Article
Mathematical & Computational Biology
Jih-Chang Yu, Yen-Tsung Huang
Summary: The natural history of hepatitis B or C involves multiple milestones such as liver cirrhosis and liver cancer. Semicompeting risks pose a common problem in fully characterizing the natural course, where liver cirrhosis and liver cancer are both of interest, but only the former may be censored by the latter. Copula, frailty, and multistate models are established analytics for semicompeting risks. In this study, we introduce a mediation framework for semicompeting risks, treating liver cirrhosis as a mediator and liver cancer as an outcome. We define indirect and direct effects as the effects of an exposure on liver cancer incidence mediated and not mediated through liver cirrhosis, respectively. Estimands and estimators are derived under copula, frailty, and multistate models, and their asymptotic results are established. The proposed framework is demonstrated in a hepatitis study, showing that hepatitis B and C increase liver cancer incidence by increasing liver cirrhosis incidence. Thus, mediation modeling provides a unified framework for various semicompeting risks models.
STATISTICS IN MEDICINE
(2023)
Article
Mathematics
Xifen Huang, Jinfeng Xu
Summary: In this study, we propose a method for estimating the transition functions in a semi-competing risks model under the illness-death model framework. The method does not rely on parametric assumptions and provides smooth estimates that are easy to compute and interpret. Compared to existing approaches, our method does not require the subjective specification of the frailty distribution or the Markov or semi-Markov assumption, making it more applicable in real-world settings.
Article
Mathematical & Computational Biology
Eric S. Kawaguchi, Gang Li, Juan Pablo Lewinger, W. James Gauderman
Summary: Individuals with different genetic profiles may have different clinical outcomes when exposed to the environment. We propose a powerful two-step hypothesis testing framework to identify gene-environment interactions in a genome-wide interaction scan.
STATISTICS IN MEDICINE
(2022)
Article
Mathematics, Applied
Jin-Jian Hsieh, Yun-Jhu Chen
Summary: This article discusses truncation data and the limitations of current analysis methods. It proposes the use of copulas to estimate the survival function of the interested event time and introduces two estimation procedures for the proportional hazard model and proportional odds model. The performance of these approaches is evaluated through simulation studies, and they are also applied to analyze real datasets.
Article
Statistics & Probability
Simon M. S. Lo, Ralf A. Wilke
Summary: A model without restrictions is introduced in this article for survival analysis situations where there is only interest in one risk and no information about risk dependence is available. An application to employment duration demonstrates that this model avoids sizable bias in the estimated gender effect.
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES C-APPLIED STATISTICS
(2023)
Article
Statistics & Probability
J. H. McVittie, V. Addona
Summary: Sporting careers can be divided into two distinct subsamples based on whether they began before or during a preset time interval. By using a proportional hazards model and a partial likelihood estimator, this study examines the effect of covariates on survival, using NBA data to validate the combined cohort proportional hazards methodology.
JOURNAL OF APPLIED STATISTICS
(2022)
Article
Biology
Lea Kats, Malka Gorfine
Summary: This study proposes a new model and estimation procedure for analyzing illness-death survival data, adopting accelerated failure time (AFT) models. A shared frailty variate is introduced to capture the unobserved dependency between nonterminal and terminal failure times given observed covariates. The proposed modeling approach aims to leverage the interpretability advantage of AFT models while benefiting from the intuitive interpretation of hazard functions. A semiparametric maximum likelihood estimation procedure is developed, and the performance is evaluated using breast cancer data and a graphical goodness-of-fit procedure.
Article
Health Care Sciences & Services
Deo Kumar Srivastava, E. Olusegun George, Zhaohua Lu, Shesh N. Rai
Summary: Clinical trials with survival endpoints are designed to enroll patients for a specified number of years with a follow-up period. Patients may be censored due to various reasons, potentially leading to unequal censoring and inaccurate conclusions about treatment effects.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2021)
Article
Mathematical & Computational Biology
Anne A. Eaton, Emily C. Zabor
Summary: This article compares the performance of Cox model estimators for right-censored data and interval-censored data in the context of component-wise censored data where the visit process differs across levels of a covariate of interest. The study found that the bias of the Cox model estimators for right-censored data and cause-specific hazard estimators increases as the frequency of visits decreases, especially when visit schedules differ according to levels of a covariate of interest. The Cox model estimator for interval-censored data with censoring at the last disease-free date is recommended in the presence of differential visit schedules.
STATISTICS IN MEDICINE
(2022)
Article
Computer Science, Interdisciplinary Applications
Huaxiang Zhou, Yizhu Li
Summary: This paper addresses a new maintenance problem involving two competing dependent risks - minor failures and major failures. The cumulative number of minor failures is incorporated as a covariant process in the proportional hazards model for major failures. The Time Discrete Markovian Approximation (TDMA) technique is introduced to overcome the curse of dimensionality in Markov Decision Processes (MDP) and simplify high-dimensional integration when computing average costs through renewal theory. A new optimal control limit policy is developed, using a mixed hazards function as the threshold, and its agreement with the MDP solution is revealed. Additionally, an iterative algorithm is developed to accelerate convergence to the optimal solution compared to policy iteration.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Statistics & Probability
Bo Zhao, Shuying Wang, Chunjie Wang
Summary: This article proposes a new frailty-based generalized estimating equation (GEE) method for proportional hazards (PH) model with informative interval-censored failure time data. The proposed method can provide an unbiased estimator of the cumulative hazards function. The performance of the method is assessed through extensive simulation study and an application to an AIDS clinical trial dataset.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2023)
Article
Biology
Peijie Wang, Danning Li, Jianguo Sun
Summary: The paper discusses regression analysis of left-truncated failure time data, proposing a pairwise pseudo-likelihood approach and demonstrating its efficiency and feasibility through a simulation study.
Article
Statistics & Probability
Alejandro R. Vasquez, Gabriel Escarela
Summary: This study developed copula-based constructions for the joint distribution of overall survival time and cause-specific failure for competing risks data analysis. Covariate effects were integrated using parametric and semiparametric proportional hazards models for survival times, and a logistic regression model for cause of failure. The methods appeal due to their adequate modeling of the proportional hazards component, ability to find parsimonious models, and the simplicity and usefulness of parameter interpretation.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2021)
Article
Mathematical & Computational Biology
Song Yang, James Troendle, Daewoo Pak, Eric Leifer
Summary: This study investigated the asymptotic distributions of event-specific win ratios in semi-competing risks data and found that the win ratios converge to hazard ratios under proportional hazards assumptions. Confidence intervals and testing procedures were developed based on bivariate normal distributions of win ratios. Proper transformations of win ratios were identified to control type one error rate and maintain competitive power in various testing procedures, with good coverage probabilities in confidence intervals. Moreover, tests for proportional hazards assumptions and equal hazard ratios were developed and illustrated in a clinical trial evaluating the effects of spironolactone in heart failure patients.
STATISTICS IN MEDICINE
(2022)
Article
Mathematical & Computational Biology
Bin Luo, Xiaoli Gao, Susan Halabi
Summary: This article introduces a method for outlier detection and robust regression in the sparse proportional hazards model, which is able to perform variable selection and outlier detection simultaneously. The method is applied to characterizing exceptional responders in a clinical trial for prostate cancer.
STATISTICS IN MEDICINE
(2022)
Article
Pharmacology & Pharmacy
Min Yuan, Yi Li, Yaning Yang, Jinfeng Xu, Fangbiao Tao, Liang Zhao, Honghui Zhou, Jose Pinheiro, Xu Steven Xu
PHARMACEUTICAL STATISTICS
(2020)
Article
Economics
Jinfeng Xu, Mu Yue, Wenyang Zhang
ECONOMETRIC THEORY
(2020)
Article
Statistics & Probability
Jinfeng Xu, Wai Keung Li, Zhiliang Ying
SCANDINAVIAN JOURNAL OF STATISTICS
(2020)
Article
Statistics & Probability
Jianhua Shi, Yutao Liu, Jinfeng Xu
Summary: This paper presents a nonparametric analysis method for length-biased and right-censored data using quantile difference estimation. By proposing a smoothed quantile difference approach, the estimating efficiency is improved and its validity is supported by asymptotic theories.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2022)
Article
Statistics & Probability
Yi Liu, Jinfeng Xu, Gang Li
Summary: A joint feature screening method has been developed for censored survival data, which retains relevant features and is more desirable than marginal screening. The proposed method is proven to be more robust than its competitors and is implemented using Matlab, with its finite sample performance illustrated through simulation studies and real data examples.
CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE
(2021)
Article
Mathematical & Computational Biology
WenWu Wang, Jinfeng Xu, Joel Schwartz, Andrea Baccarelli, Zhonghua Liu
Summary: This article discusses the identification of subgroup-specific mediation effects in biomedical studies and proposes a simple mixture modeling approach to account for latent subgroup structures. The method involves inferring individual subgroup membership based on posterior probability and selecting the number of latent subgroups using the singular Bayesian information criterion.
STATISTICS IN MEDICINE
(2021)
Article
Mathematical & Computational Biology
Yan Zhou, Li Zhang, Jinfeng Xu, Jun Zhang, Xiaodong Yan
Summary: Bulk and single-cell RNA-seq data are being used in biology and medicine research as alternatives to traditional technology. A new category encoding (CAEN) method has been proposed to select feature genes for classification, improving performance and efficiency in distinguishing important genes. Simulation studies demonstrate the superiority of the CAEN method over existing techniques in most settings.
STATISTICS IN MEDICINE
(2021)
Article
Mathematics
Xifen Huang, Chaosong Xiong, Jinfeng Xu, Jianhua Shi, Jinhong Huang
Summary: This paper introduces a flexible semiparametric mixture modeling strategy for subgroup analysis and regression analysis in survival data, utilizing nonparametric maximum likelihood method and a pair of MM algorithms with monotone ascent property for estimation procedures. The proposed methodology is demonstrated through simulation studies and empirical analysis.
Article
Mathematics
Xifen Huang, Jinfeng Xu, Yunpeng Zhou
Summary: This study proposes an advanced modeling approach for analyzing complex multivariate survival data, which incorporates a flexible frailty distribution and utilizes innovative regularization techniques for accurate estimation and effective model selection. The proposed methodology has been validated through real-world data and comprehensive simulation studies, demonstrating its practical utility and desirable theoretical properties.
Article
Mathematical & Computational Biology
Xiaoyu Zhang, Yunpeng Zhou, Jinfeng Xu, Kam Chuen Yuen
Summary: In biomedical studies, it is important to link gene expression profiles to censored survival phenotypes. Regularized methods that combine rank-based loss function and penalty are often used to identify prognostic biomarkers and prediction models for event times. However, the l(1) penalty approximation used in existing methods leads to inflated model size in practice.
STATISTICS AND ITS INTERFACE
(2022)
Article
Biology
Chun Yin Lee, Kin Yau Wong, K. F. Lam, Jinfeng Xu
Summary: This paper proposes a flexible class of semiparametric partly linear frailty transformation models for analyzing clustered interval-censored data. The estimators of the model parameters are shown to be strongly consistent and asymptotically normal, and an application to dental study is provided for illustration.
Review
Nutrition & Dietetics
Tommy H. T. Wong, Chi Ho Wong, Xiaoyu Zhang, Yunpeng Zhou, Jinfeng Xu, Kam Chuen Yuen, Jennifer M. F. Wan, Jimmy C. Y. Louie
Summary: This study re-evaluated the association between coffee consumption and metabolic syndrome, finding that there may not be a significant link between the two, especially at medium consumption levels. More longitudinal studies are needed to further explore the temporal relationship between coffee consumption and metabolic syndrome.
ADVANCES IN NUTRITION
(2021)
Article
Biochemical Research Methods
Min Yuan, Xu Steven Xu, Yaning Yang, Yinsheng Zhou, Yi Li, Jinfeng Xu, Jose Pinheiro
Summary: This study found bias in existing EBE-based GWAS methods and proposed a fast and unbiased method, SCEBE, for large-scale GWAS analysis, significantly improving computational efficiency.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Statistics & Probability
Yixin Fang, Jinfeng Xu
Article
Engineering, Electrical & Electronic
Kaiyi Ji, Jian Tan, Jinfeng Xu, Yuejie Chi
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2020)