Statistics in Biosciences最新文献

This paper develops a Bayesian model with a flexible link function connecting a binary treatment response to a linear combination of covariates and a treatment indicator and the interaction between the two. Generalized linear models allowing data-driven link functions are often called "single-index models" and are among popular semi-parametric modeling methods. In this paper, we focus on modeling heterogeneous treatment effects, with the goal of developing a treatment benefit index (TBI) incorporating prior information from historical data. The model makes inference on a composite moderator of treatment effects, summarizing the effect of the predictors within a single variable through a linear projection of the predictors. This treatment benefit index can be useful for stratifying patients according to their predicted treatment benefit levels and can be especially useful for precision health applications. The proposed method is applied to a COVID-19 treatment study.

Time-dependent covariates are often measured intermittently and with measurement errors. Motivated by the AIDS Clinical Trials Group (ACTG) 175 trial, this paper develops statistical inferences for the Cox model for partly interval censored failure times and longitudinal covariates with measurement errors. The conditional score methods developed for the Cox model with measurement errors and right censored data are no longer applicable to interval censored data. Assuming an additive measurement error model for a longitudinal covariate, we propose a nonparametric maximum likelihood estimation approach by deriving the measurement error induced hazard model that shows the attenuating effect of using the plug-in estimate for the true underlying longitudinal covariate. An EM algorithm is devised to facilitate maximum likelihood estimation that accounts for the partly interval censored failure times. The proposed methods can accommodate different numbers of replicates for different individuals and at different times. Simulation studies show that the proposed methods perform well with satisfactory finite-sample performances and that the naive methods ignoring measurement error or using the plug-in estimate can yield large biases. A hypothesis testing procedure for the measurement error model is proposed. The proposed methods are applied to the ACTG 175 trial to assess the associations of treatment arm and time-dependent CD4 cell count on the composite clinical endpoint of AIDS or death.

Supplementary information: The online version contains supplementary material available at 10.1007/s12561-023-09372-y.

Novel biomarkers, in combination with currently available clinical information, have been sought to enhance clinical decision making in many branches of medicine, including screening, surveillance and prognosis. An individualized clinical decision rule (ICDR) is a decision rule that matches subgroups of patients with tailored medical regimen based on patient characteristics. We proposed new approaches to identify ICDRs by directly optimizing a risk-adjusted clinical benefit function that acknowledges the tradeoff between detecting disease and over-treating patients with benign conditions. In particular, we developed a novel plug-in algorithm to optimize the risk-adjusted clinical benefit function, which leads to the construction of both nonparametric and linear parametric ICDRs. In addition, we proposed a novel approach based on the direct optimization of a smoothed ramp loss function to further enhance the robustness of a linear ICDR. We studied the asymptotic theories of the proposed estimators. Simulation results demonstrated good finite sample performance for the proposed estimators and improved clinical utilities when compared to standard approaches. The methods were applied to a prostate cancer biomarker study.

Assessing the impact of complex interventions on measurable health outcomes is a growing concern in health care and health policy. Interrupted time series (ITS) designs borrow from traditional case-crossover designs and function as quasi-experimental methodology able to retrospectively analyze the impact of an intervention. Statistical models used to analyze ITS designs primarily focus on continuous-valued outcomes. We propose the "Generalized Robust ITS" (GRITS) model appropriate for outcomes whose underlying distribution belongs to the exponential family of distributions, thereby expanding the available methodology to adequately model binary and count responses. GRITS formally implements a test for the existence of a change point in discrete ITS. The methodology proposed is able to test for the existence of and estimate the change point, borrow information across units in multi-unit settings, and test for differences in the mean function and correlation pre- and post-intervention. The methodology is illustrated by analyzing patient falls from a hospital that implemented and evaluated a new care delivery model in multiple units.