China
Africa
Explore chapters and articles related to this topic
What would make me an expert?
Published in Thomas A. Gerds, Michael W. Kattan, Medical Risk Prediction, 2021
Thomas A. Gerds, Michael W. Kattan
The major approaches to deal with missing data are inverse probability weighting and multiple imputation [143]. Inverse probability weighting requires nuisance models that allow us to predict if the values are missing based on the observed data. However, in the common case where multiple predictor variables have a non-monotone missingness pattern, the development of the mathematical background for inverse probability weighting is rather recent [167, 168] and there is a lack of software. The mathematical background for multiple imputation is also not fully established, and justified mostly by computer simulation (exceptions are references [62, 174]), however, there is plenty of software to perform multiple imputation. Multiple imputation requires nuisance models that allow us to predict the missing values of the predictor variables based on the observed data.
Inverse Probability Weighted Generalized Estimated Equations
Published in Craig Mallinckrodt, Geert Molenberghs, Ilya Lipkovich, Bohdana Ratitch, Estimands, Estimators and Sensitivity Analysis in Clinical Trials, 2019
Craig Mallinckrodt, Geert Molenberghs, Ilya Lipkovich, Bohdana Ratitch
Inverse probability weighting (IPW) has many applications. The focus here is on using IPW in conjunction with generalized estimating equations (GEEs) to construct weighted GEE (wGEE). A standard reference for wGEE is Robins, Rotnitzky, and Zhao (1995). For details, see also, for example, Molenberghs et al. (2015). The chapter begins with brief review of some technical details for GEE and IPW and then provides implementation examples.
Multiple Decision Treatment Regimes: Overview
Published in Anastasios A. Tsiatis, Marie Davidian, Shannon T. Holloway, Eric B. Laber, Dynamic Treatment Regimes, 2019
Anastasios A. Tsiatis, Marie Davidian, Shannon T. Holloway, Eric B. Laber
can be interpreted as the propensity for receiving treatment consistent with regime d through all K decisions, given the observed history. Thus, the inverse probability weighting is in the spirit discussed in Section 3.3.2; however, the formulation is considerably more complex and is presented formally in Section 6.4.3.
Statistical determination of cancer biomarkers: moving forward clinically
Published in Expert Review of Molecular Diagnostics, 2023
Marika Mokou, Harald Mischak, Maria Frantzi
Several key considerations should be observed to fill the gaps of insufficient or poor use of statistics in biomarker discovery and validation. A correct study design is critical for the successful clinical application of the biomarkers and depends on, among others, the selection of the proper target population, sufficient statistical power and consideration of the influence of possible confounding variables [5]. Power calculations are needed to ensure that an adequate number of samples/events are investigated, in relation to the specific clinical context of use, the specific cancer type prevalence, and the targeted performance improvement over current standards. Standard operating procedures and sampling standardization must be established prior to the initiation of sample collection for the study, to minimize the impact of experimental/analytical variability. Multiple confounding factors (e.g. sex, age, body mass index, or comorbidities) must also be accounted for to ensure the correct estimation of the biomarker performance. Statistical approaches such as inverse probability weighting or Bayesian methods can be used to reduce selection bias to the findings [6]. Biases during patient selection/specimen collection and patient evaluation can be further reduced with randomization and blinding.
Regularized regression when covariates are linked on a network: the 3CoSE algorithm
Published in Journal of Applied Statistics, 2023
Matthias Weber, Jonas Striaukas, Martin Schumacher, Harald Binder
Inverse probability weighting is a method to reweight the data, which can be used, for example, to take into account that data may be censored or truncated (it is thus in spirit similar to using a sample pre-processed with matching techniques, as for example proposed in Ho et al. [7]). Inverse probability weighting consists of two parts. First, for each observation, the probability that an event is reported (i.e. that it is neither censored nor truncated) is estimated. Second, the data is restricted to those observations with a reported event, and each observation receives the weight of the inverse estimated probability that an event is reported for this observation (thus observations for which it is likely that relatively many similar observations are censored or truncated receive a higher weight).
Natural history of Usher type 2 with the c.2299delG mutation of USH2A in a large cohort
Published in Ophthalmic Genetics, 2022
Audrey Meunier, Xavier Zanlonghi, Anne-Françoise Roux, Jean-François Fils, Laure Caspers, Isabelle Migeotte, Marc Abramowicz, Isabelle Meunier
For demographics and baseline characteristic variables, continuous data were compared by means of T-test when homogeneity of variances, tested with the Bartlett’s test, and normality of the residuals, tested with the Shapiro–Wilks test, were reached and means and standard deviations (means ± StDev) are reported. When homogeneity of the variance or normality of the residuals was not proved, Wilcoxon signed rank test was performed on ranked data and medians and interquartile ranges (median [Q25–Q75]) are reported. For count data, the Pearson Chi-Squared test was performed to compare proportions. Lastly, the R package CBPS was used for the propensity score, aiming to equal the two groups (no mutation vs one mutation) on a set of predefined covariates (age at detection and gender) estimating an Average Treatment Effect (ATE) using Covariate Balancing Propensity Score, which has been shown to be superior to traditional logistic regression approaches and boosted classification and regression trees (14). After the propensity score, groups were compared using survival analyses that included the treatment group effect and the weight resulting from the matching (15,16). For time-to-event analyses, application of propensity scores using inverse probability weighting (IPW), rather than matching, stratification, or adjustment, produces effect estimates with minimal bias (17). We used adjusted survival curves and log-rank test based on IPW proposed by Xie and Liu (18) and implemented in the IPW survival package. We used the software R, version 3.6.2 (R Core Team, 2019), to perform the statistical analyses.