Semiparametric models are often considered for analyzing longitudinal data a good balance between flexibility and parsimony. In this paper, we study class of marginal partially linear quantile with possibly varying coefficients. The functional coefficients estimated by basis function approximations. estimation procedure is easy to implement, it requires no specification the error distributions. asymptotic properties proposed estimators established as well constant We develop rank score tests...

We consider the penalized generalized estimating equations (GEEs) for analyzing longitudinal data with high-dimensional covariates, which often arise in microarray experiments and large-scale health studies. Existing regression procedures assume independent rely on likelihood function. Construction of a feasible joint function is challenging, particularly correlated discrete outcome data. The GEE procedure only requires specifying first two marginal moments working correlation structure....

Abstract We consider the generalized additive model when responses from same cluster are correlated. Incorporating correlation in estimation of nonparametric components for is important because it improves efficiency and increases statistical power selection. In our setting, there no specified likelihood function model, outcomes could be nonnormal discrete, which makes selection very challenging problems. propose consistent that incorporate structure. establish an asymptotic property with...

Identifying an informative correlation structure is important in improving estimation efficiency for longitudinal data. We approximate the empirical estimator of matrix by groups known basis matrices that represent different structures, and transform selection problem to a covariate problem. To address both complexity informativeness matrix, we minimize objective function consists two parts: difference between information model approximation penalty penalizes models with too many matrices....

We propose a shrinkage method to estimate the coefficient function in functional linear regression model when value of is zero within certain sub-regions. Besides identifying null region which zero, we also aim perform estimation and inferences for nonparametrically estimated without over-shrinking values. Our proposal consists two stages. In stage one, Dantzig selector employed provide initial location region. two, group SCAD approach refine inference procedures function. considerations...

Abstract Sociability is crucial for survival, whereas social avoidance a feature of disorders such as Rett syndrome, which caused by loss-of-function mutations in MECP2 . To understand how preference interactions encoded, we used vivo calcium imaging to compare medial prefrontal cortex (mPFC) activity female wild-type and Mecp2 -heterozygous mice during three-chamber tests. We found that mPFC pyramidal neurons -deficient are hypo-responsive both nonsocial stimuli. Hypothesizing this limited...

Quantile regression has emerged as a powerful tool in survival analysis it directly links the quantiles of patients' times to their demographic and genomic profiles, facilitating identification important prognostic factors. In view limited work on variable selection context, we develop new adaptive-lasso-based procedure for quantile with censored outcomes. To account random censoring data multivariate covariates, employ ideas redistribution-of-mass e ective dimension reduction....

The ``curse of dimensionality'' has remained a challenge for high-dimensional data analysis in statistics. sliced inverse regression (SIR) and canonical correlation (CANCOR) methods aim to reduce the dimensionality by replacing explanatory variables with small number composite directions without losing much information. However, estimated generally involve all variables, making their interpretation difficult. To simplify direction estimates, Ni, Cook Tsai [Biometrika 92 (2005) 242--247]...

In this paper, we introduce a family of robust estimates for the parametric and nonparametric components under generalized partially linear model, where data are modeled by $y_i|(\mathbf{x}_i,t_i)\sim F(\cdot,\mu_i)$ with $\mu_i=H(\eta(t_i)+\mathbf{x}_i^{$\mathrm{T}$}\beta)$, some known distribution function F link H. It is shown that $\beta$ root-n consistent asymptotically normal. Through Monte Carlo study, performance these estimators compared classical ones.

Environmental Enteropathy (EE) is a subclinical condition caused by constant fecal-oral contamination and resulting in blunting of intestinal villi inflammation. Of primary interest the clinical research to evaluate association between non-invasive EE biomarkers malnutrition cohort Bangladeshi children. The challenges are that number biomarkers/covariates relatively large, some them highly correlated. Many variable selection methods available literature, but which most appropriate for...

Early childhood is a critical stage of physical and cognitive growth that forms the foundation future wellbeing. Stunted presented in one every 4 children worldwide contributes to developmental impairment under-five mortality. Better understanding early patterns should allow for detection intervention malnutrition. We aimed characterize child quantify change curves from World Health Organization (WHO) Child Growth Standards. In cohort 626 Bangladesh children, longitudinal height-for-age...

In this paper, we propose a quantile approach to the multi-index semiparametric model for an ordinal response variable. Permitting non-parametric transformation of response, proposed method achieves root-n rate convergence and has attractive robustness properties. Further, allows additional indices remaining correlations between covariates residuals from single-index, considerably reducing error variance thus leading more efficient prediction intervals (PIs). The utility is demonstrated by...

Covariate-adaptive designs are widely used to balance covariates and maintain randomization in clinical trials. Adaptive for discrete their asymptotic properties have been well studied the literature. However, important continuous often involved studies. Simply discretizing or categorizing can result loss of information. The current understanding adaptive with lacks a theoretical foundation as existing works entirely based on simulations. Consequently, conventional hypothesis testing trials...

Motivated by a genetic investigation on the progressive decline in renal function clinical trial study of kidney disease, we develop practical test for evaluating group difference trajectories under semi-parametric modeling framework. For temporal patterns or longitudinal data, B-splines are used to approximate non-parametrically. Such approximation asymptotically converts problem testing trajectory into significance regression coefficients that can be simply estimated generalized estimating...

In prior work, we analyzed the GridFTP usage logs collected by data transfer nodes (DTNs) located at national scientific computing centers, and found significant throughput variance even among transfers between same two end hosts. The goal of this work is to quantify impact various factors on variance. Our methodology consisted executing experiments a high-speed research testbed, running large-sized instrumented operational DTNs, creating statistical models from measurements. A non-linear...

Abstract Correlation structure selection for non‐normal longitudinal data is very challenging diverging cluster size because of the high‐dimensional correlation parameters involved and complexity likelihood function data. However identifying correct important it can improve estimation efficiency testing power We propose to approximate inverse empirical matrix using a linear combination candidate basis matrices, select by non‐zero coefficients matrices. This carried out minimizing penalized...

Growth curve models have been widely used to analyse longitudinal data in social and behavioural sciences. Although growth with normality assumptions are relatively easy estimate, practical rarely normal. Failing account for non‐normal may lead unreliable model estimation misleading statistical inference. In this work, we propose a robust approach modelling using conditional medians that less sensitive outlying observations. Bayesian methods applied Based on the existing work quantile...

Alternating direction method of multipliers (ADMM) has been recognized as an efficient approach for solving many large-scale machine learning problems. However, the ADMM under master-slave mode suffers from several limitations, e.g., can't make full use multi-core cluster environment and single master load is too heavy, resulting in huge time overhead. In this paper, we propose a hierarchical communication structure. Since intra-node communications mostly shared memory, divide processes same...