SEARCH · PubMed Health

Results for “doubly robust estimation”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

9 recordsLinked to original sources

PubMed2005

Doubly robust estimation in missing data and causal inference models.

The goal of this article is to construct doubly robust (DR) estimators in ignorable missing data and causal inference models. In a missing data model, an estimator is DR if it remains consistent when either (but not necessarily both) a model for the missingness mechanism or a model for the distribution of the complete data is correctly specified. Because with observational data one can never be sure that either a missingness model or a complete data model is correct, perhaps the best that can be hoped for is to find a DR estimator. DR estimators, in contrast to standard likelihood-based or (nonaugmented) inverse probability-weighted estimators, give the analyst two chances, instead of only one, to make a valid inference. In a causal inference model, an estimator is DR if it remains consistent when either a model for the treatment assignment mechanism or a model for the distribution of the counterfactual data is correctly specified. Because with observational data one can never be sure that a model for the treatment assignment mechanism or a model for the counterfactual data is correct, inference based on DR estimators should improve upon previous approaches. Indeed, we present the results of simulation studies which demonstrate that the finite sample performance of DR estimators is as impressive as theory would predict. The proposed method is applied to a cardiovascular clinical trial.

Antidepressive Agents↗

PubMed2025

Causal Inference for Genomic Data with Multiple Heterogeneous Outcomes.

With the evolution of single-cell RNA sequencing techniques into a standard approach in genomics, it has become possible to conduct cohort-level causal inferences based on single-cell-level measurements. However, the individual gene expression levels of interest are not directly observable; instead, only repeated proxy measurements from each individual's cells are available, providing a derived outcome to estimate the underlying outcome for each of many genes. In this paper, we propose a generic semiparametric inference framework for doubly robust estimation with multiple derived outcomes, which also encompasses the usual setting of multiple outcomes when the response of each unit is available. To reliably quantify the causal effects of heterogeneous outcomes, we specialize the analysis to standardized average treatment effects and quantile treatment effects. Through this, we demonstrate the use of the semiparametric inferential results for doubly robust estimators derived from both Von Mises expansions and estimating equations. A multiple testing procedure based on Gaussian multiplier bootstrap is tailored for doubly robust estimators to control the false discovery exceedance rate. Applications in single-cell CRISPR perturbation analysis and individual-level differential expression analysis demonstrate the utility of the proposed methods and offer insights into the usage of different estimands for causal inference in genomics.

Derived outcomes↗

PubMed2001

Augmented inverse probability weighted estimator for Cox missing covariate regression.

This article investigates an augmented inverse selection probability weighted estimator for Cox regression parameter estimation when covariate variables are incomplete. This estimator extends the Horvitz and Thompson (1952, Journal of the American Statistical Association 47, 663-685) weighted estimator. This estimator is doubly robust because it is consistent as long as either the selection probability model or the joint distribution of covariates is correctly specified. The augmentation term of the estimating equation depends on the baseline cumulative hazard and on a conditional distribution that can be implemented by using an EM-type algorithm. This method is compared with some previously proposed estimators via simulation studies. The method is applied to a real example.

Analysis of Variance↗

PubMed2003

Generalized additive selection models for the analysis of studies with potentially nonignorable missing outcome data.

Rotnitzky, Robins, and Scharfstein (1998, Journal of the American Statistical Association 93, 1321-1339) developed a methodology for conducting sensitivity analysis of studies in which longitudinal outcome data are subject to potentially nonignorable missingness. In their approach, they specify a class of fully parametric selection models, indexed by a non- or weakly identified selection bias function that indicates the degree to which missingness depends on potentially unobservable outcomes. Estimation of the parameters of interest proceeds by varying the selection bias function over a range considered plausible by subject-matter experts. In this article, we focus on cross-sectional, univariate outcome data and extend their approach to a class of semiparametric selection models, using generalized additive restrictions. We propose a backfitting algorithm to estimate the parameters of the generalized additive selection model. For estimation of the mean outcome, we propose three types of estimating functions: simple inverse weighted, doubly robust, and orthogonal. We present the results of a data analysis and a simulation study.

Acquired Immunodeficiency Syndrome↗

PubMed2026

A doubly robust framework for addressing outcome-dependent selection bias in multi-cohort EHR studies.

Selection bias can hinder accurate estimation of association parameters in binary disease risk models using non-probability samples like electronic health records (EHRs). The issue is compounded when participants are recruited from multiple clinics/centers with varying selection mechanisms that may depend on the disease/outcome of interest. Traditional inverse-probability-weighted (IPW) methods, based on constructed parametric selection models, often struggle with misspecifications when selection mechanisms vary across cohorts. This paper introduces a new Joint Augmented Inverse Probability Weighted (JAIPW) method, which integrates individual-level data from multiple cohorts collected under potentially outcome-dependent selection mechanisms, with data from an external probability sample. JAIPW offers double robustness by incorporating a flexible auxiliary score model to address potential misspecifications in the selection models. We outline the asymptotic properties of the JAIPW estimator, and our simulations reveal that JAIPW achieves up to 6 times lower relative bias and 5 times lower root mean square error (RMSE) compared to the best performing joint IPW methods under scenarios with misspecified selection models. Applying JAIPW to the Michigan Genomics Initiative (MGI), a multi-clinic EHR-linked biobank, combined with external national probability samples, resulted in cancer-sex association estimates closely aligned with national benchmark estimates. We also analyzed the association between cancer and polygenic risk scores (PRS) in MGI to illustrate a situation where the exposure variable is not measured in the external probability sample.

Selection Bias↗

PubMed1996

Human energy expenditure in affluent societies: an analysis of 574 doubly-labelled water measurements.

OBJECTIVES: To describe average levels of free-living energy expenditure in people from affluent societies and to determine the influence of body weight, height, age and sex. DESIGN: Analysis of 574 measurements of total energy expenditure (TEE, assessed by the doubly-labelled water method); basal metabolic rate (BMR, directly measured or derived from similar directly measured proxy measures such as during sleep); activity energy expenditure (AEE, derived as TEE-BMR); and physical activity level (PAL, derived as TEE/BMR) from people aged 2-95 years. The dataset was extracted from 1614 published and unpublished measurements in 1156 subjects after exclusion of repeat estimates and subjects in special physiological or behavioural states (eg pregnancy, athletic or military training etc). RESULTS: A separate analysis of data from non-ambulant subjects, and from elite endurance athletes (all excluded from the main dataset) established the limits of human daily energy expenditure at around 1.2 x BMR and 4.5 x BMR. In the main analysis, the validity of PAL as an index of TEE adjusted for BMR was tested and confirmed. Regression equations were then derived to describe TEE, BMR, AEE and PAL in terms of body weight, height, age and sex. As anticipated, TEE, BMR and AEE were all positively related to weight and height, while age was a negative predictor, especially of activity. The influence of weight disappeared when TEE was expressed as PAL, but height and age remained as highly significant predictors. For all three components, females expended 11% less energy on average than males after adjustment for weight, height and age. Average levels of energy expenditure in different age and sex groups are tabulated. CONCLUSIONS: There now exists a large and robust database of energy expenditure measurements obtained by the doubly-labelled water method. Analysis of the data from affluent societies shows that, in general, levels of energy expenditure are similar to the recommendations for energy requirements adopted by FAO/WHO/UNU (1985) and UK Department of Health (1991). PAL values for active subjects tend to be higher than is currently assumed. The current analysis provides a substantial body of normal data against which other estimates can be compared.

Adolescent↗

PubMed2005

Void formation and roughening in slow fracture.

Slow crack propagation in ductile, and in certain brittle materials, appears to take place via the nucleation of voids ahead of the crack tip due to plastic yields, followed by the coalescence of these voids. Postmortem analysis of the resulting fracture surfaces of ductile and brittle materials on the microm-mm and the nm scales, respectively, reveals self-affine cracks with anomalous scaling exponent zeta approximately = 0.8 in 3 dimensions and zeta approximately = 0.65 in 2 dimensions. In this paper we present an analytic theory based on the method of iterated conformal maps aimed at modelling the void formation and the fracture growth, culminating in estimates of the roughening exponents in 2 dimensions. In the simplest realization of the model we allow one void ahead of the crack, and address the robustness of the roughening exponent. Next we develop the theory further, to include two voids ahead of the crack. This development necessitates generalizing the method of iterated conformal maps to include doubly connected regions (maps from the annulus rather than the unit circle). While mathematically and numerically feasible, we find that the employment of the stress field as computed from elasticity theory becomes questionable when more than one void is explicitly inserted into the material. Thus further progress in this line of research calls for improved treatment of the plastic dynamics.

Journal Article↗

PubMed1999

Field energetics and the estimation of pollen and nectar intake in the marsupial honey possum, Tarsipes rostratus, in heathland habitats of south-western Australia.

A method is described, based on the simultaneous turnover of both stable (18O) and radioactive isotopes (3H and 22Na), whereby the daily nectar and pollen intake of free-ranging marsupial honey possums (Tarsipes rostratus) may be estimated. The field metabolic rate is measured using doubly labelled water and nectar intake is estimated independently from the measured water and sodium fluxes. The method assumes that free-water intake is negligible (but may be accounted for if not the case), that virtually all dietary sodium is derived from nectar rather than from pollen, and that the animals are in energetic balance over the period of measurement. These assumptions have been tested and found to be robust, except during periods of heavy rain when significant intakes of free-water were recorded. Leaching experiments with pollen grains suggest that less than 10% of the sodium ingested by honey possums is derived from pollen and calculations thus assumed a 90%:10% split between nectar and pollen. Nectar intake averaged 5.9 +/- 0.6 ml.day-1 and regressing nectar intake on daily change in body mass predicts an intake of approximately 7 ml.day-1 nectar to maintain balance for a 9 g honey possum. Estimates of pollen intake averaged 660 +/- 156 mg.day-1 and a similar regression analysis of the data predicts that a daily intake of approximately 1 g pollen would be needed to maintain mass balance of honey possums. Estimated nectar and pollen intakes did not differ significantly between males and females, but nectar intake was higher in winter compared with dry periods of the year. The sugar content of nectar falls during winter, however, and the overall energy derived from nectar thus remains roughly constant. Estimates of pollen and nectar intake for individual animals were not significantly correlated, suggesting that honey possums forage selectively for these two food items.

Animals↗

PubMed2002

Validation of the DLW method in Japanese quail at different water fluxes using laser and IRMS.

In Japanese quail (Coturnix c. japonica; n = 9), the doubly labeled water (DLW) method ((2)H, (18)O) for estimation of CO(2) production (l/day) was validated. To evaluate its sensitivity to water efflux levels (r(H(2))O(e); g/day) and to assumptions of fractional evaporative water loss (x; dimensionless), animals were repeatedly fed a dry pellet diet (average r(H(2))O(e) of 34.8 g/day) or a wet mash diet (95.8 g/day). We simultaneously compared the novel infrared laser spectrometry (LS) with isotope ratio mass spectrometry. At low r(H(2))O(e), calculated CO(2) production rate exhibited little sensitivity to assumptions concerning x, with the best fit being found at 0.51, and only little error was made employing an x value of 0.25. In contrast, at high r(H(2))O(e), sensitivities were much higher with the best fit at x = 0.32. Conclusions derived from isotope ratio mass spectrometry and LS were similar, proving the usefulness of LS. Within a threefold range of r(H(2))O(e), little error in the DLW method is made when assuming one single x value of 0.25 (recommended by Speakman JR, Doubly Labelled Water. Theory and Practice. London: Chapman & Hall, 1997), indicating its robustness in comparative studies.

Animal Feed↗