false discovery rate
Recently Published Documents


TOTAL DOCUMENTS

634
(FIVE YEARS 166)

H-INDEX

57
(FIVE YEARS 9)

2021 ◽  
Author(s):  
Ron Berman ◽  
Christophe Van den Bulte

We investigate what fraction of all significant results in website A/B testing is actually null effects (i.e., the false discovery rate (FDR)). Our data consist of 4,964 effects from 2,766 experiments conducted on a commercial A/B testing platform. Using three different methods, we find that the FDR ranges between 28% and 37% for tests conducted at 10% significance and between 18% and 25% for tests at 5% significance (two sided). These high FDRs stem mostly from the high fraction of true null effects, about 70%, rather than from low power. Using our estimates, we also assess the potential of various A/B test designs to reduce the FDR. The two main implications are that decision makers should expect one in five interventions achieving significance at 5% confidence to be ineffective when deployed in the field and that analysts should consider using two-stage designs with multiple variations rather than basic A/B tests. This paper was accepted by Eric Anderson, marketing.


2021 ◽  
Author(s):  
Ye Yue ◽  
Yijuan Hu

Abstract Background: Understanding whether and which microbes played a mediating role between an exposure and a disease outcome are essential for researchers to develop clinical interventions to treat the disease by modulating the microbes. Existing methods for mediation analysis of the microbiome are often limited to a global test of community-level mediation or selection of mediating microbes without control of the false discovery rate (FDR). Further, while the null hypothesis of no mediation at each microbe is a composite null that consists of three types of null (no exposure-microbe association, no microbe-outcome association given the exposure, or neither), most existing methods for the global test such as MedTest and MODIMA treat the microbes as if they are all under the same type of null. Results: We propose a new approach based on inverse regression that regresses the (possibly transformed) relative abundance of each taxon on the exposure and the exposure-adjusted outcome to assess the exposure-taxon and taxon-outcome associations simultaneously. Then the association p-values are used to test mediation at both the community and individual taxon levels. This approach fits nicely into our Linear Decomposition Model (LDM) framework, so our new method is implemented in the LDM and enjoys all the features of the LDM, i.e., allowing an arbitrary number of taxa to be tested, supporting continuous, discrete, or multivariate exposures and outcomes as well as adjustment of confounding covariates, accommodating clustered data, and offering analysis at the relative abundance or presence-absence scale. We refer to this new method as LDM-med. Using extensive simulations, we showed that LDM-med always controlled the type I error of the global test and had compelling power over existing methods; LDM-med always preserved the FDR of testing individual taxa and had much better sensitivity than alternative approaches. In contrast, MedTest and MODIMA had severely inflated type I error when different taxa were under different types of null. The flexibility of LDM-med for a variety of mediation analyses is illustrated by the application to a murine microbiome dataset, which identified a plausible mediator.Conclusions: Inverse regression coupled with the LDM is a strategy that performs well and is capable of handling mediation analysis in a wide variety of microbiome studies.


Author(s):  
Paulo Sergio Dourado Arrais ◽  
Eudiana Vale Francelino ◽  
Eduardo Gabriel Pinheiro ◽  
Silvia Maria Freitas ◽  
Elisabeth Carmen Duarte ◽  
...  

Introdução: Os Eventos Adversos a Medicamentos-EAM representam riscos à saúde e sua subnotificação representa um desafio para a saúde pública. A busca ativa de casos suspeitos de EAM nos bancos de dados de saúde utilizando a Classificação Internacional de Doenças-CID é uma das estratégias que pode reduzir as subnotificações desses eventos. Objetivo:O objetivo desse estudo é identificar os códigos da CID mais usados como rastreadores de EAM e avaliar a sua concordância entre os pesquisadores. Métodos: Foi realizada uma revisão sistemática da literatura utilizando as bases de dados PubMed, Scopus, Web of Science, MEDLINE e LILACS com os descritores “Classificação Internacional de Doenças”, “CID-10”, “Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos”, “Envenenamento”, “Erros de Medicação”. Os artigos incluídos tiveram seus códigos CID identificados, comparados e sua qualidade avaliada. A análise de concordância dos códigos foi feita usando o modelo de ensaios de Bernoulli, testes de proporções binomial exata e a técnica de false discovery rate para analisar as hipóteses postas. A análise estatística foi feita com o software R. O estudo está registrado no PROSPERO sob n.º CRD42019120694. Resultados: Foram identificados 5.167 artigos e após os critérios de seleção, 33 foram incluídos nessa revisão. Foram identificados 1.105 códigos da CID. O coeficiente de prevalência dos EAM variou entre 0,18% e 18,4% em internações hospitalares e a taxa de mortalidade variou entre 0,12 a 45,9 óbitos por 100 mil óbitos. Somente 195 (17,7%) códigos tiveram alta concordância entre os pesquisadores. Muitos códigos CID utilizados para detectar EAM possuem baixa concordância entre os pesquisadores e produziram diferentes taxas do evento. Conclusão: Os códigos rastreadores de EAM identificados representam um método simples e eficiente para captação de eventos adversos em grandes bancos de dados em saúde, contribuindo na redução da subnotificação nos tradicionais sistemas de notificações de EAM.


2021 ◽  
Vol 22 (1) ◽  
Author(s):  
David Chardin ◽  
Olivier Humbert ◽  
Caroline Bailleux ◽  
Fanny Burel-Vandenbos ◽  
Valerie Rigau ◽  
...  

Abstract Background Supervised classification methods have been used for many years for feature selection in metabolomics and other omics studies. We developed a novel primal-dual based classification method (PD-CR) that can perform classification with rejection and feature selection on high dimensional datasets. PD-CR projects data onto a low dimension space and performs classification by minimizing an appropriate quadratic cost. It simultaneously optimizes the selected features and the prediction accuracy with a new tailored, constrained primal-dual method. The primal-dual framework is general enough to encompass various robust losses and to allow for convergence analysis. Here, we compare PD-CR to three commonly used methods: partial least squares discriminant analysis (PLS-DA), random forests and support vector machines (SVM). We analyzed two metabolomics datasets: one urinary metabolomics dataset concerning lung cancer patients and healthy controls; and a metabolomics dataset obtained from frozen glial tumor samples with mutated isocitrate dehydrogenase (IDH) or wild-type IDH. Results PD-CR was more accurate than PLS-DA, Random Forests and SVM for classification using the 2 metabolomics datasets. It also selected biologically relevant metabolites. PD-CR has the advantage of providing a confidence score for each prediction, which can be used to perform classification with rejection. This substantially reduces the False Discovery Rate. Conclusion PD-CR is an accurate method for classification of metabolomics datasets which can outperform PLS-DA, Random Forests and SVM while selecting biologically relevant features. Furthermore the confidence score provided with PD-CR can be used to perform classification with rejection and reduce the false discovery rate.


Author(s):  
Yongheng Wang ◽  
Jincheng Zhai ◽  
Xianglu Wu ◽  
Enoch Appiah Adu-Gyamfi ◽  
Lingping Yang ◽  
...  

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Hirotada Kobayashi ◽  
Karin Amrein ◽  
Jessica A. Lasky-Su ◽  
Kenneth B. Christopher

AbstractProcalcitonin is a biomarker of systemic inflammation and may have importance in the immune response. The metabolic response to elevated procalcitonin in critical illness is not known. The response to inflammation is vitally important to understanding metabolism alterations during extreme stress. Our aim was to determine if patients with elevated procalcitonin have differences in the metabolomic response to early critical illness. We performed a metabolomics study of the VITdAL-ICU trial where subjects received high dose vitamin D3 or placebo. Mixed-effects modeling was used to study changes in metabolites over time relative to procalcitonin levels adjusted for age, Simplified Acute Physiology Score II, admission diagnosis, day 0 25-hydroxyvitamin D level, and the 25-hydroxyvitamin D response to intervention. With elevated procalcitonin, multiple members of the short and medium chain acylcarnitine, dicarboxylate fatty acid, branched-chain amino acid, and pentose phosphate pathway metabolite classes had significantly positive false discovery rate corrected associations. Further, multiple long chain acylcarnitines and lysophosphatidylcholines had significantly negative false discovery rate corrected associations with elevated procalcitonin. Gaussian graphical model analysis revealed functional modules specific to elevated procalcitonin. Our findings show that metabolite differences exist with increased procalcitonin indicating activation of branched chain amino acid dehydrogenase and a metabolic shift.


Author(s):  
Abhishek Joshi ◽  
Lukas E. Schmidt ◽  
Sean A. Burnap ◽  
Ruifang Lu ◽  
Melissa V. Chan ◽  
...  

Objective: Platelets are central to acute myocardial infarction (MI). How the platelet proteome is altered during MI is unknown. We sought to describe changes in the platelet proteome during MI and identify corresponding functional consequences. Approach and Results: Platelets from patients experiencing ST-segment–elevation MI (STEMI) before and 3 days after treatment (n=30) and matched patients with severe stable coronary artery disease before and 3 days after coronary artery bypass grafting (n=25) underwent quantitative proteomic analysis. Elevations in the proteins S100A8 and S100A9 were detected at the time of STEMI compared with stable coronary artery disease (S100A8: FC, 2.00; false discovery rate, 0.05; S100A9: FC, 2.28; false discovery rate, 0.005). During STEMI, only S100A8 mRNA and protein levels were correlated in platelets ( R =0.46, P =0.012). To determine whether de novo protein synthesis occurs, activated platelets were incubated with 13C-labeled amino acids for 24 hours and analyzed by mass spectrometry. No incorporation was confidently detected. Platelet S100A8 and S100A9 was strongly correlated with neutrophil abundance at the time of STEMI. When isolated platelets and neutrophils were coincubated under quiescent and activated conditions, release of S100A8 from neutrophils resulted in uptake of S100A8 by platelets. Neutrophils released S100A8/A9 as free heterodimer, rather than in vesicles or extracellular traps. In the community-based Bruneck study (n=338), plasma S100A8/A9 was inversely associated with platelet reactivity—an effect abrogated by aspirin. Conclusions: Leukocyte-to-platelet protein transfer may occur in a thromboinflammatory environment such as STEMI. Plasma S100A8/A9 was negatively associated with platelet reactivity. These findings highlight neutrophils as potential modifiers for thrombotic therapies in coronary artery disease.


Sign in / Sign up

Export Citation Format

Share Document