Predicting childhood blood lead levels (BLLs) has had mixed success, and it is unclear if individual- or neighborhood-level variables are most predictive. An ensemble machine learning (ML) approach to identify the most relevant predictors of BLL ≥2μg/dL in urban children was implemented. A cross-sectional sample of 603 children (~7 years of age) recruited between 2009-2019 from Montevideo, Uruguay participated in the study. 77 individual- and 32 neighborhood-level variables were used to predict BLLs ≥2μg/dL. Three ensemble learners were created: one with individual-level predictors (Ensemble-I), one with neighborhood-level predictors (Ensemble-N), and one with both (Ensemble-All). Each ensemble learner comprised four base classifiers with 50% training, 25% validation, and 25% test datasets. Predictive performance of the three ensemble models was compared using area under the curve (AUC) for the receiver operating characteristic (ROC), precision, sensitivity, and specificity on the test dataset. Ensemble-I (AUC: 0.75, precision: 0.56, sensitivity: 0.79, specificity: 0.65) performed similarly to Ensemble-All (AUC: 0.75, precision: 0.63, sensitivity: 0.79, specificity: 0.69). Ensemble-N (AUC: 0.51, precision: 0.0, sensitivity: 0.0, specificity: 0.50) severely underperformed. Year of enrollment was most important in Ensemble-I and Ensemble-All, followed by household water Pb. Three neighborhood-level variables were among the top 10 important predictors in Ensemble-All (density of bus routes, dwellings with stream/other water source and distance to nearest river). The individual-level only model performed best, although precision was improved when both neighborhood and individual-level variables were included. Future predictive models of lead exposure should consider proximal predictors (i.e., household characteristics).

Primary Sjogren's disease (pSD) (also referred to as Sjogren's syndrome) is an autoimmune disease that primarily occurs in women. In addition to exocrine gland dysfunction, pSD patients exhibit B cell hyperactivity. B cell-intrinsic TLR7 activation is integral to the pathogenesis of systemic lupus erythematosus, a disease that shares similarities with pSD. The role of TLR7-mediated B cell activation in pSD, however, remains poorly understood. We hypothesized that age-associated B cells (ABCs) were expanded in pSD and that TLR7-stimulated ABCs exhibited pathogenic features characteristic of disease. Our data revealed that ABC expansion and TLR7 expression were enhanced in a pSD mouse model in a Myd88-dependent manner. Splenocytes from pSD mice showed enhanced sensitivity to TLR7 agonism as compared with those derived from control animals. Sort-purified marginal zone B cells and ABCs from pSD mice showed enhanced inflammatory cytokine secretion and were enriched for antinuclear autoantibodies following TLR7 agonism. Finally, IgG from pSD patient sera showed elevated antinuclear autoantibodies, many of which were secreted preferentially by TLR7-stimulated murine marginal zone B cells and ABCs. These data indicate that pSD B cells are hyperresponsive to TLR7 agonism and that TLR7-activated B cells contribute to pSD through cytokine and autoantibody production. Thus, therapeutics that target TLR7 signaling cascades in B cells may have utility in pSD patients. B cell TLR7 expression and ABCs are increased in a pSD model and TLR7-stimulated ABCs derived from pSD females show pathogenic potential.

The validity of adjusting dialysate bicarbonate based on pre‐hemodialysis (HD) serum bicarbonate is unclear. There are no studies of the impact of dialysate sodium on blood pH.

Modern high-dimensional statistical inference often faces the problem of missing data. In recent decades, many studies have focused on this topic and provided strategies including complete-sample analysis and imputation procedures. However, complete-sample analysis discards information of incomplete samples, while imputation procedures have accumulative errors from each single imputation. In this paper, we propose a new method, Sample-wise COmbined missing effect Model with penalization (SCOM), to deal with missing data occurring in predictors. Instead of imputing the predictors, SCOM estimates the combined effect caused by all missing data for each incomplete sample. SCOM makes full use of all available data. It is robust with respect to various missing mechanisms. Theoretical studies show the oracle inequality for the proposed estimator, and the consistency of variable selection and combined missing effect selection. Simulation studies and an application to the Residential Building Data also illustrate the effectiveness of the proposed SCOM.

PURPOSE:Frameworks for selecting exposures in high-dimensional environmental datasets, while considering confounding, are lacking. We present a two-step approach for exposure selection with subsequent confounder adjustment for statistical inference. METHODS:We measured cognitive ability in 338 children using the Woodcock-Muñoz General Intellectual Ability (GIA) score, and potential associated features across several environmental domains. Initially, 111 variables theoretically associated with GIA score were introduced into a Least Absolute Shrinkage and Selection Operator (LASSO) in a 50% feature selection subsample. Effect estimates for selected features were subsequently modeled in linear regressions in a 50% inference (hold out) subsample, first adjusting for sex and age and later for covariates selected via directed acyclic graphs (DAGs). All models were adjusted for clustering by school. RESULTS:Of the 15 LASSO selected variables, eleven were not associated with GIA score following our inference modeling approach. Four variables were associated with GIA scores, including: serum ferritin adjusted for inflammation (inversely), mother's IQ (positively), father's education (positively), and hours per day the child works on homework (positively). Serum ferritin was not in the expected direction. CONCLUSIONS:Our two-step approach moves high-dimensional feature selection a step further by incorporating DAG-based confounder adjustment for statistical inference.

Tumor derived exosomes (TEXs) have emerged as promising biomarkers for cancer liquid biopsy. Conventional methods (such as ELISA and qRT-PCR) and emerging biosensing technologies mainly detect a single type of exosomal biomarker due to the distinct properties of different biomolecules. Sensitive detection of two different types of TEX biomarkers, i.e., protein and microRNA combined biomarkers, may greatly improve cancer diagnostic accuracy. We developed an exosome protein microRNA one-stop (Exo-PROS) biosensor that not only selectively captured TEXs but also enabled in situ, simultaneous detection of TEX protein-microRNA pairs via a surface plasmon resonance mechanism. Exo-PROS assay is a fast, reliable, low sample consumption, and user-friendly test. With a total of 175 cancer patients and normal controls, we demonstrated that TEX protein-microRNA pairs measured by Exo-PROS assay detected lung cancer and breast cancer with 99% and 96% accuracy, respectively. Exo-PROS assay also showed superior diagnostic performance to conventional ELISA and qRT-PCR methods. Our results demonstrated that Exo-PROS assay is a potent liquid biopsy assay for cancer diagnosis.

Weighted nearest neighbors (WNN) classifiers are popular non-parametric classifiers. Despite the significant progress in WNN, most existing WNN classifiers are designed for traditional supervised learning problems where both training samples and test samples are assumed to be independent and identically distributed. However, in many real applications, it could be difficult or expensive to obtain training samples from the distribution of interest. Therefore, data collected from some related distributions are often used as supplementary training data for the classification task under the distribution of interest. It is essential to develop effective classification methods that could incorporate both training samples from the distribution of interest (if they exist) and supplementary training samples from a different but related distribution. To address this challenge, we propose a novel Transfer learning weighted Nearest Neighbors (TNN) classifier. As a WNN classifier, TNN determines the weights on the class labels of training samples for different test samples adaptively by minimizing an upper bound on the conditional expectation of the estimation error of the regression function. It puts decreasing weights on the class labels of the successive more distant neighbors. To accommodate the difference between training samples from the distribution of interest and supplementary training samples, TNN adds a non-negative offset to the distance between each supplementary training sample and the test sample, and thus constrains the excessive influence of the supplementary training samples on the prediction. Our theoretical studies show that, under certain conditions, TNN is consistent and minimax optimal (up to a logarithmic factor) in the covariate shift setting. In the posterior drift or the more general setting where both covariate shift and posterior drift exist, the excess risk of TNN depends on the maximum posterior discrepancy between the distribution of the supplementary training samples and the distribution of interest. Both our simulation studies and an application to the land use/land cover mapping problem in geography demonstrate that TNN outperforms other existing methods. It can serve as an effective tool for transfer learning.

Tree-based machine learning models based on environmental features offer low-cost and timely solutions for predicting microbial fecal contamination in beach water to inform the public of the health risk. However, many of these models are black boxes that are difficult for humans to understand, which may cause severe consequences such as unexplained decisions and failure in accountability. To develop interpretable predictive models for beach water quality, we evaluate five tree-based models, namely classification tree, random forest, CatBoost, XGBoost, and LightGBM, and employ a state-of-the-art explanation method SHAP to explain the models. When tested on the Escherichia coli (E. coli) concentration data collected from three beach sites along Lake Erie shores, LightGBM, followed by XGBoost, achieves the highest averaged precision and recall scores. For all three sites, both models suggest lake turbidity as the most important predictor, and elucidate the crucial role of accurate local data of wave height and rainfall in the model development. Local SHAP values further reveal the robustness of the importance of lake turbidity as its SHAP value increases nearly monotonically with its value and is minimally affected by other environmental factors. Moreover, we found an intriguing interaction between lake turbidity and day-of-year. This work suggests that the combination of LightGBM and SHAP has a promising potential to develop interpretable models for predicting microbial water quality in freshwater lakes.

Lead exposure and neighborhoods can affect children’s behavior, but it is unclear if neighborhood characteristics modify the effects of lead on behavior. Understanding these modifications has important intervention implications. Blood lead levels (BLLs) in children (~7 years) from Montevideo, Uruguay, were categorized at 2 µg/dL. Teachers completed two behavior rating scales (n = 455). At one-year follow-up (n = 380), caregivers reported child tantrums and parenting conflicts. Multilevel generalized linear models tested associations between BLLs and behavior, with neighborhood disadvantage, normalized difference vegetation index (NDVI), and distance to nearest greenspace as effect modifiers. No effect modification was noted for neighborhood disadvantage or NDVI. Children living nearest to greenspace with BLLs < 2 µg/dL were lower on behavior problem scales compared to children with BLLs ≥ 2 µg/dL. When furthest from greenspace, children were similar on behavior problems regardless of BLL. The probability of daily tantrums and conflicts was ~20% among children with BLLs < 2 µg/dL compared to ~45% among children with BLLs ≥ 2 µg/dL when closest to greenspace. Furthest from greenspace, BLLs were not associated with tantrums and conflicts. Effect modification of BLL on child behavior by distance to greenspace suggests that interventions should consider both greenspace access and lead exposure prevention.

In modern biomedical classification applications, data are often collected from multiple modalities, ranging from various omics technologies to brain scans. As different modalities provide complementary information, classifiers using multi-modality data usually have good classification performance. However, in many studies, due to the high cost of measures, in a lot of samples, some modalities are missing and therefore all data from those modalities are missing completely. In this case, the training data set is a block-missing multi-modality data set. In this paper, considering such classification problems, we develop a new weighted nearest neighbors classifier, called the integrative nearest neighbor (INN) classifier. INN harnesses all available information in the training data set and the feature vector of the test data point effectively to predict the class label of the test data point without deleting or imputing any missing data. Given a test data point, INN determines the weights on the training samples adaptively by minimizing the worst-case upper bound on the estimation error of the regression function over a convex class of functions. Our simulation study shows that INN outperforms common weighted nearest neighbors classifiers that only use complete training samples or modalities that are available in each sample. It performs better than methods that impute the missing data as well, even for the case where some modalities are missing not at random. The effectiveness of INN has been also demonstrated by our theoretical studies and a real application from the Alzheimer’s disease neuroimaging initiative.

Co-exposure to inorganic arsenic (iAs) and fluoride (F-) and their collective actions on cardiovascular systems have been recognized as a global public health concern. Emerging studies suggest an association between the perturbation of gut bacterial microbiota and adverse cardiovascular effects (CVEs), both of which are the consequence of iAs and F- exposure in human and experimental animals. The aim of this study was to fill the gap of understanding the relationship among co-exposure to iAs and F-, gut microbiota perturbation, and adverse CVEs. We systematically assessed cardiac morphology and functions (blood pressure, echocardiogram, and electrocardiogram), and generated gut microbiota profiles using 16S rRNA gene sequencing on rats exposed to iAs (50 mg/L NaAsO2), F- (100 mg/L NaF) or combined iAs and F- (50 mg/L NaAsO2 + 100 mg/L NaF), in utero and during early postnatal periods (postnatal day 90). Correlation analysis was then performed to examine relationship between significantly altered microbiota and cardiac performance indices. Our results showed that co-exposure to iAs and F- resulted in more prominent effects in CVEs and perturbation of gut microbiota profiles, compared to iAs or F- treatment alone. Furthermore, nine bacterial genera (Adlercreutzia, Clostridium sensu stricto 1, Coprococcus 3, Romboutsia, [Bacteroides] Pectinophilus group, Lachnospiraceae NC2004 group, Desulfovibrio, and two unidentified genera in Muribaculaceae and Ruminococcaceae family), which differed significantly in relative abundance between control and iAs and F- co-exposure group, were strongly correlated with the higher risk of CVEs (correlation coefficient = 0.70-0.88, p < 0.05). Collectively, these results suggest that co-exposure to iAs and F- poses a higher risk of CVEs, and the part of the mode of action is potentially through inducing gut microbiota disruption, and the strong correlations between them indicate a high potential for the development of novel microbiome-based biomarkers of iAs and/or F- associated CVEs.

Myd88 activation is an important driver of autoimmunity. Primary Sjögren's syndrome (pSS) is an autoimmune disease characterized by exocrine gland dysfunction in combination with serious systemic disease manifestations. Myd88-dependent signaling networks remain incompletely understood in the context of pSS. The objective of this study was to establish the contribution of tissue-specific Myd88 activation to local (exocrine) and systemic pSS manifestations. To this end, we generated two novel conditional knockout pSS mouse models; one lacking Myd88 in hematopoietic cells and a second strain in which Myd88 was deleted in the stromal compartment. Spontaneous production of inflammatory mediators was altered in salivary tissue, and nephritis was diminished in both conditional knockout strains. In contrast, pulmonary inflammation was increased in mice lacking Myd88 in hematopoietic cells and was reduced when Myd88 was ablated in stromal cells. Finally, anti-nuclear autoantibodies (ANAs) were attenuated in pSS mice lacking Myd88 in immune cells. Additionally, the ANA-specific B cell repertoire was skewed in the Myd88-deficient strains. Collectively, these data demonstrate that Myd88 activation in specific cell types is essential for distinct aspects of pSS pathology.

Studies in mice and humans identify Myd88 activation as an important driver of autoimmune disease, although the role of this adapter molecule in primary Sjögren’s syndrome (pSS) is poorly understood. In pSS, patients exhibit exocrine gland dysfunction along with serious systemic disease manifestations. Although Myd88-dependent signaling networks are essential for pSS disease pathogenesis, these pathways remain incompletely understood in the context of pSS. The objective of this study was to establish the contribution of Myd88 activation in immune cells to local and systemic pSS manifestations. To this end, we generated a novel conditional knockout pSS mouse model lacking Myd88 in the hematopoietic compartment. We found that ablation of Myd88 in immune cells altered inflammatory mediators in salivary tissue, but did not affect the degree of lymphocytic infiltration in exocrine glands or salivary production. Nephritis was diminished and pulmonary inflammation was increased in mice lacking Myd88 in hematopoietic cells. Finally, we found total IgM titers and autoreactive IgM and IgG were attenuated in pSS mice lacking Myd88 in immune cells. Thus, we demonstrated that Myd88 activation in the hematopoietic compartment is essential for distinct aspects of disease. This study establishes a rationale for design of targeted therapeutics that modulate Myd88 activation in a tissue-specific manner to ameliorate salivary inflammation and systemic disease manifestations characteristic of pSS.

Objectives > To evaluate the correlation between craniofacial structures, anthropometric measurements, and bony and soft tissue nasopharyngeal dimensions in African Black adolescents. Methods > This retrospective cross-sectional study was conducted on 483 healthy adolescents (250 females and 233 males), randomly selected from one dental clinic. The inclusion criteria were skeletal and dental Class I, Black ethnicity, pubertal growth period as determined by the Greulich and Pyle atlas criteria, and no history of orthodontic treatment. Anthropometric measurements (stature, upper body height, lower body height, and BMI) and radiographic records (hand-wrist radiographs, and lateral cephalograms) were obtained. One investigator traced and analysed all cephalograms to determine three skeletal craniofacial parameters (maxillary length [Ar- ANS], mandibular length [Ar-Gn], and lower anterior facial height [ANS-Me]), and 14 (8 skeletal and 6 soft tissue) nasopharyngeal parameters. Pearson correlation coefficients and stepwise multiple linear regression analyses were conducted. Results > The mean skeletal ages of females and males were 11.31 +/- 2.31y and 12.66 +/- 1.85y, respectively. Multiple linear regression analyses showed that stature, posterior height of nasal cavity (S-PNS), length of nasal floor (AA-PNS), and mean area of bony nasopharynx (Area 1) were significantly correlated with maxillary length, P<.001. Stature, BMI, S-PNS, vertical angle of nasopharynx (Ba-S-PNS), Area 1, adenoid height (AD), and linear hyoid bone measurements (H-AA, H-RGN, H-Ax) were all correlated with mandibular length, P<.05. Lower facial height showed sexual dimorphism and was significantly associated with vertical nasopharyngeal measurements, BMI and upper body height. Conclusions > Craniofacial structures were significantly associated with stature and upper body height. Maxillary growth was associated with bony nasopharyngeal variables. Mandibular growth and lower facial height were associated with bony and soft tissue nasopharyngeal variables. The sexual dimorphism in lower facial height warrants future studies to fully understand and manage the craniofacial complex and nasopharyngeal airway in African Black adolescents.

Primary Sjögren’s syndrome is an autoimmune disease that is predominantly seen in women. The disease is characterized by exocrine gland dysfunction in combination with serious systemic manifestations. At present, the causes of pSS are poorly understood. Pulmonary and renal inflammation are observed in pSS mice, reminiscent of a subset of pSS patients. A growing body of evidence indicates that inflammation mediated by Damage-Associated Molecular Patterns (DAMPs) contributes to autoimmunity, although this is not well-studied in pSS. Degraded extracellular matrix (ECM) constituents can serve as DAMPs by binding pattern-recognition receptors and activating Myd88-dependent signaling cascades, thereby exacerbating and perpetuating inflammatory cascades. The ECM components biglycan (Bgn) and decorin (Dcn) mediate sterile inflammation and both are implicated in autoimmunity. The objective of this study was to determine whether these ECM components and anti-ECM antibodies are altered in a pSS mouse model, and whether this is dependent on Myd88 activation in immune cells. Circulating levels of Bgn and Dcn were similar among pSS mice and controls and tissue expression studies revealed pSS mice had robust expression of both Bgn and Dcn in the salivary tissue, saliva, lung and kidney. Sera from pSS mice displayed increased levels of autoantibodies directed against ECM components when compared to healthy controls. Further studies using sera derived from conditional knockout pSS mice demonstrated that generation of these autoantibodies relies, at least in part, on Myd88 expression in the hematopoietic compartment. Thus, this study demonstrates that ECM degradation may represent a novel source of chronic B cell activation in the context of pSS.

Abstract Budget constraints become an important consideration in modern predictive modeling due to the high cost of collecting certain predictors. This motivates us to develop cost-constrained predictive modeling methods. In this article, we study a new high-dimensional cost-constrained linear regression problem, that is, we aim to find the cost-constrained regression model with the smallest expected prediction error among all models satisfying a budget constraint. The nonconvex budget constraint makes this problem NP-hard. In order to estimate the regression coefficient vector of the cost-constrained regression model, we propose a new discrete first-order continuous optimization method. In particular, our method delivers a series of estimates of the regression coefficient vector by solving a sequence of 0-1 knapsack problems. Theoretically, we prove that the series of the estimates generated by our iterative algorithm converge to a first-order stationary point, which can be a globally optimal solution under some conditions. Furthermore, we study some extensions of our method that can be used for general statistical learning problems and problems with groups of variables. Numerical studies using simulated datasets and a real dataset from a diabetes study indicate that our proposed method can solve problems of fairly high dimensions with promising performance.

Microbial pollution of beach water can expose swimmers to harmful pathogens. Predictive modeling provides an alternative method for beach management that addresses several limitations associated with traditional culture-based methods of assessing water quality. Widely-used machine learning methods often suffer from high variability in performance from one year or beach to another. Therefore, the best machine learning method varies between beaches and years, making method selection difficult. This study proposes an ensemble machine learning approach referred to as model stacking that has a two-layered learning structure, where the outputs of five widely-used individual machine learning models (multiple linear regression, partial least square, sparse partial least square, random forest, and Bayesian network) are taken as input features for another model that produces the final prediction. Applying this approach to three beaches along eastern Lake Erie, New York, USA, we show that generally the model stacking approach was able to generate reliably good predictions compared to all of the five base models. The accuracy rankings of the stacking model consistently stayed 1st or 2nd every year, with yearly-average accuracy of 78%, 81%, and 82.3% at the three studied beaches, respectively. This study highlights the value of the model stacking approach in predicting beach water quality and solving other pressing environmental problems.

Many “hot spot” geographic areas across the world with drinking water co-contaminated with inorganic arsenic (iAs) and fluoride (F-), two of the most common natural contaminants in drinking water. Both iAs and F- are known neurotoxins and affect neurodevelopment of children. However, very few studies have investigated the neurodevelopmental effects of concurrent exposure to iAs and F-, which could potentially pose a greater risk than iAs or F- exposure alone. Further, perturbations of gut microbiota, which plays a regulatory role in neurodevelopment, resulting from iAs and F- exposure has been reported in numerous studies. There is lacking of information regarding to the relationship among concurrent iAs and F- exposure, microbiome disruption, and neurodevelopmental impacts. To fill these gaps, we treated offspring rats to iAs (50 mg/L NaAsO2) and F- (100 mg/L NaF), alone or combined from early life (in utero and childhood) to puberty. We applied Morris water maze test to assess spatial learning and memory of these rats and generated gut microbiome profiles using 16S rRNA gene sequencing. We showed that concurrent iAs and F- exposure caused more prominent neurodevelopmental effects in rats than either iAs or F- exposure alone. Moreover, Unsupervised Principal Coordinates Analysis (PCoA) and Linear Discriminant Analysis Effect Size (LEfSe) analysis of gut microbiome sequencing results separated concurrent exposure group from others, indicating a more sophisticated change of gut microbial communities occurred under the concurrent exposure condition. Further, a correlation analysis between indices of the water maze test and microbial composition at the genus level identified featured genera that were clearly associated with neurobehavioral performance of rats. 75% (9 out of 12) genera, which had a remarkable difference in relative abundance between the control and combined iAs and F- exposure groups, showed significantly strong correlations (r = 0.70–0.90) with the water maze performance indicators. Collectively, these results suggest that concurrent iAs and F- exposure led to more prominent effects on neurodevelopment and gut microbiome composition structures in rats, and the strong correlation between them indicates a high potential for the development of novel microbiome-based biomarkers of iAs and/or F- associated neurodevelopmental deficits.

Toll-like receptors (TLRs) are important mediators of chronic inflammation in numerous autoimmune diseases, although the role of these receptors in primary Sjögren's syndrome (pSS) remains incompletely understood. Previous studies in our laboratory established Myd88 as a crucial mediator of pSS, although the disease-relevant ligands and the upstream signaling events that culminate in Myd88 activation have yet to be established. The objective of this study was to identify specific Myd88-dependent TLR-related pathways that are dysregulated both locally and systemically in a mouse model of pSS [NOD.B10Sn-H2b/J (NOD.B10)]. We performed RNA-sequencing on spleens derived from NOD.B10 mice. We then harvested salivary tissue and spleens from Myd88-sufficient and deficient C57BL/10 (BL/10) and NOD.B10 mice and performed flow cytometry to determine expression of Myd88-dependent TLRs. We cultured splenocytes with TLR2 and TLR4 agonists and measured production of inflammatory mediators by ELISA. Next, we evaluated spontaneous and TLR4-mediated inflammatory cytokine secretion in NOD.B10 salivary tissue. Finally, we assessed spontaneous Myd88-dependent cytokine secretion by NOD.B10 salivary cells. We identified dysregulation of numerous TLR-related networks in pSS splenocytes, particularly those employed by TLR2 and TLR4. We found upregulation of TLRs in both the splenic and salivary tissue from pSS mice. In NOD.B10 splenic tissue, robust expression of B cell TLR1 and TLR2 required Myd88. Splenocytes from NOD.B10 mice were hyper-responsive to TLR2 ligation and the endogenous molecule decorin modulated inflammation via TLR4. Finally, we observed spontaneous secretion of numerous inflammatory cytokines and this was enhanced following TLR4 ligation in female NOD.B10 salivary tissue as compared to males. The spontaneous production of salivary IL-6, MCP-1 and TNFα required Myd88 in pSS salivary tissue. Thus, our data demonstrate that Myd88-dependent TLR pathways contribute to the inflammatory landscape in pSS, and inhibition of such will likely have therapeutic utility.

Linear discriminant analysis (LDA) is widely used for various binary classification problems. In contrast to the LDA that estimates the precision matrix Ω and the mean difference vector δ in the classification rule separately, the linear programming discriminant (LPD) rule estimates the product Ωδ directly through a constrained ℓ1 minimization. The LPD rule has very good classification performance on many high‐dimensional binary classification problems. However, to estimate β* = Ωδ, the LPD rule uses equal weights for all the elements of β* in the constrained ℓ1 minimization. It may not deliver the optimal estimate of β*, and therefore the estimated discriminant direction can be suboptimal. In order to obtain better estimates of β* and the discriminant direction, we can heavily penalize βj in the constrained ℓ1 minimization if we suspect the jth feature is useless for the classification while moderately penalize βj if we suspect the jth feature is useful. In this paper, based on the LPD rule and some popular feature screening methods, we propose a new weighted linear programming discriminant (WLPD) rule for the high‐dimensional binary classification problem. The screening statistics used in the marginal two‐sample t‐test screening, Kolmogorov–Smirnov filter, and the maximum marginal likelihood screening will be used to construct appropriate weights for different elements of β* flexibly. Besides the linear programming algorithm, we develop a new alternating direction method of multipliers algorithm to solve the high‐dimensional constrained ℓ1 minimization problem efficiently. Our numerical studies show that our proposed WLPD rule can outperform LPD and serve as an effective binary classification tool.