency hospital admission: a prospective cohort study

BackgroundReduced mobility may be responsible for functional decline and acute sarcopenia in older hospitalised patients. The drivers of reduced in-hospital mobility are poorly understood, especially during the early phase of acute hospitalisation. We investigated predictors of in-hospital activity during a 24-h period in the first 48 h of hospital admission in older adults.MethodsThis was a secondary analysis of a prospective repeated measures cohort study. Participants aged 75 years or older were recruited within the first 24 h of admission. At recruitment, patients underwent a baseline assessment including measurements of pre-morbid functional mobility, cognition, frailty, falls efficacy, co-morbidity, acute illness severity, knee extension strength and grip strength, and consented to wear accelerometers to measure physical activity during the first 7 days (or until discharge if earlier). In-hospital physical activity was defined as the amount of upright time (standing or walking). To examine the predictors of physical activity, we limited the analysis to the first 24 h of recording to maximise the sample size as due to discharge from hospital there was daily attrition. We used a best subset analysis including all baseline measures. The optimal model was defined by having the lowest Bayesian information criterion in the best-subset analyses. The model specified a maximum of 5 covariates and used an exhaustive search.ResultsSeventy participants were recruited but eight were excluded from the final analysis due to lack of accelerometer data within the first 24 h after recruitment. Patients spent a median of 0.50 h (IQR: 0.21; 1.43) standing or walking. The optimal model selected the following covariates: functional mobility as measured by the de Morton Mobility Index and two measures of illness severity, the National Early Warning Score, and serum C-reactive protein.ConclusionsPhysical activity, particularly in the acute phase of hospitalisation, is very low in older adults. The association between illness severity and physical activity may be explained by symptoms of acute illness being barriers to activity. Interdisciplinary approaches are required to identify early mobilisation opportunities.

Multisensory Integration Develops Prior to Crossmodal Recalibration.

It has been hypothesized that crossmodal recalibration plays a crucial role for the development of multisensory integration capabilities [1]. To test the developmental trajectory of multisensory integration and crossmodal recalibration, we used a combined ventriloquist/ventriloquist aftereffect paradigm [2] in children aged 5-9 years. The ventriloquist effect (indicating multisensory integration), that is, the shift of auditory localization toward simultaneously presented but spatially discrepant visual stimuli, was larger in children than in adults, which was attributed to a lower auditory localization precision in the children. In fact, the size of the ventriloquist effect depended on the visual stimulus reliability in both children and adults. In all groups, the ventriloquist effect was best explained by a causal inference model. In contrast to their multisensory integration capabilities, 5-year-old children did not recalibrate. The immediate ventriloquist aftereffect (indicating recalibration after a single exposure to a spatially discrepant audio-visual stimulus) emerged in 6- to 7-year-old children, whereas the cumulative ventriloquist aftereffect (reflecting recalibration to the audio-visual spatial discrepancies over the complete experiment) was not observed before the age of 8 years. First, in contrast to common beliefs, the present results provide evidence that multisensory integration precedes rather than follows crossmodal recalibration during development. Second, we report developmental evidence for a dissociation of the processes involved in multisensory integration and immediate as well as cumulative recalibration. We speculate that multisensory integration is a prerequisite for crossmodal recalibration, because the multisensory percept, rather than unimodal cues, might comprise a crucial signal for the calibration of the sensory systems.

Improved Parsimonious Topic Modeling Based on the Bayesian Information Criterion.

In a previous work, a parsimonious topic model (PTM) was proposed for text corpora. In that work, unlike LDA, the modeling determined a subset of salient words for each topic, with topic-specific probabilities, with the rest of the words in the dictionary explained by a universal shared model. Further, in LDA all topics are in principle present in every document. In contrast, PTM gives sparse topic representation, determining the (small) subset of relevant topics for each document. A customized Bayesian information criterion (BIC) was derived, balancing model complexity and goodness of fit, with the BIC minimized to jointly determine the entire model—the topic-specific words, document-specific topics, all model parameter values, and the total number of topics—in a wholly unsupervised fashion. In the present work, several important modeling and algorithm (parameter learning) extensions of PTM are proposed. First, we modify the BIC objective function using a lossless coding scheme with low modeling cost for describing words that are non-salient for all topics—such words are essentially identified as wholly noisy/uninformative. This approach increases the PTM’s model sparsity, which also allows model selection of more topics and with lower BIC cost than the original PTM. Second, in the original PTM model learning strategy, word switches were updated sequentially, which is myopic and susceptible to finding poor locally optimal solutions. Here, instead, we jointly optimize all the switches that correspond to the same word (across topics). This approach jointly optimizes many more parameters at each step than the original PTM, which in principle should be less susceptible to finding poor local minima. Results on several document data sets show that our proposed method outperformed the original PTM model with respect to multiple performance measures, and gave a sparser topic model representation than the original PTM.

Abstract P131: Optimal Phenomapping Strategy to Identify Novel Subgroups of Patients With Type 2 Diabetes and High Cardiovascular Disease Risk

Introduction: Phenomapping can be used to identify subgroups of patients with similar risk factor profiles; however, there is no widely accepted clustering method. We compared 3 phenomapping strategies to identify groups of individuals with diabetes mellitus (DM) and high cardiovascular disease (CVD) risk. Methods: Participants with DM and no baseline CVD from the ACCORD trial (N = 6591) were included in this study. Among 45 mixed data variables, we used an informed variable selection method with optimization of the Wald index to select the top 20 covariates associated with survival. Gaussian mixture models (GMM), Latent Class Analysis (LCA), and finite mixture model-based clustering (FMM) were performed to identify mutually exclusive phenogroups. Bayesian information criterion (BIC) and Dunn index were calculated for each method and compared. The primary composite endpoint was cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke. Results: FMM clustering with 3 phenogroups was the optimal clustering strategy as determined by the lowest BIC and highest Dunn index (Figs A-B). Phenogroup 1 had a higher burden of comorbidities and DM complications; phenogroup 2 were older with intermediate CV comorbidity burden, but lowest complications from DM; and phenogroup 3, the largest group, had a lower prevalence of CV-related comorbidities. In adjusted Cox models, over 4.9 years median follow-up, phenogroup 1 and 2 participants (vs. phenogroup 3) had a significantly higher risk of primary composite endpoint (Fig C). In Cox analysis, when group membership was added to 10 optimal predictors of primary composite endpoint, the C -index improved from 0.68 to 0.70 ( P = 0.02). Similar risk patterns were also observed in an external cohort (Look AHEAD trial, N = 4211) with the highest risk of adverse outcome seen in phenogroup 1 (Fig D). Conclusion: Semi-supervised clustering using FMM efficiently identifies phenogroups of patients with DM with distinct clinical characteristics and CVD risk.

Parametric Frailty Model with Time-Dependent Covariates

The parametric frailty model has been used in this study where, the term frailty is used to represent an unobservable random effect shared by subjects with similar (unmeasured) risks in the analysis of mortality rate. In real-life environment, the application of frailty models have been widely used by biostatistician, economists and epidemiologist to donate proneness to disease, accidents and other events because there are persistent differences in susceptibility among individuals. When heterogeneity is ignored in a study of survival analysis the result will produce an incorrect estimation of parameters and standard errors. This study used gamma and Weibull distribution for the frailty model. The first objective of this study is to investigate parametric model with time dependent covariates on frailty model. The derivation is using either classical maximum likelihood or Monte Carlo integration. The second objective is to measure the effectiveness of Gamma and Weibull frailty model with and without time-dependent covariates. This is done by calculating the root mean square error (RMSE). The last objective is to assess the goodness of fit of Gamma and Weibull frailty model with and without time-dependent covariates using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Simulation is used in order to obtain the RMSE, AIC ad BIC value if time-dependent covariate does not exists. Between both models with time-dependent covariate, Weibull frailty distribution has lower AIC and BIC compared to Gamma frailty distribution. Therefore, Weibull frailty distribution with time-dependent covariate is preferable when a time-dependent covariate exists in a data.

Temporal analysis of visceral leishmaniasis between 2000 and 2019 in Ardabil Province, Iran: A time-series study using ARIMA model.

Background:Visceral leishmaniasis in human (VLH) also known as kala-azar is a neglected disease of humans that mainly occurs in more than 50 countries mostly located in the Eastern Mediterranean and the Northern America.Objective:The purpose of this study was to determine the temporal patterns and predict of occurrence of VL in Ardabil Province, in northwestern Iran using autoregressive integrated moving average (ARIMA) models.Methods:This descriptive study employed yearly and monthly data of 602 cases of VLH in the province between January 2000 to December 2019, which was provided by the leishmaniasis national surveillance system. The monthly occurrences case constructed the ARIMA model of time-series model. The insignificance of the correlation in the lags of 12, 24 and 36 months, and Chi-square test showed the occurrence of VLH does not have a seasonal pattern. Eleven potential ARIMA models were examined for VLH cases. Finally, the best model was selected with the lower Akaike Information Criteria (AIC) and Bayesian information criterion (BIC) value. Then, the selected model was used to forecast frequency of monthly occurrences case. The forecasting precision was estimated by mean absolute percentage error (MAPE). Data analysis was performed using Stata14 and its package time series analysis.Results:ARIMA (5, 0, 1) model with AIC (25.7) and BIC (43.35) was selected. The MAPE value was 26.89% and the portmanteau test for white noise was (Q = 23.02, P = 0.98) for the residuals of the selected model showed that the data were fully modelled. The total cumulative VLH cases in the next 24 months’ in Ardabil province predicted 14 cases (95% CI: 4-54 case).Conclusion:The ARIMA (5, 0, 1) model can be a useful tool to predict VLH cases as early warning system and the results are helpful for policy makers and primary care physicians in the readiness of public health problems before the outbreak of the disease.

Worry less about the algorithm, more about the sequence of events

Background: Many algorithms exist for learning network structure and parameters from data. Some of these algorithms include Grow Shrink, Incremental Association Markov Blanket, IAMB, Fast IAMB, and Interleaved IAMB, Hill Climbing, Restricted Maximization and Maximum-Minimum Hill Climbing. These algorithms optimize the fit to the data, while ignoring the order of occurrences of the variables in the network structure. Objective: This paper examines if sequence information (i.e., one variable occurs before another) can make algorithms for learning directed acyclical graph networks more accurate. Methods: A 13- variable network was simulated, where information on sequence of occurrence of some of the variables was assumed to be known. In each simulation 10,000 observations were generated. These observations were used by 4 conditional dependency and 4 search and score algorithms to discover the network from the simulated data. Partial sequence was used to prohibit a directed arc from a later variable to an earlier one. The Area under the Receiver Operating Curve (AROC) was used to compare the accuracy of the sequence-constrained and unconstrained algorithms in predicting the last node in the network. In addition, we examined the performance of sequence constrained algorithms in a real data set. We analyzed 1.3 million disability assessments done on 296,051 residents in Veterans Affairs nursing homes; where the sequence of occurrence of variables was inferred from the average age of occurrence of disabilities. We constructed three networks using Grow-Shrink algorithm, one without and the other two use two permutation of the observed sequence. The fit of these three models to data was examined using Bayesian Information Criterion (BIC). Results: In simulated data, algorithms that used sequenced constraints (AROC = 0.94, confidence intervals, C.I. = 0.86 to 1) were significantly more accurate than the same algorithm without use of sequence constraints (AROC = 0.74, C.I. = 0.65 to 0.83). The agreement between discovered and observed networks improved from range of 0.54 to 0.97 to range of 0.88 to 1. In the real data set, the Bayesian network constructed with use of sequence had 6% lower BIC scores. Conclusions: Sequence information improved accuracy of all eight learning algorithms and should be routinely examined in learning network structure from data.

Model Prediction and Optimization of Waste Lube Oil Treated with Natural Clay

In this work, used lube oil was treated using natural acid-free clay. Clay was added at different amounts (5, 10, and 20 g) to 100 mL of waste engine oil at various temperatures (250, 350, 400, and 450 °C) and mixed at a speed of 800 rpm for 30 min. After settling and separation, the treated oil was diluted with kerosene before being examined using a Ultraviolet–visible (UV) spectrophotometer. In order to achieve cost-effective recycling, this process is modeled using the response surface method (RSM). Five regression models (linear, quadratic, Two Factor Interactions (2FI), cubic, and reduced-order quadratic model) were developed, then tested, and examined by calculating the statistical performance indicators (R2, R2adj, Akaike’s Information Criterion corrected (AICc), Bayesian Information Criterion (BIC), and Root Mean Square Error (RMSE)). The results obtained reveal that the modified quadratic model outperforms the rest of the models in terms of the low value of RMSE, the lowest AICc, lowest BIC, and the highest R2 and R2adj. The developed modified quadratic model is optimized successfully to predict optimum operation conditions. Results show that optimum operation conditions are at the minimum area under the curve for UV absorption at 223.358; this can be achieved with a process temperature of 266.246 °C and clay quantity of 5.331 g. This model agreed with experimental data regardless of the effectiveness of red clay in the therapy of lube oil.

Relative effects of breastfeeding intention and practice on maternal responsiveness.

ABSTRACTOur objective was to examine the differential effects of antenatal breastfeeding intention (BI) and breastfeeding practice (BP) on maternal postnatal responsiveness. We conducted a secondary analysis of longitudinal data from a subsample of 962 mother–infant dyads from a U.K.‐based birth cohort study the Avon Longitudinal Study of Parents and Children. Exposures were BI and BPs measured at 32 weeks of gestation and 18 months’ postpartum. The outcome was maternal responsiveness assessed at 12 months’ postpartum. We used logistic regression analyses unadjusted and adjusted for confounders. Intention to breastfeed was associated with increased odds of postnatal maternal responsiveness independent of BP, adjusted odds ratio (OR) = 2.34, 95% CI [1.42, 3.86]. There was no evidence that BP was an independent predictor of maternal responsiveness, OR = 0.93, 95% CI [0.55, 1.57]. Life‐course epidemiology analyses demonstrated that maternal responsiveness is most positive when both BI and BP are present. This is the first population‐based study to provide evidence that BI during pregnancy is more strongly associated with maternal postnatal responsiveness than is BP. Further research is needed to understand the determinants of BI in pregnancy and its relationships with maternal responsiveness.

Evaluating the most suitable nonlinear growth model for turbot (Scophthalmus maximus) in aquaculture 2 (weight application): Multi‐criteria model selection and growth prediction

Seeking the most suitable model to describe the growth of turbot, we analysed growth data of two different turbot (Scophthalmus maximus) strains reared communally in a recirculating aquaculture system. We fitted 10 different nonlinear growth models to individual weight gain data (n = 2,010) during the grow-out phase. Analyses were carried out for each strain, for sexes within strains and for a pooled data set containing both strains and sexes. To assess the model performance, three different criteria are used. Further, a growth-simulation was performed to evaluate the shape of the generated curve. This way we could assess the capability of the models to predict future growth. The 3-parametric Gompertz model achieved the best fit in 42.9% of all cases tested and the lowest Bayesian information criterion in 100% of cases. The model produced realistically shaped curves and asymptotic values matching the biological attributes of the species. In contrast, 5-parametric functions projected unrealistically shaped curves and predicted improbable mature sizes. Our results show that increasing number of parameters do not lead to increasing goodness of fit, but tend to result in overfitting, and demonstrate the advantages of the 3-parametric Gompertz model for describing the growth of turbot.

LO75: The impact of snowfall on patient attendance at an urban academic emergency department

Introduction: Accurate forecasting of emergency department (ED) patient visits can inform better resource matching. Calendar variables such as day of week and time of day are routinely used as predictors of ED volume. Further improvement in forecasting will likely come from dynamic variables. The effect of snowfall on ED volumes in colder climates remains poorly understood. We sought to determine whether accounting for snowfall improves ED patient volume forecasting. Our secondary objective was to characterize the magnitude of effect of snowfall on ED volume. Methods: This was a retrospective observational study using historical patient volume data and local snowfall records from April 1st, 2011 to March 31st, 2018 (2,542 days) at a single urban ED. We fit a series of four generalized linear models: a baseline model which included calendar variables and three different snowfall models which contained the variables in the baseline model plus an indicator variable for modelling snowfall. Each snowfall model had a different daily threshold for its indicator variable: any snowfall ( &amp;gt;0cm), moderate snowfall ( &amp;gt; = 1 cm), or high snowfall ( &amp;gt; = 5 cm). We modeled daily ED volume as the dependent variable using a Poisson distribution. To evaluate model fit, we examined the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) in each of the four models. In both cases, a lower number indicates better model fit. Incident rate ratios were calculated to determine the effect of snowfall. We used the delta method to calculate confidence intervals. Results: A total of 2542 days were used to develop the model. All three snowfall models demonstrated improved model fit compared to the baseline model with lower AIC and BIC values. The best fitting model included a binary variable for moderate snowfall ( &amp;gt; = 1cm/day). This model showed a statistically significant decrease in ED volume of 2.65% (95% CI: 1.23% -4.00%) on snowfall days, representing 5.4 (95% CI: 2.5 -8.2) patients per day at our hospital with an average daily volume of 205 patients. Conclusion: The addition of a snowfall variable results in improved forecasting model performance in ED volume forecasting with optimal threshold set at 1 cm of snow in our setting. Snowfall is associated with a modest, but statistically significant reduction in ED volume.

Identifying Optimal Factor Scores on the Bereaved Family Survey: Implications for Practice and Policy

ContextThe Bereaved Family Survey (BFS) is used to evaluate the quality of end-of-life (EOL) care in Veterans Affairs inpatient settings. The BFS consists of a global Performance Measure (BFS-PM) and three factors that relate to specific aspects of EOL care. ObjectiveThe purpose of this study was to identify empirically based target scores on each BFS factor that are most strongly related to a rating of “excellent” on the BFS-PM. MethodsWe conducted a cross-sectional analysis of BFS and Veteran clinical data from January 2012 to January 2016. Logistic regression models were constructed for each potential cut point on the three BFS factors and accounted for facility case-mix and nonresponse bias. Model fit was assessed primarily using the Liu Index, Bayesian Information Criterion (BIC), and classification accuracy values. ResultsOur analytic sample included 40,180 Veterans whose next-of-kin completed a BFS. The mean BFS response rate across study years was 58%. A score of 14 or higher on the Respectful Care and Communication factor (range 0–15) had the lowest BIC (121355) and highest percent correctly classified (81.2%). The Emotional and Spiritual Support factor (range 0–9) had an optimal score of 8 or higher (BIC = 133685; % correctly classified = 77.1%). An optimal cut point on the Benefits factor was not identified. ConclusionThe identification of data-driven targets makes BFS factor scores more useful to clinicians and administrators focused on improving quality of EOL care in their facilities. Our results lend support for prioritizing quality improvement efforts related to respectful care and communication.