Dayu Sun

2025

1. JASA

   Kernel Meets Sieve: Transformed Hazards Models with Sparse Longitudinal Covariates

   Dayu Sun, Zhuowei Sun, Xingqiu Zhao, and Hongyuan Cao

   Journal of the American Statistical Association, 2025

   Abs DOI Bib Code

   Abstract We study the transformed hazards model with time-dependent covariates observed intermittently for the censored outcome. Existing work assumes the availability of the whole trajectory of the time-dependent covariates, which is unrealistic. We propose combining kernel-weighted log-likelihood and sieve maximum log-likelihood estimation to conduct statistical inference. The method is robust and easy to implement. We establish the asymptotic properties of the proposed estimator and contribute to a rigorous theoretical framework for general kernel-weighted sieve M-estimators. Numerical studies corroborate our theoretical results and show that the proposed method performs favorably over competing methods. The analysis of a dataset from a COVID-19 study in Wuhan identifies clinical predictors that otherwise cannot be obtained using existing methods. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

   @article{Sun2025,
     author = {Sun, Dayu and Sun, Zhuowei and Zhao, Xingqiu and Cao, Hongyuan},
     journal = {Journal of the American Statistical Association},
     title = {Kernel Meets Sieve: Transformed Hazards Models with Sparse Longitudinal Covariates},
     year = {2025},
     volume = {120},
     number = {552},
     pages = {2580--2591},
     author+an = {1=highlight},
     doi = {10.1080/01621459.2025.2476781},
     publisher = {Informa UK Limited},
   }

2. Nat. Commun.

   The HM-TARGET personalised real-time haemodynamic targets in critical care

   Yanhua Sun, Jiangqiong Li, Xiang Liu, Genevieve A. Mortensen, Xiaoping Gu, David C. Adams, Haixu Tang, Jing Su, Ziyue Liu, Dayu Sun^†, and Lingzhong Meng^†

   Nature Communications, 2025

   Abs DOI Bib

   Haemodynamic management in critical care typically relies on static, population-based targets that overlook patient-specific physiology and the evolving nature of illness. We develop and validate a framework for real-time, personalised haemodynamic management using a time-dependent Cox model that integrates static and dynamic clinical data to predict survival probabilities and derive optimal heart rate and systolic blood pressure targets over time. Trained on the electronic Intensive Care Unit dataset and externally validated with Medical Information Mart for Intensive Care IV and Indiana University Health cohorts, the model demonstrates high predictive accuracy (c-index up to 0.931) and generalisability across diverse populations. Patients with heart rate and systolic blood pressure values closer to model-predicted targets exhibit significantly lower intensive care unit mortality than those aligned with fixed, population-based thresholds. Exploratory dose-response and propensity score-matched analyses confirm outcome relevance, while case studies illustrate feasibility in critical care settings. This personalised, dynamic approach-termed Haemodynamic Management by Time-Adaptive, Risk-Guided Estimation of Targets (HM-TARGET)-offers a scalable framework for precision haemodynamic management in critically ill patients. Prospective trials are warranted to evaluate clinical impact.

   @article{Sun2025a,
     author = {Sun, Yanhua and Li, Jiangqiong and Liu, Xiang and Mortensen, Genevieve A. and Gu, Xiaoping and Adams, David C. and Tang, Haixu and Su, Jing and Liu, Ziyue and Sun, Dayu and Meng, Lingzhong},
     journal = {Nature Communications},
     title = {The {HM-TARGET} personalised real-time haemodynamic targets in critical care},
     year = {2025},
     issn = {2041-1723},
     number = {1},
     pages = {7307},
     volume = {16},
     author+an = {10=highlight,corresponding;11=corresponding},
     doi = {10.1038/s41467-025-62527-x},
     keywords = {colab},
     publisher = {Springer Science and Business Media LLC},
   }

2024

1. A robust approach for regression analysis of panel count data with time-varying covariates

   Dayu Sun, Yuanyuan Guo, Yang Li, Wanzhu Tu, and Jianguo Sun

   Bernoulli, 2024

   Abs DOI Bib

   The validity of statistical inference for panel count data with time-varying covariates depends on the correct specification of within-subject correlation structures; misspecification often leads to questionable inference. To alleviate, robust inference has been proposed for mean models, which implicitly assume monotone mean functions. When covariate values fluctuate with time, however, the assumed monotonicity becomes unrealistic. In this research, we propose a robust inference based on rate models that are free of such constraints. Since the asymptotic variance has no closed form under the rate model, we further develop computationally efficient robust variance estimators using the Expectation-Maximization (EM) algorithm, thus sidestepping the need for computationally intensive numerical methods, which could undermine the robustness. Rigorous theoretical development is provided in support of parameter estimation and inference. Extensive simulation studies demonstrate the superiority of the proposed method. We present a real clinical application to illustrate the use of the proposed method.

   @article{Sun2024,
     author = {Sun, Dayu and Guo, Yuanyuan and Li, Yang and Tu, Wanzhu and Sun, Jianguo},
     journal = {Bernoulli},
     title = {A robust approach for regression analysis of panel count data with time-varying covariates},
     year = {2024},
     number = {4},
     pages = {3251-3275},
     volume = {30},
     author+an = {1=highlight},
     doi = {10.3150/23-bej1713},
     publisher = {Bernoulli Society for Mathematical Statistics and Probability},
   }

2. JASA

   Partial quantile tensor regression

   Dayu Sun, Zhiping Qiu, Limin Peng, Ying Guo, and Amita Manatunga

   Journal of the American Statistical Association, 2024

   Abs DOI Bib Code

   Tensors, characterized as multidimensional arrays, are frequently encountered in modern scientific studies. Quantile regression has the unique capacity to explore how a tensor covariate influences different segments of the response distribution. In this work, we propose a partial quantile tensor regression (PQTR) framework, which novelly applies the core principle of the partial least squares technique to achieve effective dimension reduction for quantile regression with a tensor covariate. The proposed PQTR algorithm is computationally efficient and scalable to a large tensor covariate. Moreover, we uncover an appealing latent variable model representation for the PQTR algorithm, justifying a simple population interpretation of the resulting estimator. We further investigate the connection of the PQTR procedure with an envelope quantile tensor regression (EQTR) model, which defines a general set of sparsity conditions tailored to quantile tensor regression. We prove the root- n consistency of the PQTR estimator under the EQTR model, and demonstrate its superior finite-sample performance compared to benchmark methods through simulation studies. We demonstrate the practical utility of the proposed method via an application to a neuroimaging study of post traumatic stress disorder (PTSD). Results derived from the proposed method are more neurobiologically meaningful and interpretable as compared to those from existing methods.

   @article{Sun2024a,
     author = {Sun, Dayu and Qiu, Zhiping and Peng, Limin and Guo, Ying and Manatunga, Amita},
     journal = {Journal of the American Statistical Association},
     title = {Partial quantile tensor regression},
     year = {2024},
     issn = {1537-274X},
     number = {551},
     pages = {1724--1735},
     volume = {120},
     author+an = {1=highlight},
     doi = {10.1080/01621459.2024.2422129},
     publisher = {Informa UK Limited},
   }