Damla Şentürk

Inference for covariate adjusted regression via varying coefficient models

We consider covariate adjusted regression (CAR), a regression method for situations where predictors and response are observed after being distorted by a multiplicative factor. The distorting factors are unknown functions of an observable covariate, where one specific distorting function is associated with each predictor or response. The dependence of both response and predictors on the same confounding covariate may alter the underlying regression relation between undistorted but unobserved predictors and response. We consider a class of highly flexible adjustment methods for parameter estimation in the underlying regression model, which is the model of interest. Asymptotic normality of the estimates is obtained by establishing a connection to varying coefficient models. These distribution results combined with proposed consistent estimates of the asymptotic variance are used for the construction of asymptotic confidence intervals for the regression coefficients. The proposed approach is illustrated with data on serum creatinine, and finite sample properties of the proposed procedures are investigated through a simulation study.We consider covariate adjusted regression (CAR), a regression method for situations where predictors and response are observed after being distorted by a multiplicative factor. The distorting factors are unknown functions of an observable covariate, where one specific distorting function is associated with each predictor or response. The dependence of both response and predictors on the same confounding covariate may alter the underlying regression relation between undistorted but unobserved predictors and response. We consider a class of highly flexible adjustment methods for parameter estimation in the underlying regression model, which is the model of interest. Asymptotic normality of the estimates is obtained by establishing a connection to varying coefficient models. These distribution results combined with proposed consistent estimates of the asymptotic variance are used for the construction of asymptotic confidence intervals for the regression coefficients. The proposed approach is illustrated with data on serum creatinine, and finite sample properties of the proposed procedures are investigated through a simulation study.

Multicovariate-adjusted regression models

We introduce multicovariate-adjusted regression (MCAR), an adjustment method for regression analysis, where both the response (Y) and predictors (X 1, …, X p ) are not directly observed. The available data have been contaminated by unknown functions of a set of observable distorting covariates, Z 1, …, Z s , in a multiplicative fashion. The proposed method substantially extends the current contaminated regression modelling capability, by allowing for multiple distorting covariate effects. MCAR is a flexible generalisation of the recently proposed covariate-adjusted regression method, an effective adjustment method in the presence of a single covariate, Z. For MCAR estimation, we establish a connection between the MCAR models and adaptive varying coefficient models. This connection leads to an adaptation of a hybrid backfitting estimation algorithm. Extensive simulations are used to study the performance and limitations of the proposed iterative estimation algorithm. In particular, the bias and mean square error of the proposed MCAR estimators are examined, relative to a baseline and a consistent benchmark estimator. The method is also illustrated with a Pima Indian diabetes data set, where the response and predictors are potentially contaminated by body mass index and triceps skin fold thickness. Both distorting covariates measure aspects of obesity, an important risk factor in type 2 diabetes.

Modeling time-varying effects with generalized and unsynchronized longitudinal data.

We propose novel estimation approaches for generalized varying coefficient models that are tailored for unsynchronized, irregular and infrequent longitudinal designs/data. Unsynchronized longitudinal data refer to the time-dependent response and covariate measurements for each individual measured at distinct time points. Data from the Comprehensive Dialysis Study motivate the proposed methods. We model the potential age-varying association between infection-related hospitalization status and the inflammatory marker, C-reactive protein, within the first 2 years from initiation of dialysis. We cannot directly apply traditional longitudinal modeling to unsynchronized data, and no method exists to estimate time-varying or age-varying effects for generalized outcomes (e.g., binary or count data) to date. In addition, through the analysis of the Comprehensive Dialysis Study data and simulation studies, we show that preprocessing steps, such as binning, needed to synchronize data to apply traditional modeling can lead to significant loss of information in this context. In contrast, the proposed approaches discard no observation; they exploit the fact that although there is little information in a single subject trajectory because of irregularity and infrequency, the moments of the underlying processes can be accurately and efficiently recovered by pooling information from all subjects using functional data analysis. We derive subject-specific mean response trajectory predictions and study finite sample properties of the estimators.

Cardiovascular event risk dynamics over time in older patients on dialysis: a generalized multiple-index varying coefficient model approach.

Among patients on dialysis, cardiovascular disease and infection are leading causes of hospitalization and death. Although recent studies have found that the risk of cardiovascular events is higher after an infection-related hospitalization, studies have not fully elucidated how the risk of cardiovascular events changes over time for patients on dialysis. In this work, we characterize the dynamics of cardiovascular event risk trajectories for patients on dialysis while conditioning on survival status via multiple time indices: (1) time since the start of dialysis, (2) time since the pivotal initial infection-related hospitalization, and (3) the patients age at the start of dialysis. This is achieved by using a new class of generalized multiple-index varying coefficient (GM-IVC) models. The proposed GM-IVC models utilize a multiplicative structure and one-dimensional varying coefficient functions along each time and age index to capture the cardiovascular risk dynamics before and after the initial infection-related hospitalization among the dynamic cohort of survivors. We develop a two-step estimation procedure for the GM-IVC models based on local maximum likelihood. We report new insights on the dynamics of cardiovascular events risk using the United States Renal Data System database, which collects data on nearly all patients with end-stage renal disease in the United States. Finally, simulation studies assess the performance of the proposed estimation procedures.

Covariate Adjusted Correlation Analysis with Application to FMR1 Premutation Female Carrier Data

Motivated by molecular data on female premutation carriers of the fragile X mental retardation 1 (FMR1) gene, we present a new method of covariate adjusted correlation analysis to examine the association of messenger RNA (mRNA) and number of CGG repeat expansion in the FMR1 gene. The association between the molecular variables in female carriers needs to adjust for activation ratio (ActRatio), a measure which accounts for the protective effects of one normal X chromosome in females carriers. However, there are inherent uncertainties in the exact effects of ActRatio on the molecular measures of interest. To account for these uncertainties, we develop a flexible adjustment that accommodates both additive and multiplicative effects of ActRatio nonparametrically. The proposed adjusted correlation uses local conditional correlations, which are local method of moments estimators, to estimate the Pearson correlation between two variables adjusted for a third observable covariate. The local method of moments estimators are averaged to arrive at the final covariate adjusted correlation estimator, which is shown to be consistent. We also develop a test to check the nonparametric joint additive and multiplicative adjustment form. Simulation studies illustrate the efficacy of the proposed method. (Application to FMR1 premutation data on 165 female carriers indicates that the association between mRNA and CGG repeat after adjusting for ActRatio is stronger.) Finally, the results provide independent support for a specific jointly additive and multiplicative adjustment form for ActRatio previously proposed in the literature.

Measurement Error Case Series Models With Application to Infection-Cardiovascular Risk in Older Patients on Dialysis

Infection and cardiovascular disease are leading causes of hospitalization and death in older patients on dialysis. Our recent work found an increase in the relative incidence of cardiovascular outcomes during the ∼ 30 days after infection-related hospitalizations using the case series model, which adjusts for measured and unmeasured baseline confounders. However, a major challenge in modeling/assessing the infection-cardiovascular risk hypothesis is that the exact time of infection, or more generally “exposure,” onsets cannot be ascertained based on hospitalization data. Only imprecise markers of the timing of infection onsets are available. Although there is a large literature on measurement error in the predictors in regression modeling, to date, there is no work on measurement error on the timing of a time-varying exposure to our knowledge. Thus, we propose a new method, the measurement error case series (MECS) models, to account for measurement error in time-varying exposure onsets. We characterized the general nature of bias resulting from estimation that ignores measurement error and proposed a bias-corrected estimation for the MECS models. We examined in detail the accuracy of the proposed method to estimate the relative incidence of cardiovascular events. Hospitalization data from the United States Renal Data System, which captures nearly all (>99%) patients with end-stage renal disease in the United States over time, are used to illustrate the proposed method. The results suggest that the estimate of the relative incidence of cardiovascular events during the 30 days after infections, a period where acute effects of infection on vascular endothelium may be most pronounced, is substantially attenuated in the presence of infection onset measurement error.

A Multi-Dimensional Functional Principal Components Analysis of EEG Data

The electroencephalography (EEG) data created in event-related potential (ERP) experiments have a complex high-dimensional structure. Each stimulus presentation, or trial, generates an ERP waveform which is an instance of functional data. The experiments are made up of sequences of multiple trials, resulting in longitudinal functional data and moreover, responses are recorded at multiple electrodes on the scalp, adding an electrode dimension. Traditional EEG analyses involve multiple simplifications of this structure to increase the signal-to-noise ratio, effectively collapsing the functional and longitudinal components by identifying key features of the ERPs and averaging them across trials. Motivated by an implicit learning paradigm used in autism research in which the functional, longitudinal, and electrode components all have critical interpretations, we propose a multidimensional functional principal components analysis (MD-FPCA) technique which does not collapse any of the dimensions of the ERP data. The proposed decomposition is based on separation of the total variation into subject and subunit level variation which are further decomposed in a two-stage functional principal components analysis. The proposed methodology is shown to be useful for modeling longitudinal trends in the ERP functions, leading to novel insights into the learning patterns of children with Autism Spectrum Disorder (ASD) and their typically developing peers as well as comparisons between the two groups. Finite sample properties of MD-FPCA are further studied via extensive simulations.

Robust functional clustering of ERP data with application to a study of implicit learning in autism

Motivated by a study on visual implicit learning in young children with Autism Spectrum Disorder (ASD), we propose a robust functional clustering (RFC) algorithm to identify subgroups within electroencephalography (EEG) data. The proposed RFC is an iterative algorithm based on functional principal component analysis, where cluster membership is updated via predictions of the functional trajectories obtained through a non-parametric random effects model. We consider functional data resulting from event-related potential (ERP) waveforms representing EEG time-locked to stimuli over the course of an implicit learning experiment, after applying a previously proposed meta-preprocessing step. This meta-preprocessing is designed to increase the low signal-to-noise ratio in the raw data and to mitigate the longitudinal changes in the ERP waveforms which characterize the nature and speed of learning. The resulting functional ERP components (peak amplitudes and latencies) inherently exhibit covariance heterogeneity due to low data quality over some stimuli inducing the averaging of different numbers of waveforms in sliding windows of the meta-preprocessing step. The proposed RFC algorithm incorporates this known covariance heterogeneity into the clustering algorithm, improving cluster quality, as illustrated in the data application and extensive simulation studies. ASD is a heterogeneous syndrome and identifying subgroups within ASD children is of interest for understanding the diverse nature of this complex disorder. Applications to the implicit learning paradigm identify subgroups within ASD and typically developing children with diverse learning patterns over the course of the experiment, which may inform clinical stratification of ASD.

Identifying longitudinal trends within EEG experiments.

Differential brain response to sensory stimuli is very small (a few microvolts) compared to the overall magnitude of spontaneous electroencephalogram (EEG), yielding a low signal-to-noise ratio (SNR) in studies of event-related potentials (ERP). To cope with this phenomenon, stimuli are applied repeatedly and the ERP signals arising from the individual trials are averaged at the subject level. This results in loss of information about potentially important changes in the magnitude and form of ERP signals over the course of the experiment. In this article, we develop a meta-preprocessing step utilizing a moving average of ERP across sliding trial windows, to capture such longitudinal trends. We embed this procedure in a weighted linear mixed effects model to describe longitudinal trends in features such as ERP peak amplitude and latency across trials while adjusting for the inherent heteroskedasticity created at the meta-preprocessing step. The proposed unified framework, including the meta-processing and the weighted linear mixed effects modeling steps, is referred to as MAP-ERP (moving-averaged-processed ERP). We perform simulation studies to assess the performance of MAP-ERP in reconstructing existing longitudinal trends and apply MAP-ERP to data from young children with autism spectrum disorder (ASD) and their typically developing counterparts to examine differences in patterns of implicit learning, providing novel insights about the mechanisms underlying social and/or cognitive deficits in this disorder.

Time‐varying effect modeling with longitudinal data truncated by death: conditional models, interpretations, and inference

Recent studies found that infection-related hospitalization was associated with increased risk of cardiovascular (CV) events, such as myocardial infarction and stroke in the dialysis population. In this work, we develop time-varying effects modeling tools in order to examine the CV outcome risk trajectories during the time periods before and after an initial infection-related hospitalization. For this, we propose partly conditional and fully conditional partially linear generalized varying coefficient models (PL-GVCMs) for modeling time-varying effects in longitudinal data with substantial follow-up truncation by death. Unconditional models that implicitly target an immortal population is not a relevant target of inference in applications involving a population with high mortality, like the dialysis population. A partly conditional model characterizes the outcome trajectory for the dynamic cohort of survivors, where each point in the longitudinal trajectory represents a snapshot of the population relationships among subjects who are alive at that time point. In contrast, a fully conditional approach models the time-varying effects of the population stratified by the actual time of death, where the mean response characterizes individual trends in each cohort stratum. We compare and contrast partly and fully conditional PL-GVCMs in our aforementioned application using hospitalization data from the United States Renal Data System. For inference, we develop generalized likelihood ratio tests. Simulation studies examine the efficacy of estimation and inference procedures.