2023-2024

4/12/2024 

12:30-1:30pm;  1:30-2:45pm

Youtube Link: https://www.youtube.com/watch?v=jFVwtpxiaCE&ab_channel=JohnsHopkinsSLAMWorkingGroup

Title: Two-sample survival probability curves: A graphical approach for the analysis of time-to-event data in clinical trials

Speaker:  Xianghua Luo, Professor, Division of Biostatistics & Health Data Science, University of Minnesota

Abstract: With the aim to improve the communication of trial results, we introduced a novel graphical approach that complements the analysis of time-to-event outcomes in two-arm randomized trial setting. We first proposed a nonparametric estimator of the two-sample survival curve for right-censored data using Kaplan-Meier (K-M) survival estimates for each arm. Later, we extended the method to the analysis of arbitrarily censored and truncated data by using one-sample nonparametric maximum likelihood estimator (NPMLE), the analog of the K-M estimator. We are currently working on applying Bayesian models to the estimation of the curve and the area under the curve (AUC) measure (i.e., concordance). The estimated two-sample survival curve can be used to visualize treatment effect as well as potential effect modification of factors of interest and to graphically assess model fit of semiparametric and parametric models such as the Cox model. The proposed two-sample survival probability plot puts trials in a standardized [0,1] X [0,1] space, allowing for a simple visualization of the main effect, effect modification, and the adequacy of a model fit.

10/6/2023

Speaker: Maria Kamenetsky,, National Cancer Institute, NIH 

Title: Spatial Methods in Cluster Detection and Environmental Mixtures

[Abstract] Patterns in the grouping of diseases and exposures across space and time are important to epidemiologists and health professionals because they may indicate elevated disease risk or unmeasured environmental exposures. I will present on two spatial methods focusing on cluster detection and environmental mixtures. The first part of the talk will focus on spatial cluster detection. Here, we develop a novel stacking approach to detect spatial clusters via overlapping circular windows that in turn generate a set of single-cluster models. We use likelihood-based weights to stack single-cluster relative risk estimates into a meta-model, where the optimal number of parameters/space-time clusters is identified using information criteria. We illustrate our proposed method using female breast cancer incidence data at the municipality level in Japan. In the second half of the presentation, I will focus on a proposed method for environmental mixtures. Mixtures are pervasive in the environment and refer to compounds of multiple concurrent exposures such as air pollution or water contamination. These mixtures are often subject to censoring due to reporting or instrument limitations. We are developing a flexible latent spatial model to predict measurements of an environmental mixture at unsampled locations, accounting for censoring at the limit of reporting. Our proposed method is illustrated using well water samples in North Carolina. This is a work in progress. By identifying spatial patterns in both disease risk and environmental exposures, we can develop models to better identify spatio-temporal trends in disease burdens, predict measurements of censored complex compounds and, ultimately, improve public health.

9/22/2023 

Speaker: Fangya Mao, National Cancer Institute, NIH 

Title: Design and analysis of two-phase studies with lifetime data

[Abstract] Two-phase designs have emerged as invaluable tools for effectively allocating resources when faced with costly measurements of covariates, such as biomarkers, across an entire cohort. Typically, these designs encompass two sequential steps. Phase I involves collecting information on responses and affordable covariates from a large cohort, while phase II focuses on a subsample in which the marker of interest is assayed via biospecimen examination. In this presentation, I will share my recent research on the design and analysis of two-phase studies involving different types of lifetime data, collected from time-to-event and multistate processes with various observational schemes. Design efficiency is evaluated in terms of the precision when estimating the biomarker's effect on event processes of interest, such as the disease progression. Various two-phase sampling schemes are investigated and compared under distinct analysis frameworks such as maximum likelihood and inverse probability weighting

5/12/2023

Title:  Multiply robust estimation of principal causal effects with noncompliance and time-to-event outcomes

Presenter: Fan Li, Yale School of Public Health

Abstract: This presentation considers analyzing randomized trials and observational studies with a time-to-event outcome in the presence of treatment noncompliance. We develop a multiply robust estimator for the principal causal effects measured on the counterfactual survival probability scale within principal strata characterized by the joint potential compliance status. The multiply robust estimator requires specification of several working models to characterize the treatment assignment mechanism, the compliance status, time to censoring, and time to the primary outcome of interest. We demonstrate that the proposed estimator is consistent even if one, and sometimes two, of the working models are incorrectly specified. A sensitivity analysis strategy is also developed for investigating robustness of the proposed estimator for violations of the identification assumption. We applied the proposed approach to the ADAPTABLE study to evaluate effectiveness of low- versus high-dosage aspirin intake on patients’ death and hospitalization from cardiovascular diseases. It is observed that, comparing to low-aspirin dosage, high-aspirin dosage exerted differential effects among always high-dosage takers, compliers, and always low-dosage takers. This finding suggests that policy makers should design personalized strategies targeting patients with different compliance status to maximize the benefits to entire population.

4/28/2023

Title: Nonparametric Inference in the Accelerated Failure Time Model Using Restricted Means

Speaker: Dr. Ted Karrison, Department of Public Health Sciences, University of Chicago

Abstract: We propose a nonparametric estimate of the scale-change parameter for characterizing the difference between two survival functions under the accelerated failure time model using an estimating equation based on restricted means. Advantages of our restricted means based approach compared to current nonparametric procedures is the strictly monotone nature of the estimating equation as a function of the scale-change parameter, leading to a unique root, as well as the availability of a direct standard error estimate, avoiding the need for hazard function estimation or re-sampling to conduct inference. We derive the asymptotic properties of the proposed estimator for fixed and for random point of restriction. In a simulation study, we compare the performance of the proposed estimator with parametric and nonparametric competitors in terms of bias, efficiency, and accuracy of coverage probabilities. The restricted means based approach provides unbiased estimates and accurate confidence interval coverage rates with efficiency ranging from 81% to 95% relative to fitting the correct parametric model. An example from a randomized clinical trial in head and neck cancer is provided to illustrate an application of the methodology in practice.

4/21/2023

Title: Disease assessment schedule, a unique challenge in using real world or historical progression-free survival data in clinical trials

Speaker: Dr. Jian Zhu, Servier Pharmaceuticals

Abstract:  The past decade has witnessed an increasing trend in utilizing external control data in clinical trials, especially in the form of synthetic control arms (SCA) derived from real-world or historical trial data. Including such data in clinical trial analysis can improve trial feasibility and efficiency, provided the issues caused by non-randomization and systematic differences are appropriately addressed. Motivated by the comparative analysis of SCA progression-free survival (PFS) and trial arm PFS, we focused on a unique issue for external time-to-event data that depend on disease assessment schedules (DAS). Specifically, since DAS are generally inconsistent across different data sources, it may lead to severe under-estimation or over-estimation of the treatment effect size (and subsequently power loss or type I error inflation). To address this issue, we propose a proper statistical inference framework that harmonizes the DAS through data augmentation by multiple imputation. We demonstrate that the proposed method is able to correct the aforementioned bias and brings us one step closer to valid inference.

3/31/2023

Title: Restricted Mean Survival Time Estimation Using Bayesian Nonparametric Dependent Mixture Models

Speaker: Dr. Ruizhe Chen, Division of Oncology Biostatistics, Johns Hopkins University School of Medicine

    Restricted mean survival time (RMST) is an intuitive summary statistic for time-to-event random variables and can be used for measuring treatment effects. Compared to hazard ratio, its estimation procedure is robust against the non-proportional hazards assumption. We propose nonparametric Bayeisan (BNP) estimators for RMST using a dependent stick-breaking process prior mixture model that adjusts for mixed-type covariates. The proposed Bayesian estimators can yield both group-level causal estimate and subject-level predictions. Besides, we propose a novel dependent stick-breaking process prior that on average results in narrower credible intervals while maintaining similar coverage probability compared to a dependent probit stick-breaking process prior. We conduct simulation studies to investigate the performance of the proposed BNP RMST estimators compared to existing frequentist approaches and under different Bayesian modeling choices. The proposed framework is applied to estimate the treatment effect of an immuno therapy among KRAS wild-type colorectal cancer patients.

3/1/2023

Title: Joint semiparametric models for case-cohort designs 

Speaker: Guoqing Diao, Professor, Department of Biostatistics and Bioinformatics, George Washington University.

Abstract: Two-phase studies such as case-cohort and nested case-control studies are widely used cost-effective sampling strategies. In the first phase, the observed failure/censoring time and inexpensive exposures are collected. In the second phase, a subgroup of subjects is selected for measurements of expensive exposures based on the information from the first phase. One challenging issue is how to utilize all the available information to conduct efficient regression analyses of the two-phase study data. This paper proposes a joint semiparametric modeling of the survival outcome and the expensive exposures. Specifically, we assume a class of semiparametric transformation models and a semiparametric density ratio model for the survival outcome and the expensive exposures, respectively. The class of semiparametric transformation models includes the proportional hazards model and the proportional odds model as special cases. The density ratio model is flexible in modeling multivariate mixed-type data. We develop efficient likelihood-based estimation and inference procedures and establish the large sample properties of the nonparametric maximum likelihood estimators. Extensive numerical studies reveal that the proposed methods perform well under practical settings. The proposed methods also appear to be reasonably robust under various model mis-specifications. An application to the National Wilms Tumor Study is provided.