China
Africa
Explore chapters and articles related to this topic
Cancer Informatics
Published in Trevor F. Cox, Medical Statistics for Cancer Studies, 2022
We digress to explain false discovery rate. The problem with carrying out so many t-tests or other hypothesis tests is that there will be many Type I errors made. So, we will have a high false discovery rate. We have so many very small p-values and we need to adjust these to allow for the FDR.
Clinical Development in the Light of Bayesian Statistics
Published in Emmanuel Lesaffre, Gianluca Baio, Bruno Boulanger, Bayesian Methods in Pharmaceutical Research, 2020
As previously mentioned, a common mistake is the failure to consider the prior probability of a null or small treatment effect. To illustrate this further, Figure 4.2 considers the quantitative decision rule associated with a Phase II trial (e.g. achieving statistical significance) to be a diagnostic test for further drug development. The figure shows the false discovery rate is driven by the background success rate. Unfortunately, in practice this rate is unknown and difficult to estimate or model. In other settings (e.g. genomics), where hundreds of thousands of candidate drugs are assessed simultaneously, it is possible to estimate the background null rate empirically (Storey and Tibshirani, 2003). However, without a large clear pool of relevant cases, the drug development setting is much more challenging. Possible solutions include expert elicitation (Kinnersley and Day, 2013) or using historical data from the same disease to form the prior. In either case, subjective assumptions must be made.
Multiple Comparisons, Multiple Primary Endpoints and Subpopulation Analysis
Published in Susan Halabi, Stefan Michiels, Textbook of Clinical Trials in Oncology, 2019
Ekkehard Glimm, Dong Xi, Paul Gallo
Another type of error rate is becoming more frequently considered, especially when exploring many hypotheses. The false discovery rate is the expected proportion of falsely rejected hypotheses among the rejected hypotheses. As the focus of this chapter involves confirmatory trials, we only consider the FWER in this chapter and multiple testing procedures that control it in the strong sense.
The COVID-19 pandemic: asthma control, tobacco use, and mental health among African American and Latinx college students
Published in Journal of Asthma, 2023
Mayra S. Ramos, Rosalie Corona, Katherine W. Dempster, Sarah C. M. Morton, Robin S. Everhart
To examine the association between COVID-19 impact scores and asthma control on other study variables, linear and logistic regression analyses were performed. Age, gender, and race/ethnicity were included as covariates in all analyses. Given the established associations among stress, anxiety, depressive symptoms, and tobacco use, for any significant associations between COVID-19 impact scores/asthma control and tobacco use outcome variables, we reran these regression models controlling for stress, anxiety, and depressive symptoms. For significant regression analyses, we used the Benjamini-Hochberg (44) correction to control for Type 1 error in multiple comparisons. This type of correction controls for false discovery rate by ranking observed p values from lowest to highest. The false discovery rate, in this case 0.05, is multiplied by p values rank for each test over the total number of tests. The original p values for the test is then compared to the adjusted p values; if the original p values is less than the adjusted p values using the Benjamini-Hochberg correction, the result for that comparison remains significant.
Dysbiotic but nonpathogenic shift in the fecal mycobiota of patients with rheumatoid arthritis
Published in Gut Microbes, 2022
Eun Ha Lee, Hyun Kim, Jung Hee Koh, Kwang Hyun Cha, Kiseok Keith Lee, Wan-Uk Kim, Cheol-Ho Pan, Yong-Hwan Lee
Statistical analysis was performed using R statistical software, version 3.5.2.52 After multiple hypothesis tests had been corrected using the false discovery rate method, significant results were determined using a p-value threshold of 0.05. First, OTU tables were scaled by cumulative-sum scaling (CSS) and log-transformed (for normalization) using the cumNum and MRcounts functions in the metagenomeSeq package in R.53 Next, rarefication of bacterial (5425 reads) and fungal (4256 reads) reads was conducted using the rarefy_even_depth function in the Phyloseq package in R; this was followed by calculation of the Shannon and Simpson indices using the diversity function in the Vegan (version 2.5–3) package in R. The Wilcoxon rank-sum test and one-way analysis of variance were also used. A Bray–Curtis dissimilarity matrix was produced for use in two separate principal coordinates analyses; CAP was then performed using RA and HC constraints, respectively, with the capscale and ordinate functions from the Vegan and Phyloseq packages. PERMANOVA using the adonis function in the Vegan package (version 2.5–3) was also used for analysis.54 Subsequently, the core OTUs of RA and HC groups were identified using a prevalence threshold of 85% for bacteria and 70% for fungi. Differentially abundant OTUs between the RA and HC groups were identified using LEfSe (https://huttenhower.sph.harvard.edu/galaxy/).55 Differences in OTU abundance were considered significant when p-values were < 0.05.
Toll-like receptor agonist combinations augment mouse T-cell anti-tumor immunity via IL-12- and interferon ß-mediated suppression of immune checkpoint receptor expression
Published in OncoImmunology, 2022
Donghwan Jeon, Douglas G. McNeel
Data quality was examined by FastQC26 with per-base sequence quality scores. Data that passed the quality control were aligned to the mouse reference genome using RNA STAR.27 The expression level of each gene was calculated by FeatureCounts,28 and heat-maps for genes of interest were generated using R 3.3.1. Gene expression profiles were subsequently used for differential gene expression analysis using DESeq2.29 The false discovery rate was controlled using the Benjamini–Hochberg procedure. Rank lists for Gene Set Enrichment Analysis (GSEA) were generated from DESeq2 results, with the following formula: ‘Sign(log2FoldChange) X -log10(p-value)’. Pre-ranked GSEA was performed with the Molecular Signatures Database (MSigDB) immunologic signature gene sets.30 RNAseq data (BioProject ID PRJNA792998) is publicly available at http://www.ncbi.nlm.nih.gov/bioproject/792998.