Collinear Variables Research Articles

Non-Destructive Identification of Wool and Cashmere Fibers Based on Cascade Optimizations of Interval-Wavelength Selection Using NIR Spectroscopy

ABSTRACT Near-infrared (NIR) spectroscopy is an effective method for identifying wool and cashmere fibers, with high spectral data providing a wealth of information. However, a key issue is that the accuracy and robustness of subsequent estimates can be reduced by redundant and interfering wavelengths. For this reason, a novel interval-wavelength cascaded optimization method is proposed. Initially, the collected spectral data are preprocessed by standard normal variate transformation (SNV) to eliminate the scattering effect. Then, the backward interval partial least squares (BiPLS) algorithm is applied for the preliminary selection of spectral intervals, followed by the application of three different variable selection algorithms, competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA) and whale optimization algorithm (WOA), for secondary wavelength optimization, respectively. Finally, both support vector machine (SVM) and random forest (RF) discriminant models are built to identify the extracted subset of wavelengths. In the experimental stage, the cascade method BiPLS-WOA selects 36 wavelengths, in SVM, the accuracy of the validation set reaches 96.9%, and the area under the ROC curve (AUC) can reach 99.3%. The results demonstrate that the proposed method can eliminate redundant and collinear variables, thereby validating the effectiveness of distinguishing wool and cashmere fibers.

Sleep Health among Community-Recruited Opioid-Using People Who Inject Drugs in Los Angeles, CA and Denver, CO.

Chronic insufficient and poor-quality sleep are linked to hypertension, diabetes, depression, heart attack, and stroke. While studies on substance use and sleep typically focus on people in or entering treatment, there is a lack of research on sleep health among community-recruited people who inject drugs (PWID). To address this literature gap, we examined factors associated with insufficient and poor-quality sleep among community-recruited PWID. We recruited and interviewed 472 active opioid-using PWID (injected within the last 30days) in Los Angeles, CA and Denver, CO between 2021 and 2022. Participants completed computer-assisted interviews covering demographics, subsistence measures, drug use patterns, injection-related behaviors, health risks, and sleep duration and quality in the last 3months. Descriptive statistics were used to analyze all variables for subjects with complete responses to sleep items (n = 464). Bivariate analyses determined factors associated with sleep measures using chi-square and t-tests. Collinear variables were removed, and binomial linear multivariable regression calculated risk ratios (RR) for insufficient and poor-quality sleep in the last 3months. Participants exhibited low sleep duration (mean = 4.99, standard deviation (SD) = 2.70), with 76% reporting insufficient sleep and 62% poor-quality sleep. Bivariate analyses associated both sleep measures with drug use, high subsistence scores, violent victimization, and poor health outcomes. Multivariable analyses showed a high subsistence score predicting insufficient (RR = 1.31) and poor-quality sleep (RR = 1.69) compared to low subsistence. Poor sleep health is common among structurally vulnerablecommunity-recruited PWID, asmeasured by subsistence index associated with adverse sleep outcomes. Further research on structural interventions to address sleep and subsequent health outcomes among PWID is imperative.

Construction and validation of a risk prediction model for extrauterine growth restriction in preterm infants born at gestational age less than 34 weeks.

This study aimed to develop and validate a model for predicting extrauterine growth restriction (EUGR) in preterm infants born ≤34 weeks gestation. Preterm infants from Guangxi Maternal and Child Health Hospital (2019-2021) were randomly divided into training (80%) and testing (20%) sets. Collinear clinical variables were excluded using Pearson correlation coefficients. Predictive factors were identified using Lasso regression. Random forest (RF), support vector machine (SVM), and logistic regression (LR) models were then built and evaluated using the confusion matrix, area under the curve (AUC), and the F1 score. Additionally, calibration curves and decision curve analysis (DCA) were plotted to assess the performance and practical utility of the models. The study included 387 infants, with no significant baseline differences between training (n = 310) and testing (n = 77) sets. LR identified gestational age, birth weight, premature rupture of membranes, patent ductus arteriosus, cholestasis, and neonatal sepsis as key EUGR predictors. The RF model (19 variables) demonstrated an accuracy of greater than 90% during training, and superior AUC (0.62), F1 score (0.80), and accuracy (0.72) in testing compared to other models. Gestational age, birth weight, premature rupture of membranes, patent ductus arteriosus, cholestasis, and neonatal sepsis are significant EUGR predictors in preterm infants ≤34 weeks. The model shows promise for early EUGR prediction in clinical practice, potentially enhancing screening efficiency and accuracy, thus saving medical resources.

Forest above-ground biomass estimation based on strongly collinear variables derived from airborne laser scanning data

Airborne laser scanning technique (ALS) is the most appealing remote sensing technique for the precise estimation of forest above-ground biomass (AGB). Significantly strong correlations (collinearity) among the independent variables derived from ALS data decrease the accuracy of developed AGB models. To address this issue, we propose a novel variable selection algorithm-an improved sure independence screening (SPV), which integrates the Pearson correlation coefficient, threshold μ, and variance inflation factor. We further compared the performance of SPV-based, stepwise feature selection (SFS)-based, and least absolute shrinkage and selection operator (LASSO)-based AGB models developed with different regression approaches that were sensitive to strong collinearity. Field-measured data and corresponding ALS data, acquired from 1002 sample plots distributed across four distinct forest types within the Guangxi Zhuang Autonomous Region in Southern China, were used to evaluate variable selection techniques and develop AGB models. Results indicated that ALS variables selected by SPV exhibited weaker collinearity compared to those selected by SFS and LASSO. SPV-based AGB models outperformed SFS-based AGB models with higher leave-one-out cross-validation adjusted R2 (LOOCV Radj2) by 0.1% − 27.8%. SPV-based AGB models outperformed LASSO-based AGB models with higher LOOCV Radj2 by 0.4% − 16.3%. Hence, for variable selection in constructing AGB models (linear regression model, log–log regression model, and generalized additive model (GAM)) based on strongly collinear ALS variables, SPV is most preferred, followed by SFS and LASSO. The smooth curves from our GAMs developed using SPV-selected variables revealed that five canopy height variables (whp40, hmean, hp25, hp70, hskew), one canopy density variable (whp85), three density-related variables (nVegPts, bcc, and mcc), and one vertical structural variable (LADmean) were positively correlated with AGB. The canopy height variables (whp40, hmean, hp25, and hp70) were identified as the most important variables in estimating AGB for four forest types. The canopy density variable wdp85 showed a strong effect on AGB of the coniferous forests, whereas it had almost no effect on the AGB of broadleaved forests. Overall, this manuscript proposes a novel variable selection algorithm named SPV, aimed at addressing collinearity of variables derived from ALS data, which has significant implications for the application of ALS in forest inventory and forest modeling.

Development and validation of a TAAbs and TAAs based non-invasive model for diagnosing lung cancer

BackgroundSingle Tumor-associated autoantibodies (TAAbs) and tumor-associated antigens (TAAs) have been found to have lower diagnostic efficacy in lung cancer. Our objective is to develop and validate a lung cancer prediction model that utilizes TAAbs and TAAs and to enhance the accuracy of lung cancer detection. Methods1830 subjects were randomly divided into training and validation sets at a 7:3 ratio for this study. Lasso regression analysis was used to remove collinear variables, whereas univariate logistic regression analysis was employed to identify potential independent risk factors for lung cancer. A diagnostic model was constructed using multivariate logistic analysis. The results were presented as a nomogram and assessed for various performance measures, including area under the curve, calibration curve, and decision curve analysis. ResultsThe diagnostic model was developed using gender, age, GAGE7, MAGE-A1, CA125, and CEA as variables. The training set had an AUC of 0.787, while the validation set had an AUC of 0.750. The calibration curves of the training and validation sets showed a strong agreement between anticipated and observed values. The nomogram performed better than any individual variable in both the training and validation sets in terms of net benefits for lung cancer detection, according to DCA analysis. ConclusionsThis study proposes a diagnostic model for lung cancer that uses TAAbs and TAAs and incorporates individual characteristics. This model can be easily applied to personalized diagnosis.

Developing and validating a clinlabomics-based machine-learning model for early detection of retinal detachment in patients with high myopia.

Retinal detachment (RD) is a vision-threatening disorder of significant severity. Individuals with high myopia (HM) face a 2 to 6 times higher risk of developing RD compared to non-myopes. The timely identification of high myopia-related retinal detachment (HMRD) is crucial for effective treatment and prevention of additional vision impairment. Consequently, our objective was to streamline and validate a machine-learning model based on clinical laboratory omics (clinlabomics) for the early detection of RD in HM patients. We extracted clinlabomics data from the electronic health records for 24,440 HM and 5607 HMRD between 2015 and 2022. Lasso regression analysis assessed fifty-nine variables, excluding collinear variables (variance inflation factor > 10). Four models based on random forest, gradient boosting machine (GBM), generalized linear model, and Deep Learning Model were trained for HMRD diagnosis and employed for internal validation. An external test of the models was done. Three random data sets were further processed to validate the performance of the diagnostic model. The primary outcomes were the area under the receiver operating characteristic curve (AUC) and the area under the precision-recall curve (AUCPR) to diagnose HMRD. Nine variables were selected by all models. Given the AUC and AUCPR values across the different sets, the GBM model was chosen as the final diagnostic model. The GBM model had an AUC of 0.8550 (95%CI = 0.8322-0.8967) and an AUCPR of 0.5584 (95%CI = 0.5250-0.5879) in the training set. The AUC and AUCPR in the internal validation were 0.8405 (95%CI = 0.8060-0.8966) and 0.5355 (95%CI = 0.4988-0.5732). During the external test evaluation, it reached an AUC of 0.7579 (95%CI = 0.7340-0.7840) and an AUCPR of 0.5587 (95%CI = 0.5345-0.5880). A similar discriminative capacity was observed in the three random data sets. The GBM model was well-calibrated across all the sets. The GBM-RD model was implemented into a web application that provides risk prediction for HM individuals. GBM algorithms based on nine features successfully predicted the diagnosis of RD in patients with HM, which will help ophthalmologists to establish a preliminary diagnosis and to improve diagnostic accuracy in the clinic.

Patterns of democracy and democratic satisfaction: Results from a comparative conjoint experiment

AbstractThis study presents the results from a series of conjoint experiments evaluating how the various components of political systems affect citizens’ satisfaction with democracy (SWD). Compared to earlier studies, our approach is unique in that we were able to disentangle the effects of otherwise highly collinear variables corresponding to three defining features of any political system, namely (1) access to power, (2) the policy‐making process and (3) performance. We fielded identical conjoint experiments in countries with significant variations across the key independent variables: France, Hungary and Italy. Overall, our study supports the view that citizens are more satisfied with democracy in consensus systems than in majoritarian democracies, while it also contributes to identifying the respective weight of the specific components of political systems. Respondents across all countries recognize that proportionality and party system fragmentation magnify their voices by creating a more representative political climate. With regard to output legitimacy, we find that respondents identify good economic performance and public probity as important features of a well‐performing democracy. The positive effect of the legislators’ constituency orientation on SWD is a particularly noteworthy result that is currently understudied in the literature. Moreover, the state of the economy has by far the biggest impact on SWD, which indicates a highly materialistic view of democracy in all three countries.

Transformation of Human Capital as a Driver of Innovative Economy

The purpose of this study was to identify key factors that have a significant impact on the level of high-tech production in the region and the development of human capital in seven countries of the Caucasus and Central Asia. A comprehensive literature analysis was carried out to achieve this goal, including data from several countries of the Caucasus and Central Asia, including demographic data, education levels, labor market indicators, health indicators, and research activities. The authors constructed a model of multiple spatial regression, including seven spatial objects and a 5-time series. A qualitative model of cumulative panel regression without signs of multicollinearity was obtained as a result of the step-by-step elimination of collinear and nonessential variables. The results of the study revealed three key factors that are closely and directly related to the level of production of high-tech products. Two relate to the education sector – "the number of university graduates in the field of science and technology", and "the number of researchers per 1 million inhabitants" and one indicator characterizes the labor market – "the employment level in the economy". The authors did not identify a close relationship between the level of high-tech production and economic factors, such as cost and productivity, which contradicts the results of other research and requires more careful study. The results of the research can be useful in the development of measures to promote human capital and innovation policy of the countries of the Caucasus and Central Asia.
 
 Received: 7 September 2023 / Accepted: 19 October 2023 / Published: 5 November 2023

Multicollinearity in Multiple Linear Regression: Detection, Consequences, and Remedies

Abstract: Multiple linear regression is a widely used statistical tool for modeling relationships between a dependent variable and multiple explanatory variables. However, it assumes that these explanatory variables are independent, which is not always the case in practical scenarios, leading to a phenomenon known as multicollinearity. Multicollinearity occurs when explanatory variables in a regression model are strongly correlated with each other, causing several issues in regression analysis. This paper discusses the detection and remedies for multicollinearity in detail. Detection methods include examining the determinant of the correlation matrix, inspecting correlation coefficients, using partial regression coefficients, calculating Variance Inflation Factors (VIFs), and assessing the condition number and condition index. These techniques help researchers identify the presence and severity of multicollinearity in their dataset. To address multicollinearity, several remedies are proposed, including obtaining more data, dropping collinear variables, using relevant prior information, employing generalized inverses, and employing principal component regression. Ridge regression, which introduces bias to reduce variance, is also discussed as an effective technique to combat multicollinearity. Understanding multicollinearity and employing appropriate detection and remediation strategies is crucial for obtaining reliable and meaningful results from multiple linear regression models

#4784 IDENTIFICATION OF FACTORS ASSOCIATED WITH DEATH IN DIALYSIS PATIENTS USING A MACHINE LEARNING-BASED PREDICTIVE MODEL

Abstract Background and Aims Few studies including unplanned dialysis starts have used machine learning for the prediction of death in dialysis patients. The objective of the study was to use R software algorithms to develop machine learning predictive models for the identification of death-related factors in patients undergoing hemodialysis (HD) and peritoneal dialysis (PD). Method This study included adult patients undergoing HD and PD started in a planned or urgent manner in a dialysis center between January 2014 and January 2019. Epidemiological, clinical and laboratory data were collected. Univariate analysis was followed by ML-based analyses. Then, multivariate regressions were obtained using stepwise and Cox regression analyses. Finally, a Random Forest predictive model was generated after variables with missing values >30% were removed. Results Of 581 patients included, 170 died (29,2%). On univariate analysis death was associated with age, number of comorbidities, dialysis modality switching, creatinine, PTH and albumin values at dialysis initiation, presence of diabetes (DM), hospitalization, function recovery and central venous catheter (CVC) for dialysis access. Patients who started dialysis with a CVC had a worse survival (p = 0.0034) than those who did not use CVC, started HD with AVF, or received PD (Figure 1). Data were split into 20% for testing the regression model, and 80% for training the model. Data preprocessing for Cox regression included imputing some values using bag impute (decision trees), creating dummy variables, and removing collinear variables. Death was associated with older age (p < 0.001), fewer ESI-free months (p < 0.001) and lower initial creatinine (p = 0.008) (Table 1). The model C-index was 0.8099. Random forest ranked the following variables predictive of death in descending order of importance: ESI-free months; age; initial levels of creatinine, PTH and albumin; number of comorbidities; dialysis-related infection; initial phosphorus and hemoglobin; hospitalizations; male gender; modality switching (Figure 2). The agreement of the model obtained was 0.8110. Conclusion ESI-free months, age and initial levels of creatinine were associated with death on both multivariate and ML-based analyses.

Development and multi-center validation of machine learning model for early detection of fungal keratitis

Fungal keratitis (FK) is a leading cause of corneal blindness in developing countries due to poor clinical recognition and laboratory identification. Here, we aimed to identify the distinct clinical signature of FK and develop a diagnostic model to differentiate FK from other types of infectious keratitis. We reviewed the electronic health records (EHRs) of all patients with suspected infectious keratitis in Beijing Tongren Hospital from January 2011 to December 2021. Twelve clinical signs of slit-lamp images were assessed by Lasso regression analysis and collinear variables were excluded. Three models based on binary logistic regression, random forest classification, and decision tree classification were trained for FK diagnosis and employed for internal validation. Independent external validation of the models was performed in a cohort of 420patients from seven different ophthalmic centers to evaluate the accuracy, specificity, and sensitivity in real world. Three diagnostic models of FK based on binary logistic regression, random forest classification, and decision tree classification were established and internal validation were achieved with the mean AUC of 0.916, 0.920, and 0.859, respectively. The models were well-calibrated by external validation using a prospective cohort including 210FK and 210 non-FK patients from seven eye centers across China. The diagnostic model with the binary logistic regression algorithm classified the external validation dataset with a sensitivity of 0.907 (0.774, 1.000), specificity 0.899 (0.750, 1.000), accuracy 0.905 (0.805, 1.000), and AUC 0.903 (0.808, 0.998). Our model enables rapid identification of FK, which will help ophthalmologists to establish a preliminary diagnosis and to improve the diagnostic accuracy in clinic. The Open Research Fund from the National Key Research and Development Program of China (2021YFC2301000) and the Open Research Fund from Beijing Advanced Innovation Center for Big Data-Based Precision Medicine, Beijing Tongren Hospital, Beihang University &Capital Medical University (BHTR-KFJJ-202001) supported this study.

P758 Impact of co-medication on rates of endoscopic remission: post-hoc analysis of randomised non-inferiority trial of 1600 mg versus 400 mg tablets of mesalazine for the treatment of mild-to-moderate ulcerative colitis

Abstract Background Some studies reported that anti-TNF treatment in biologic-naïve elderly onset inflammatory bowel disease patients may be less effective than in non-elderly onset patients (Amano et al. Scientic Reports. 2022 12:5324). We assessed known factors as well as presence of co-morbidities and polypharmacy associated with endoscopic remission in mild-to-moderate (Mayo clinic score ≥5) adult ulcerative colitis (UC) patients treated with mesalazine. Methods In a post-hoc analysis of data from a phase 3 non-inferiority trial ( D’Haens et al. Aliment Pharmacol Ther.2017;46:292–302)we examined the strength of association between number of co-medications and comorbidities by calculating Spearman’s rho. We used multivariable logistic regression modeling to determine the associations between endoscopic remission as outcome (yes vs. no; defined as either Mayo endoscopic subscore of 0 or ≤1 according to blinded central reading) and independent variables (disease duration, Mayo endoscopic score at baseline, leucocyte concentration at baseline, Robarts Histopathologic Index at baseline, age [≤60 vs. >60 years of age], gender, comedications at the time of the trial [none, ≤2, ≥3], comorbidities [none, 1-2, 3-5, >5], body mass index, disease extent prior to the trial, smoking status [current, ex, and never] at week 8 and 38. Collinear variables were examined in separate models. P-value of <0.05 was considered to be statisically significant. Results We analyzed data on 774 patients (330 female, 43.5±14.3 [SD] years of age, median disease duration of 3.00 years [IQR 0.56; 7.88], 175 and 92 persons with 1-2 and ≥3 comedications, respectively) and 679 patients (289 female, 43.7±14.5 [SD] years of age, median disease duration of 2.99 years [0.57; 7.62]; 150 and 85 persons with 1-2 and ≥3 comedications, respectively) that had colonoscopy/sigmoidoscopy at 8 weeks of double-blind induction and 38 weeks open-label maintenance phases of the study, respectively. We observed a moderate association between number of comorbidities and comedications (week 8: Spearman rho=0.545, P-value of < 0.001; week 38: Spearman rho=0.512, P-value < 0.001). Patients with baseline Mayo clinic endoscopic score of 3 and increased baseline leucocyte concentration, were less likely to be in endoscopic remission irrespective of endoscopic remission definition and time post treatment (Table 1). In addition, patients with ≥3 co-medications were less likely to be in endoscopic remission of 1 at week 8 and 38 and in endoscopic remission of ≤1 at week 38 (Table 1). Conclusion The increased number of co-medications rather than the actual age may be responsible for decreased effectiveness of medications, including mesalazine, in elderly UC patients highlighting the importance of healthy aging.

The effect of governance on entrepreneurship: from all income economies perspective

The purpose of this study is to analyze the effect of governance indicators on Entrepreneurship. Explanatory research design with Pearson correlation and multiple linear regression models were applied. Five-year World Bank data (2014–2018) of 126 countries from all economic development levels were used. Worldwide governance indicators considered are voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and corruption control. Gross net income was taken as a control variable. To measure entrepreneurship, the number of formally registered limited liability businesses as a percentage of the working-age population, was used. To make highly skewed time series data of dependent variable (entrepreneurship) closer to normal, logarithmic transformation was made and heteroscedasticity of residuals was checked. The finding of Pearson correlation shows that there are moderate to strong significant correlations (0.466 ≤ r ≤ 0.806, p < 0.01) between predictors and the outcome variable. Predictor variables have moderate to very strong correlation among each other (0.506 ≤ r ≤ 0.956, p < 0.01). Regression analysis was computed after two highly collinear variables were dropped from the model using the variance inflation factor (VIF) test. The study found that the remaining four independent variables and the control variable predict 71.5% of the variance in the outcome variable. Except for voice and accountability, all predictors have their own statistically significant influence on entrepreneurship. Thus, working on each predictor up to the standard application can bring incremental changes in new business formation and entry. The researchers believe that this study is of significant interest to policymakers, program developers, entrepreneurs, analysis, and supporters, since it provides useful insight on how governance indicators influence entrepreneurship.

Pain-related fear of movement dynamics in individuals with and without low back pain participating in weightlifting and/or powerlifting training.

Pain-free adults in the general population have been shown to possess unhelpful beliefs that certain movements and postures are harmful to the spine, potentially reinforcing fear-avoidance behaviour. Whether such beliefs occur in individuals undertaking regular powerlifting (PL) and Olympic weightlifting (OWL) training is unclear. In a cross-sectional study design, 67 individuals who participate in OWL and PL training completed an online survey. Demographic characteristics, training history, and self-reported perceptions of harm, on the 40-item Photograph Series of Daily Activities shortened electronic version (PHODA-SeV), were collected. After removing collinear variables, 13 items were entered into a network analysis, in which the adjusted correlations between items, and the centrality indices of each item (i.e., the degree of connection with other symptoms in the network) were quantified. Twenty-one (31.3%) participants had LBP symptoms. The pairwise correlations with the greatest magnitudes were between images of 'leg stretch' and 'jumping' (0.32 [95%CI 0.08 to 0.45]) and two images depicting ironing (0.32 [95%CI 0.05 to 0.54]) respectively. The three most Central (connected) items were 'stair ascend', 'walking with groceries', and 'mopping with spine flexion'. For individuals training in OWL and PL, images reflecting walking, rather than those depicting high spinal flexion angle, had greater connectivity to other activity items. In addition, the strongest correlations were not between items reflecting high spinal flexion angle. Future studies that investigate the relationship between different intensities of OWL and PL training and the dynamics of pain-related fear are warranted.

Prevalence and Factors Associated with Hypertension among Community-Dwelling Adolescents in a Rural Area of Haryana, India, Using AAP 2017 and NHBPEP Criteria.

To estimate the prevalence and determinants of hypertension among adolescents in a rural area in North India. This was a cross-sectional study conducted among community-dwelling adolescents (10-19 y) and defined blood pressure as per the National High Blood Pressure Education Program (NHBPEP) and American Academy of Pediatrics (AAP) 2017 criteria. A pretested, semistructured questionnaire was used to elicit sociodemographic details and risk factors of hypertension. Association of factors with hypertension, z scores of systolic blood pressure (SBP) and diastolic blood pressure (DBP) were assessed by logistic and linear regression. In multivariable regression model, for both logistic and linear regression, variables with p value < 0.2 were included, excluding collinear variables. Eight hundred sixty-four adolescents participated with mean (SD) age, BMI of 14.0 ± 2.6 y, 18.3 ± 3.5kg/m2. Hypertension prevalence was 22.5% (95% CI: 19.7, 25.5) as per AAP and 15.2% (95% CI: 12.9, 17.8) as per NHBPEP criteria. Being hypertensive as per the AAP was associated with family history of noncommunicable disease (NCD) (adjusted odds ratio, AOR 1.80, 95% CI: 1.11, 2.86) and BMI (AOR = 1.09, 95% CI: 1.04, 1.14), and as per the NHBPEP, was only associated with BMI (AOR = 1.11, 95% CI: 1.05, 1.17). SBP z score was significantly associated with male sex (adjusted beta coefficient AC = 0.79, 95% CI: 0.69, 0.90), and weight z scores (AC = 2.00, 95% CI: 1.22, 3.28); DBP z score was significantly associated with z scores of height (AC = 0.70, 95% CI: 0.51, 0.96) and waist (AC = 1.15, 95% CI: 1.02, 1.30). The prevalence of hypertension among adolescents was high and was associated with BMI and family history of NCD.

An Intelligent Time Series Model Based on Hybrid Methodology for Forecasting Concentrations of Significant Air Pollutants

Rapid industrialization and urban development are the main causes of air pollution, leading to daily air quality and health problems. To find significant pollutants and forecast their concentrations, in this study, we used a hybrid methodology, including integrated variable selection, autoregressive distributed lag, and deleted multiple collinear variables to reduce variables, and then applied six intelligent time series models to forecast the concentrations of the top three pollution sources. We collected two air quality datasets from traffic and industrial monitoring stations and weather data to analyze and compare their results. The results show that a random forest based on selected key variables has better classification metrics (accuracy, AUC, recall, precision, and F1). After deleting the collinearity of the independent variables and adding the lag periods using the autoregressive distributed lag model, the intelligent time-series support vector regression was found to have better forecasting performance (RMSE and MAE). Finally, the research results could be used as a reference by all relevant stakeholders and help respond to poor air quality.

Modelling potential distribution of an endemic plant (Ferula Microcolea) based on bioclimatic variables in Iran

Ferula microcolea (Boiss.) Boiss. is an endemic plant in Iran that some of its habitats have been destroyed in recent decades. Since the bioclimatic variables which determine its potential distribution, are poorly defined, a specific analysis is needed. In this study, the species distribution modelling was used for reaching these goals: (i) identifying the bioclimatic factors that constrain the distribution of this species in Iran, (ii) generating a potential habitat suitability map for F. microcolea using Maxent (iii) determining the high suitable areas where this species could be present (iv) evaluating the final model. In all, 66 records of F. microcolea in Iran were used as the occurrence data. Nineteen bioclimatic variables were obtained from the WorldClim database and collinear variables were removed in a sequential manner with regard to the ecological knowledge of the plant. The maxent parameters were optimised with ENMeval R package. For evaluating the performance of the Maxent model, the Area under curve value (AUC) was calculated. The results showed that the model performance was excellent. Analysis of variable contribution demonstrated that the distribution of this species is most influenced by the Annual Mean Temperature. We revealed that the area about 22,005.5 km2 is highly suitable for F. microcolea that is principally located in Chaharmahal and Bakhtiari province. Although this region is rich in biodiversity, greater focus should be paid to its conservation. Our findings provide a scientific basis for the habitats conservation of this species in Iran.

A greedy regression algorithm with coarse weights offers novel advantages

Regularized regression analysis is a mature analytic approach to identify weighted sums of variables predicting outcomes. We present a novel Coarse Approximation Linear Function (CALF) to frugally select important predictors and build simple but powerful predictive models. CALF is a linear regression strategy applied to normalized data that uses nonzero weights + 1 or − 1. Qualitative (linearly invariant) metrics to be optimized can be (for binary response) Welch (Student) t-test p-value or area under curve (AUC) of receiver operating characteristic, or (for real response) Pearson correlation. Predictor weighting is critically important when developing risk prediction models. While counterintuitive, it is a fact that qualitative metrics can favor CALF with ± 1 weights over algorithms producing real number weights. Moreover, while regression methods may be expected to change most or all weight values upon even small changes in input data (e.g., discarding a single subject of hundreds) CALF weights generally do not so change. Similarly, some regression methods applied to collinear or nearly collinear variables yield unpredictable magnitude or the direction (in p-space) of the weights as a vector. In contrast, with CALF if some predictors are linearly dependent or nearly so, CALF simply chooses at most one (the most informative, if any) and ignores the others, thus avoiding the inclusion of two or more collinear variables in the model.

P494 A serum metabolite panel predicts Crohn’s disease relapse in patients discontinuing infliximab and continuing antimetabolite therapy: sub-analysis of the SPARE Trial

Abstract Background Combination therapy with infliximab and anti-metabolites is an effective strategy for maintaining remission in patients with Crohn’s disease (CD); however, the implications of long-term combination therapy has led clinicians to consider discontinuing one of the two agents after sustained remission has been achieved. The SPARE investigators have addressed this in an unblinded clinical trial in which 211 patients were randomised to either continuation of combination therapy, withdrawal of infliximab, or withdrawal of antimetabolite therapy. Here, we aimed to determine whether nuclear magnetic resonance (NMR)-based metabolomics analysis of patient sera at baseline could determine which patients randomised to infliximab withdrawal would relapse over 2-years on anti-metabolite monotherapy. Methods The study cohort comprised patients randomised to infliximab cessation who had a serum aliquot available for analysis (n=63). All patients had been treated with a combination therapy of infliximab (IFX) and anti-metabolite &amp;gt;8 months and had been in sustained steroid-free remission &amp;gt;6 months. Disease relapse was defined by CDAI&amp;gt;250 (or between 150–250 with 70 points increase over two consecutive weeks) alongside serum CRP and faecal calprotectin levels. Untargeted metabolomic profiling was carried out using NMR spectroscopy. Key metabolites were identified by their variable importance in projection (VIP) score using orthogonal partial least squares discriminant analysis (OPLS-DA) with 10-fold cross-validation on independent data. Multiple logistic regression on the metabolites with highest VIP score was used to report a receiver operator characteristic (ROC) curve integral (AUC) value after collinear variables were removed. Results Of 63 patients, 21 (33%) relapsed. The mean time to relapse after infliximab discontinuation was 7.9 months. Four independent (multicollinearity R2&amp;lt;0.4) serum metabolites (alanine, glutamine, acetoacetate, glucose) identified relapse with an AUC of 0.84 (95% confidence interval 0.74–0.94, p&amp;lt;0.0001). The positive predictive value was 66.7%, and the negative predictive value was 77.1%. All serum metabolites identified were independent of baseline CRP level (p&amp;gt;0.05, Pearson correlation). Patients who relapsed following infliximab discontinuation had significantly elevated serum acetoacetate and decreased alanine and glutamine levels (post-hoc students t-test, p&amp;lt;0.05) at baseline. Conclusion Here, we identify a panel of serum metabolites independent of baseline serum CRP, which predict high risk of disease relapse after infliximab discontinuation. These data may help in stratifying patients suitable for drug withdrawal, together with other clinical and multi-omic parameters.

Risk factors for unchanged ventricles during pediatric shunt malfunction.

Children whose ventricles do not change during shunt malfunction present a diagnostic dilemma. This study was performed to identify risk factors for unchanged ventricular size at shunt malfunction. This retrospective 1:1 age-matched case-control study identified children with shunted hydrocephalus who underwent shunt revision with intraoperative evidence of malfunction at one of the three participating institutions from 1997 to 2019. Cases were defined as patients with a change of < 0.05 in the frontal-occipital horn ratio (FOR) between malfunction and baseline, and controls included patients with FOR changes ≥ 0.05. The presence of infection, abdominal pseudocyst, pseudomeningocele, or wound drainage and lack of baseline cranial imaging at the time of malfunction warranted exclusion. Of 450 included patients, 60% were male, 73% were Caucasian, and 67% had an occipital shunt. The median age was 4.3 (IQR 0.97-9.21) years at malfunction. On univariable analysis, unchanged ventricles at malfunction were associated with a frontal shunt (41% vs 28%, p < 0.001), programmable valve (17% vs 9%, p = 0.011), nonsiphoning shunt (85% vs 66%, p < 0.001), larger baseline FOR (0.44 ± 0.12 vs 0.38 ± 0.11, p < 0.001), no prior shunt infection (87% vs 76%, p = 0.003), and no prior shunt revisions (68% vs 52%, p < 0.001). On multivariable analysis with collinear variables removed, patients with a frontal shunt (OR 1.67, 95% CI 1.08-2.70, p = 0.037), programmable valve (OR 2.63, 95% CI 1.32-5.26, p = 0.007), nonsiphoning shunt at malfunction (OR 2.76, 95% CI 1.63-4.67, p < 0.001), larger baseline FOR (OR 3.13, 95% CI 2.21-4.43, p < 0.001), and no prior shunt infection (OR 2.34, 95% CI 1.27-4.30, p = 0.007) were more likely to have unchanged ventricles at malfunction. In a multicenter cohort of children with shunt malfunction, those with a frontal shunt, programmable valve, nonsiphoning shunt, baseline large ventricles, and no prior shunt infection were more likely than others to have unchanged ventricles at shunt failure.