Lifetime Data Science (LiDS) Section Student Paper Awards

Joint models are widely used to link longitudinal biomarker trajectories with time-to-event outcomes. A key question is how well the longitudinal component discriminates between subjects who experience events earlier versus later. The concordance index (C-index) is the standard measure for this purpose, but its classical definition assumes a time-invariant risk score—an assumption violated when risk evolves with the biomarker trajectory.
We propose a generalized C-index grounded in stochastic-order preference that directly accommodates time-dependent risk scores from joint models with random effects. Our measure quantifies the probability that a subject with a higher predicted risk at any given time experiences the event before a subject with lower risk, fully accounting for how risk rankings may change over time. The generalization is interpretable, handles non-monotonic risk dynamics, and reduces to the classical C-index when risk is time-invariant.
The empirical performance of the proposed metric was assessed via simulation studies. Real-world data analysis was conducted on the Mayo Clinic's Primary Biliary Cirrhosis (PBC) clinical trial data to further demonstrate the robustness of the proposed index. 

We develop a joint model of multivariate longitudinal biomarkers with a change point at an interval-censored event time. Our model allows us to simultaneously understand the causal effect of longitudinal biomarkers on the event time and the causal effect of event time on the changes of longitudinal biomarkers post the event. A simulation study is carried out to demonstrate the satisfactory finite-sample performance of the proposed method for making inferences. Finally, the method is applied to PREDICT-HD data from a multisite observational cohort study of prodromal Huntington's disease individuals to ascertain the effects of cognitive impairments on the onset of Huntington's disease that subjects to interval censoring and how the disease onset accelerates the cognitive impairments.  

Biological research and clinical evidence suggest that treatment response may vary substantially across individuals with different characteristics, such as comorbidities, genetic variants, environmental, or socio-economic factors. Future precision medicine requires accurate assessment of explainable variability in treatment effects, known as heterogeneous treatment effects (HTE), to guide optimal clinical decisions at the individual level. Measuring HTE by the ratio of survival probabilities under the structural failure time model, we develop a martingale R-learner to estimate HTE. Our martingale R-learner incorporates flexible estimators for (1) marginal survival or cumulative hazards for the association between outcome and confounders, and (2) time-varying propensity scores in risk sets, which enables leveraging advances in machine learning. To reduce the impact of estimation bias in these two nuisance models on HTE, we propose a Neyman-orthogonal score based on an orthogonal decomposition of conditional model martingale residuals into residuals of the propensity score and the marginal model martingale. The resulting martingale R-learner attains an oracle property, i.e., estimation errors of the nuisance models have no impact on HTE if their estimators are consistent at rate o(n^(-1/4)). Numerical experiments across various settings demonstrate empirical performance consistent with the theory. We investigate the effect of alcohol on dementia through the Honolulu-Asia Aging Study data. 

Left-truncated time-to-event data are frequently encountered in health sciences research, e.g., epidemiologic studies of prevalent cohorts. In certain clinical settings, patients are at risk for death as well as a serious but nonfatal event, and a composite endpoint, defined as the time from disease onset until the earlier of the nonfatal event and death, might be desirable. Component-wise censoring of a composite endpoint arises when each component is subject to a different censoring mechanism. For example, the nonfatal event may be interval censored between visits for assessments, and death is right censored. Methods to estimate the event-free survival (EFS) with left-truncated and right-censored data are available in the literature, but they cannot handle component-wise censoring. We propose a kernel smoothing method for left-truncated and component-wise-censored data to provide a nonparametric estimator for EFS, which takes component-wise censoring into account by treating the nonfatal event as a time-dependent binary variable that is observed intermittently during follow-up. Our method can also estimate and test for differences in the restricted mean EFS time and can leverage two types of supplemental data that may be available in the same study: death-only data (if some participants are only followed for death but not for the nonfatal event) and incident cohort data (if a study also enrolls participants who have not yet experienced the index event at the enrollment time). We assess the proposed methods by simulations of different study designs and demonstrate the method using the Atherosclerosis Risk in Communities Study to estimate post-myocardial infarction, dementia-free survival probability. 

Understanding how vaccines perform against different pathogen genotypes is crucial for developing effective prevention strategies, particularly for highly genetically diverse pathogens like HIV. Sieve analysis is a statistical framework used to determine whether a vaccine selectively prevents acquisition of certain genotypes while allowing breakthrough of other genotypes that evade immune responses. Traditionally, these analyses are conducted with a single sequence available per individual acquiring the pathogen. However, modern sequencing technology can provide detailed characterization of intra-individual viral diversity by capturing up to hundreds of pathogen sequences per person. In this work, we introduce methodology that extends sieve analysis to account for intra-individual viral diversity. Our approach estimates vaccine efficacy against viral populations with varying true (unobservable) frequencies of vaccine-mismatched mutations. To account for differential resolution of information from differing sequence counts per person, we use competing risks Cox regression with modeled causes of failure and propose an empirical Bayes approach for the classification model. Simulation studies demonstrate that our approach reduces bias, provides nominal confidence interval coverage, and improves statistical power compared to conventional methods. We apply our method to the HVTN 705 Imbokodo trial, which assessed the efficacy of a heterologous vaccine regimen in preventing HIV-1 acquisition.