A5006102391- Statistical Methods and Inference
- Statistical Methods in Clinical Trials
- Advanced Causal Inference Techniques
- Statistical Methods and Bayesian Inference
- Advanced Statistical Methods and Models
- HIV Research and Treatment
- Optimal Experimental Design Methods
- Bayesian Methods and Mixture Models
- Health Systems, Economic Evaluations, Quality of Life
- Vaccine Coverage and Hesitancy
- Statistics Education and Methodologies
- Fault Detection and Control Systems
- Statistical Distribution Estimation and Applications
- Advanced Statistical Process Monitoring
- HIV/AIDS drug development and treatment
- Pesticide Residue Analysis and Safety
- Spectroscopy and Chemometric Analyses
- Hepatitis C virus research
- COVID-19 epidemiological studies
- Scientific Measurement and Uncertainty Evaluation
- Genetic and phenotypic traits in livestock
- Advanced Biosensing Techniques and Applications
- HIV/AIDS Research and Interventions
- Mathematical and Theoretical Epidemiology and Ecology Models
- SARS-CoV-2 and COVID-19 Research
North Carolina State University
2014-2024
Duke University
2014-2021
North Central State College
2014-2017
John Wiley & Sons (United Kingdom)
2015
Duke University Hospital
2014
Clinical Research Institute
2002-2014
Duke Medical Center
2004-2014
American Statistical Association
2012
Hasselt University
2009
KU Leuven
2009
(2003). Hierarchical Linear Models: Applications and Data Analysis Methods. Journal of the American Statistical Association: Vol. 98, No. 463, pp. 767-768.
Estimation of treatment effects with causal interpretation from observational data is complicated because exposure to may be confounded subject characteristics. The propensity score, the probability conditional on covariates, basis for two approaches adjusting confounding: methods based stratification observations by quantiles estimated scores and weighting inverse scores. We review popular versions these related offering improved precision, describe theoretical properties highlight their...
Nonlinear mixed effects models for data in the form of continuous, repeated measurements on each a number individuals, also known as hierarchical nonlinear models, are popular platform analysis when interest focuses individual-specific characteristics.This framework first enjoyed widespread attention within statistical research community late 1980s, and 1990s saw vigorous development new methodological computational techniques these emergence general-purpose software, broad application...
Doubly robust estimation combines a form of outcome regression with model for the exposure (i.e., propensity score) to estimate causal effect an on outcome. When used individually effect, both and score methods are unbiased only if statistical is correctly specified. The doubly estimator these 2 approaches such that 1 models need be specified obtain estimator. In this introduction estimators, authors present conceptual overview estimation, simple worked example, results from simulation study...
Abstract Heteroscedastic regression models are used in fields including economics, engineering, and the biological physical sciences. Often, heteroscedasticity is modeled as a function of covariates or other structural parameters. Standard asymptotic theory implies that how one estimates variance function, particular parameters, has no effect on first-order properties parameter estimates; there evidence, however, both practice higher-order to suggest does matter. Further, some settings,...
Summary A treatment regime is a rule that assigns treatment, among set of possible treatments, to patient as function his/her observed characteristics, hence “personalizing” the patient. The goal identify optimal that, if followed by entire population patients, would lead best outcome on average. Given data from clinical trial or observational study, for single decision, can be found assuming regression model expected conditional and covariates, where, given one yields most favorable...
Abstract There is considerable debate regarding whether and how covariate‐adjusted analyses should be used in the comparison of treatments randomized clinical trials. Substantial baseline covariate information routinely collected such trials, one goal adjustment to exploit covariates associated with outcome increase precision estimation treatment effect. However, concerns are raised over potential for bias when selected post hoc based on a model relationship between outcome, covariates,...
Considerable recent interest has focused on doubly robust estimators for a population mean response in the presence of incomplete data, which involve models both propensity score and regression outcome covariates. The usual estimator may yield severely biased inferences if neither these is correctly specified can exhibit nonnegligible bias estimated close to zero some observations. We propose alternative that achieve comparable or improved performance relative existing methods, even with scores zero.
A treatment regime maps observed patient characteristics to a recommended treatment. Recent technological advances have increased the quality, accessibility, and volume of patient-level data; consequently, there is growing need for powerful flexible estimators an optimal that can be used with either observational or randomized clinical trial data. We propose novel general framework transforms problem estimating into classification wherein classifier corresponds regime. show commonly employed...
In clinical practice, physicians make a series of treatment decisions over the course patient's disease based on his/her baseline and evolving characteristics. A dynamic regime is set sequential decision rules that operationalizes this process. Each rule corresponds to point dictates next action accrued information. Using existing data, key goal estimating optimal regime, that, if followed by patient population, would yield most favorable outcome average. Q- A-learning are two main...
Summary. Normality of random effects is a routine assumption for the linear mixed model, but it may be unrealistic, obscuring important features among-individual variation. We relax this by approximating density seminonparameteric (SNP) representation Gallant and Nychka (1987, Econometrics55, 363–390), which includes normality as special case provides flexibility in capturing broad range nonnormal behavior, controlled user-chosen tuning parameter. An advantage that marginal likelihood...
Summary. Joint models for a time-to-event (e.g., survival) and longitudinal response have generated considerable recent interest. The data are assumed to follow mixed effects model, proportional hazards model depending on the random other covariates is survival endpoint. Interest may focus inference process, which informatively censored, or hazard relationship. Several methods fitting such been proposed, most requiring parametric distributional assumption (normality) effects. A natural...
Many studies have shown that patients infected with hepatitis C virus (HCV) of genotype 2 better response to interferon (IFN)-α treatment than 1 patients; however, the mechanisms responsible for this difference are not understood. In study, viral dynamics during high-dose IFN induction were compared between genotypes. Patients in each group received 10 MU IFN-α2b 14 days, and HCV RNA levels frequently determined. Nonlinear fitting, both individually patient using a mixed-effects approach,...
Journal Article The nonlinear mixed effects model with a smooth random density Get access MARIE DAVIDIAN, DAVIDIAN Department of Statistics, North Carolina State UniversityCampus Box 8203, Raleigh, 27695-8203, U. S.A. Search for other works by this author on: Oxford Academic Google Scholar A. RONALD GALLANT Biometrika, Volume 80, Issue 3, September 1993, Pages 475–488, https://doi.org/10.1093/biomet/80.3.475 Published: 01 1993
Abstract Heteroscedastic regression models are used in fields including economics, engineering, and the biological physical sciences. Often, heteroscedasticity is modeled as a function of covariates or other structural parameters. Standard asymptotic theory implies that how one estimates variance function, particular parameters, has no effect on first-order properties parameter estimates; there evidence, however, both practice higher-order to suggest does matter. Further, some settings,...
Journal Article Robust estimation of optimal dynamic treatment regimes for sequential decisions Get access Baqun Zhang, Zhang School Statistics, Renmin University China, Beijing 100872, zhangbaqun@ruc.edu.cn Search other works by this author on: Oxford Academic Google Scholar Anastasios A. Tsiatis, Tsiatis Department North Carolina State University, Raleigh, Carolina, 27695-8203, U.S.A., tsiatis@ncsu.edu Eric B. Laber, Laber eblaber@ncsu.edu Marie Davidian davidian@ncsu.edu Biometrika,...
Rationale: Lung transplantation is an accepted and increasingly employed treatment for advanced lung diseases, but the anticipated survival benefit of poorly understood.Objectives: To determine whether which patients confers a in modern era U.S. allocation.Methods: Data on 13,040 adults listed between May 2005 September 2011 were obtained from United Network Organ Sharing. A structural nested accelerated failure time model was used to over time. The effects patient, donor, transplant center...
Summary A treatment regime formalizes personalized medicine as a function from individual patient characteristics to recommended treatment. high-quality can improve outcomes while reducing cost, resource consumption, and burden. Thus, there is tremendous interest in estimating regimes observational randomized studies. However, the development of for application clinical practice requires long-term, joint effort statisticians scientists. In this collaborative process, statistician must...