Publications | Dayu Sun

Please see CV for more detailed information on publications and working papers.

2026

Divergent Effects of Communication Skills on Aggression Among Justice-Involved Youth with and without ADHD: A Propensity Score Matching Analysis

Lin Liu, Dayu Sun, Jorge Luis Hernández, and Michael T. Baglivio

Criminal Justice and Behavior, 2026

Abs DOI Bib

Higher rates of aggressive behavior have been reported among youth with attention-deficit/hyperactivity disorder (ADHD) compared to their peers without ADHD. However, it remains an under-explored question whether protective factors against aggression operate differently for youth with and without ADHD. Drawing on 10 years of statewide longitudinal data, this study examined communication skills as a protective factor against aggression for justice-involved youth with and without ADHD. Propensity score matching was used to minimize the difference in 20 observed covariates to increase the rigor of the estimates. We found significant protective effects of communication skills on aggression among justice-involved youth in juvenile justice residential facilities. Furthermore, youth with ADHD received amplified protective effects of communication skills. Our findings provided important implications for prevention and intervention.
@article{Liu2025a, author = {Liu, Lin and Sun, Dayu and Hern\'andez, Jorge Luis and Baglivio, Michael T.}, journal = {Criminal Justice and Behavior}, title = {Divergent Effects of Communication Skills on Aggression Among Justice-Involved Youth with and without ADHD: A Propensity Score Matching Analysis}, year = {2026}, pages = {00938548261430847}, author+an = {2=highlight}, doi = {10.1177/00938548261430847}, keywords = {colab}, }
The Last Straw: Sudden Shifts in Substance Use During the First 6 Months After Prison Release

Lin Liu, Christy A. Visher, and Dayu Sun

Crime & Delinquency, 2026

Abs DOI Bib

Substance use during reentry is a complex and dynamic phenomenon. Although using substance and returning to prison are what post-incarcerated individuals are determined to avoid, stressors such as financial hardship, family estrangement, and other challenges can put them at an elevated risk of substance use. This study examined whether substance use during reentry is best characterized as a cusp catastrophe process in which substance use, rather than steady escalation, exhibits a shift from stable increase to a sudden, catastrophic increase when certain situation occurs. Longitudinal data of post-incarcerated adult males were leveraged to answer the research question. Results indicated that cusp catastrophe model accounted for ten times more variance in substance use than a linear model. Financial difficulties and mental health emerged as significant bifurcation factors, propelling individuals “over the edge” into a sudden escalation of use. In contrast, family bonds, neighborhood environment, and social isolation functioned as stable background factors that exerted linear effects on substance use. Findings suggest that substance use during reentry is marked by abrupt escalation rather than steady, moderate increases. Prevention efforts should therefore prioritize mitigating the acute stressors that trigger the sudden escalation of substance use among post-incarcerated individuals.
@article{Liu2026, author = {Liu, Lin and Visher, Christy A. and Sun, Dayu}, journal = {Crime \& Delinquency}, title = {The Last Straw: Sudden Shifts in Substance Use During the First 6 Months After Prison Release}, year = {2026}, pages = {00111287261434485}, author+an = {3=highlight}, doi = {10.1177/00111287261434485}, keywords = {colab}, }
Learning Health Systems and Substance Use Care Cascade Achievement Among Justice-Involved Youth: A Cluster-Randomized Stepped-Wedge Clinical Trial

Matthew C. Aalsma, Katherine Schwartz, Dayu Sun, Lauren M. O’Reilly, Steven A. Brown, Patrick O. Monahan, Lisa Saldana, Sarah E. Wiehe, Tamika C. B. Zapolski, Leslie A. Hulvershorn, Zachary W. Adams, and Allyson L. Dir

JAMA Network Open, 2026

Abs DOI Bib

Importance: Adolescents involved in the youth legal system (YLS) rarely use community-based behavioral health services, despite their disproportionate risk for substance use and substance use disorders (SU/D). A care cascade framework quantifies deficits in the process by which 2 systems—YLS and behavioral health care—ensure that youths engage in indicated treatment. Objective: To test effectiveness of a cross-system learning health systems (LHS) intervention on S/UD care cascade outcomes among YLS-involved youth. Design, Setting, and Participants: This cluster-randomized stepped-wedge clinical trial was designed to improve use of SU/D treatment by youths. Eight counties in a single Midwest state were randomly assigned to 1 of 3 cohorts stepped in from preintervention control to intervention in 9-month intervals. Each county-level LHS team included juvenile probation department and community mental health center representatives. Administrative YLS records were collected from all youths aged 11 to 17 years arrested in participating counties from January 1, 2019, through March 31, 2025, and linked to Medicaid claims data. Interventions: LHS interventions were based on principles of continuous quality improvement and data-driven decision-making. Here, LHS principles were innovatively applied among collaborative teams to identify and resolve local gaps in SU/D care cascade achievement. Main Outcomes and Measures: The main outcome was timeliness of care cascade step achievement after arrest among YLS-involved adolescents. Cascade steps included screening for SU/D risk, identification with need for services, referral for services, initiation of services, and engagement in services. Results: Cascade step achievement was recorded for 5731 youths with linked YLS and Medicaid records; participants included 3538 males (62%) with a mean (SD) age of 15.4 (1.7) years at the time of arrest. A total of 1010 youths (18%) were Black, 614 (11%) were Hispanic, and 4362 (76%) were White. The LHS was associated with a significant reduction in the number of days between youth first arrest and risk screening (hazard ratio, 1.67; 95% CI, 1.12-2.23); significant interactions showed reductions in time from arrest to service initiation and service engagement for youths arrested later in the study (ie, 3.5-4.0 years after study start through end of study). Conclusions and Relevance: This cluster-randomized stepped-wedge clinical trial demonstrated that the LHS improved the timeliness of SU/D care cascade achievement among YLS-involved youth. Trial Registration: ClinicalTrials.gov Identifier: NCT04499079
@article{Aalsma2026, author = {Aalsma, Matthew C. and Schwartz, Katherine and Sun, Dayu and O'Reilly, Lauren M. and Brown, Steven A. and Monahan, Patrick O. and Saldana, Lisa and Wiehe, Sarah E. and Zapolski, Tamika C. B. and Hulvershorn, Leslie A. and Adams, Zachary W. and Dir, Allyson L.}, journal = {JAMA Network Open}, title = {Learning Health Systems and Substance Use Care Cascade Achievement Among Justice-Involved Youth: A Cluster-Randomized Stepped-Wedge Clinical Trial}, year = {2026}, issn = {2574-3805}, number = {2}, pages = {e2558222}, volume = {9}, author+an = {3=highlight}, doi = {10.1001/jamanetworkopen.2025.58222}, keywords = {colab,service}, publisher = {American Medical Association (AMA)}, }

2025

modelSSE: An R Package for Characterizing Infectious Disease Superspreading from Contact Tracing Data

Shi Zhao, Zihao Guo, Kai Wang, Shengzhi Sun, Dayu Sun, Weiming Wang, Daihai He, Marc KC Chong, Yuantao Hao, and Eng-Kiong Yeoh

Bulletin of Mathematical Biology, 2025

Abs DOI Bib

Infectious disease superspreading is a phenomenon where few primary cases generate unexpectedly large numbers of secondary cases. Superspreading, is frequently documented in epidemiology literature, and is considered a consequence of heterogeneity in transmission. Since understanding the risks of superspreading became a rising concern from both statistical modelling and public health aspects, the R package modelSSE provides comprehensive analytical tools to characterize transmission heterogeneity. The package modelSSE integrates recent advances in statistical methods, such as decomposition of reproduction number, for modelling infectious disease superspreading using various types and sources of contact tracing data that allow models to be grounded in real-world observations. This study provided an overview of the theoretical background and implementation of modelSSE, designed to facilitate learning infectious disease transmission, and explore novel research questions for transmission risks and superspreading potentials. Detailed examples of classic, historical infectious disease datasets are given for demonstration and model extensions.
@article{Zhao2025, author = {Zhao, Shi and Guo, Zihao and Wang, Kai and Sun, Shengzhi and Sun, Dayu and Wang, Weiming and He, Daihai and Chong, Marc KC and Hao, Yuantao and Yeoh, Eng-Kiong}, journal = {Bulletin of Mathematical Biology}, title = {modelSSE: An R Package for Characterizing Infectious Disease Superspreading from Contact Tracing Data}, year = {2025}, number = {4}, volume = {87}, author+an = {5=highlight}, doi = {10.1007/s11538-025-01421-5}, publisher = {Springer Science and Business Media LLC}, }
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 arXiv 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}, number = {552}, pages = {2580--2591}, volume = {120}, author+an = {1=highlight}, doi = {10.1080/01621459.2025.2476781}, publisher = {Informa UK Limited}, }
Robust Principal Component Analysis with Truncated Weighted Nuclear Norm and Adaptive Histogram Equalization: A Novel Method for Low-Quality Retinal Image Enhancement

Habte Tadesse Likassa, Ding-Geng Chen, Kewei Chen, Yalin Wang, Wenhui Zhu, Oana Dumitrascu, and Dayu Sun

Statistics and Data Science in Imaging, 2025

Abs DOI Bib

Robust Principal Component Analysis (RPCA) methods are widely used in data-driven applications, including biomedical imaging, where retinal images often suffer from noise and artifacts that reduce diagnostic accuracy. We previously published a novel RPCA-based method that improved image quality beyond existing techniques. However, it lacked robustness against substantial noise and struggled to preserve fine image details critical for early screening and medical diagnosis. To address these limitations, we now propose a novel RPCA method that combines Truncated Weighted Nuclear Norm (TWNN) minimization with adaptive histogram-based enhancement techniques. Specifically, Adaptive Histogram Equalization (AHE) enhances the visibility of clinically relevant features, while TWNN minimization suppresses noise and anomalies, thereby substantially improving the reconstruction quality of degraded retinal images. This problem is formulated as an optimization task, where an efficient solution is obtained using the Alternating Direction Method of Multipliers (ADMM) to iteratively update the model parameters. The proposed method demonstrates superior performance in enhancing retinal image quality compared to recent baseline methods, as reflected in significantly improved PSNR, SSIM, and RAE scores. Extensive evaluations on synthetic, EyeQ, DRIVE, and STARE datasets further confirm its effectiveness, showing notable gains in Kappa, CR, CNR, and RMSE metrics, underscoring its strong potential for clinical application. In addition, we developed an app that enhances degraded retinal images using our method and automatically saves the results to support early screening and medical diagnosis.
@article{Likassa2025, author = {Likassa, Habte Tadesse and Chen, Ding-Geng and Chen, Kewei and Wang, Yalin and Zhu, Wenhui and Dumitrascu, Oana and Sun, Dayu}, journal = {Statistics and Data Science in Imaging}, title = {Robust Principal Component Analysis with Truncated Weighted Nuclear Norm and Adaptive Histogram Equalization: A Novel Method for Low-Quality Retinal Image Enhancement}, year = {2025}, issn = {2997-9676}, number = {1}, pages = {47}, volume = {2}, author+an = {7=highlight}, doi = {10.1080/29979676.2025.2538438}, publisher = {Informa UK Limited}, }
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}, }
Temporal trends in suicide ideation and attempt among youth in juvenile detention, 2016–2021

Lin Liu, Melissa Padron, Dayu Sun, and Jeremy W. Pettit

Suicide and Life-Threatening Behavior, 2025

Abs DOI Bib

Introduction: Data from the general population of youth show increases in suicide ideation and attempt in recent years, with rates of increase differing across sex and racial/ethnic groups. This study assessed trends in suicide ideation and attempt from 2016 to 2021 in youth in juvenile detention, across sex, age, and racial/ethnic groups. Methods: We leveraged state‐wide suicide screening data of all detained youth (n = 53,769) from 2016 to 2021. We analyzed data for periods defined by statistically significant changes in trends of lifetime suicide attempt, past 6‐month suicide attempt, and current suicide ideation. Results: The prevalence of lifetime, but not past six‐month, attempts increased, whereas the prevalence of current suicide ideation decreased annually from 2016 to 2021. Overall trends were qualified by distinct patterns among subgroups: rates of lifetime attempt increased among male, adolescent, and Black youth, while rates of current ideation decreased among male, adolescent, and White and Hispanic youth. Conclusion: These data document increasing rates of lifetime suicide attempts in detained youth from 2016 to 2021, especially among male and Black adolescents, concomitant with decreasing rates of current suicide ideation. Suicide prevention approaches for detained youth may need to prioritize variables besides or in addition to suicide ideation.
@article{Liu2024, author = {Liu, Lin and Padron, Melissa and Sun, Dayu and Pettit, Jeremy W.}, journal = {Suicide and Life-Threatening Behavior}, title = {Temporal trends in suicide ideation and attempt among youth in juvenile detention, 2016--2021}, year = {2025}, issn = {1943-278X}, number = {1}, pages = {e13133}, volume = {55}, author+an = {3=highlight}, doi = {10.1111/sltb.13133}, keywords = {colab}, publisher = {Wiley}, }
Functional Connectome Signatures of Patients with Asymptomatic and Typical Alzheimer’s

Xiaoqing Huang, Rishit Puri, Dayu Sun, Yi Zhao, Jie Zhang, Kun Huang, and Yijie Wang

Alzheimer’s & Dementia, 2025

Abs DOI Bib

Background: Functional magnetic resonance imaging (fMRI) has emerged as a powerful modality for investigating brain activity, offering superior spatial resolution and a non‐invasive means to probe functional connectivity. In the context of Alzheimer’s disease (AD), studying these disruptions is crucial for understanding how functional connectivity changes in salient brain activation networks for different subtypes. This study compares asymptomatic and typical AD groups to identify early alterations in brain networks that may inform future diagnostic and therapeutic strategies. Method: Three analytical pipelines were employed to characterize functional connectivity comprehensively from ADNI resting state fMRI data. The first pipeline parcellated the brain using functional atlases, then edge quantification through Graphical Lasso and group sparse covariance estimation, thereby capturing inter‐regional dependencies, using Nilearn. The second pipeline used independent component analysis (ICA) to decompose the fMRI data into distinct spatial components; mutual information was then applied to quantify the statistical relationships among these components. The third pipeline utilized FSL to perform advanced brain decomposition techniques like ICA and dual regression to generate time series that were subsequently analyzed to discern significant connectivity patterns between asymptomatic and typical AD cohorts. Result: Across all pipelines, heatmaps were generated to visualize regions of high and low brain activity, while network diagrams demonstrated varying levels of connectivity strength and hub distribution. For instance, when compared to typical AD, patients having asymptomatic AD have more regions with negative covariance. These regions are the right posterior temporal region, the default mode network. On the contrary, the regions with high positive connections or brain activity for typical AD are the auditory cortex, intraparietal sulcus, dorsal, ventral anterior cingulate cortex, and left lateral occipital complex. The posterior occipital region has strong negative connections with other regions of the brain in Typical AD. Conclusion: This study underscores the utility of fMRI‐based techniques for elucidating connectivity differences in asymptomatic and symptomatic stages of AD. By mapping out network‐level alterations, the resulting brain graphs offer valuable insights into the pathophysiology of AD, highlighting regions and pathways that may be critical in the early detection and treatment of the disease, potentially in a clinical setting utilizing fMRI as a biomarker.
@article{Huang2025, author = {Huang, Xiaoqing and Puri, Rishit and Sun, Dayu and Zhao, Yi and Zhang, Jie and Huang, Kun and Wang, Yijie}, journal = {Alzheimer's \& Dementia}, title = {Functional Connectome Signatures of Patients with Asymptomatic and Typical Alzheimer's}, year = {2025}, issn = {1552-5279}, number = {S2}, pages = {e103445}, volume = {21}, author+an = {3=highlight}, doi = {10.1002/alz70856_103445}, keywords = {colab}, publisher = {Wiley}, }

2024

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}, }
A flexible time-varying coefficient rate model for panel count data

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

Lifetime Data Analysis, 2024

Abs DOI Bib

Panel count regression is often required in recurrent event studies, where the interest is to model the event rate. Existing rate models are unable to handle time-varying covariate effects due to theoretical and computational difficulties. Mean models provide a viable alternative but are subject to the constraints of the monotonicity assumption, which tends to be violated when covariates fluctuate over time. In this paper, we present a new semiparametric rate model for panel count data along with related theoretical results. For model fitting, we present an efficient EM algorithm with three different methods for variance estimation. The algorithm allows us to sidestep the challenges of numerical integration and difficulties with the iterative convex minorant algorithm. We showed that the estimators are consistent and asymptotically normally distributed. Simulation studies confirmed an excellent finite sample performance. To illustrate, we analyzed data from a real clinical study of behavioral risk factors for sexually transmitted infections.
@article{Sun2024b, author = {Sun, Dayu and Guo, Yuanyuan and Li, Yang and Sun, Jianguo and Tu, Wanzhu}, journal = {Lifetime Data Analysis}, title = {A flexible time-varying coefficient rate model for panel count data}, year = {2024}, number = {4}, pages = {721--741}, volume = {30}, author+an = {1=highlight}, doi = {10.1007/s10985-024-09630-1}, publisher = {Springer Science and Business Media LLC}, }
A Novel RPCA Method Using Log-Weighted Nuclear and L_(2,1) Norms Combined with Contrast-Limited Adaptive Histogram Equalization (CLAHE) for High Dimensional Natural and Medical Image Data

Habte Tadesse Likassa, Ding Geng Chen, and Dayu Sun

International Journal of Statistics in Medical Research, 2024

Abs DOI Bib

Estimating the true underlying images from distorted high-dimensional data is crucial for applications in high-profile fields such as crime detection in security, clinical settings and medical diagnosis in healthcare, and radar imaging in signal processing. Existing statistical methods often struggle with robustness and image reconstruction quality when processing high-dimensional image data. While Robust Principal Component Analysis (RPCA) is widely used for image recovery, its reliance on uniform weights with singular value decomposition (SVD) weakens performance, especially in noisy environments. The L1 norm also fails to capture image details and recovery under high noise levels, a critical limitation for applications like medical diagnoses, where detail is essential. These challenges emphasize the need for improved methods to handle noise and enhance image quality in sensitive fields. Therefore, this paper proposes a novel RPCA method that integrates CLAHE with Log weighted nuclear norm (LWNN) and the L2,1 norm for high-dimensional natural and medical imaging. To reduce the computational load, our novel method is formulated into a new optimization problem and solved using the Alternating Direction Method of Multipliers (ADMM). This method leverages LWNN for enhanced low-rank approximation to drastically prune out the anomalies in images and the norm for improved sparse component recovery. Our approach has superior performance in image reconstruction compared to other state-of-the-art methods (SOTAs), showing significant advancements with real-world datasets. An interesting finding of this research is that combining the LWNN with the L2,1 norm is highly effective at removing noise from images. Furthermore, when the CLAHE technique is combined with LWNN and the L2,1 norm, it significantly enhances the extraction of previously unseen features, making blood vessels in medical images much clearer and more distinguishable. This combination proves to be a powerful approach for medical image analysis, revealing details that are otherwise difficult to detect. This method will be used for crime detection in security intelligence, and clinical settings and medical diagnosis in human retinal eyes.
@article{Likassa2024, author = {Likassa, Habte Tadesse and Chen, Ding Geng and Sun, Dayu}, journal = {International Journal of Statistics in Medical Research}, title = {A Novel {RPCA} Method Using Log-Weighted Nuclear and $L_{(2,1)}$ Norms Combined with Contrast-Limited Adaptive Histogram Equalization ({CLAHE}) for High Dimensional Natural and Medical Image Data}, year = {2024}, issn = {1929-6029}, pages = {275--290}, volume = {13}, author+an = {3=highlight}, doi = {10.6000/1929-6029.2024.13.25}, publisher = {Lifescience Global}, }
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}, }
Impact of learning health systems on cross-system collaboration between youth legal and community mental health systems: a type II hybrid effectiveness-implementation trial

Lauren O’Reilly, Dayu Sun, Katherine Schwartz, Logan Gillenwater, Allyson Dir, Patrick Monahan, Gregory A. Aarons, Lisa Saldana, Zachary Adams, Tamika Zapolski, Leslie Hulvershorn, and Matthew C. Aalsma

Implementation Science Communications, 2024

Abs DOI Bib

Background: Youth involved in the legal system have disproportionately higher rates of problematic substance use than non-involved youth. Identifying and connecting legal-involved youth to substance use intervention is critical and relies on the connection between legal and behavioral health agencies, which may be facilitated by learning health systems (LHS). We analyzed the impact of an LHS intervention on youth legal and behavioral health personnel ratings of their cross-system collaboration. We also examined organizational climate toward evidence-based practice (EBP) over and above the LHS intervention. Methods: Data were derived from a type II hybrid effectiveness trial implementing an LHS intervention with youth legal and community mental health centers (CMHCs) in eight Indiana counties. Using a stepped wedge design, counties were randomly assigned to one of three cohorts and stepped in at nine-month intervals. Counties were in the treatment phase for 18 months, after which they were in the maintenance phase. Youth legal system and CMHC personnel completed five waves of data collection (n=307 total respondents, ranging from 108-178 per wave). Cross-system collaboration was measured via the Cultural Exchange Inventory, organizational EBP climate via the Implementation Climate Scale and Implementation Citizenship Behavior Scale, and intervention via a dummy-coded indicator variable. We conducted linear mixed models to examine: 1) the treatment indicator, and 2) the treatment indicator and organizational EBP climate variables on cross-system collaboration. Results: The treatment indicator was not significantly associated with cross-system collaboration. When including the organizational EBP climate variables, the treatment indicator significantly predicted cross-system collaboration. Compared to the control phase, treatment (B=0.41, standard error [SE]=0.20) and maintenance (B=0.60, SE=0.29) phases were associated with greater cross-system collaboration output. Conclusions: The analysis may have been underpowered to detect an effect; third variables may have explained variance in cross-system collaboration, and, thus, the inclusion of important covariates may have reduced residual errors and increased the estimation precision. The LHS intervention may have affected cross-system collaboration perception and offers a promising avenue of research to determine how systems work together to improve legal-involved-youth substance use outcomes. Future research is needed to replicate results among a larger sample and examine youth-level outcomes. Trial registration: Clinicaltrials.gov identifier: NCT04499079. Registered 30 July 2020. https://clinicaltrials.gov/study/NCT04499079 .
@article{OReilly2024, author = {O'Reilly, Lauren and Sun, Dayu and Schwartz, Katherine and Gillenwater, Logan and Dir, Allyson and Monahan, Patrick and Aarons, Gregory A. and Saldana, Lisa and Adams, Zachary and Zapolski, Tamika and Hulvershorn, Leslie and Aalsma, Matthew C.}, journal = {Implementation Science Communications}, title = {Impact of learning health systems on cross-system collaboration between youth legal and community mental health systems: a type II hybrid effectiveness-implementation trial}, year = {2024}, issn = {2662-2211}, number = {1}, volume = {5}, author+an = {2=highlight}, doi = {10.1186/s43058-024-00686-6}, keywords = {colab,service}, publisher = {Springer Science and Business Media LLC}, }

2023

Regression analysis of panel count data with both time-dependent covariates and time-varying effects

Yuanyuan Guo, Dayu Sun, and Jianguo Sun

Statistica Sinica, 2023

Abs DOI Bib

Panel count data occur in many fields including clinical, demographic and industrial studies, and extensive literature has been established for their regression analysis.However, most of the existing methods apply only to the situations where both covariates and their effects are constant or one of them may be time-dependent.This paper considers a situation where both covariates and their effects may be time-dependent, and an estimating equationbased approach is developed for estimating these time-varying effects.In the method, the B-splines are employed to approximate time-dependent coefficients, and the asymptotic properties of the proposed estimators are established.To assess the finite sample performance of the proposed estimators, an extensive simulation study is conducted and suggests that the proposed method works well in practical situations.An application to the China Health and Nutrition Survey is provided.
@article{Guo2023, author = {Guo, Yuanyuan and Sun, Dayu and Sun, Jianguo}, journal = {Statistica Sinica}, title = {Regression analysis of panel count data with both time-dependent covariates and time-varying effects}, year = {2023}, number = {2}, pages = {961--981}, volume = {33}, author+an = {1=first; 2=highlight,first}, doi = {10.5705/ss.202021.0036}, publisher = {Statistica Sinica (Institute of Statistical Science)}, }
Regression analysis of general mixed recurrent event data

Ryan Sun, Dayu Sun, Liang Zhu, and Jianguo Sun

Lifetime Data Analysis, 2023

Abs DOI Bib

In modern biomedical datasets, it is common for recurrent outcomes data to be collected in an incomplete manner. More specifically, information on recurrent events is routinely recorded as a mixture of recurrent event data, panel count data, and panel binary data; we refer to this structure as general mixed recurrent event data. Although the aforementioned data types are individually well-studied, there does not appear to exist an established approach for regression analysis of the three component combination. Often, ad-hoc measures such as imputation or discarding of data are used to homogenize records prior to the analysis, but such measures lead to obvious concerns regarding robustness, loss of efficiency, and other issues. This work proposes a maximum likelihood regression estimation procedure for the combination of general mixed recurrent event data and establishes the asymptotic properties of the proposed estimators. In addition, we generalize the approach to allow for the existence of terminal events, a common complicating feature in recurrent event analysis. Numerical studies and application to the Childhood Cancer Survivor Study suggest that the proposed procedures work well in practical situations.
@article{Sun2023, author = {Sun, Ryan and Sun, Dayu and Zhu, Liang and Sun, Jianguo}, journal = {Lifetime Data Analysis}, title = {Regression analysis of general mixed recurrent event data}, year = {2023}, number = {4}, pages = {807--822}, volume = {29}, author+an = {2=highlight}, doi = {10.1007/s10985-023-09604-9}, publisher = {Springer Science and Business Media {LLC}}, }

2022

Violent victimization during reentry: prevalence, triggers, and impact on mental health

Lin Liu, Thomas J. Mowen, Christy A. Visher, and Dayu Sun

Justice Quarterly, 2022

Abs DOI Bib

Abstract Victimization is associated with a cascade of negative outcomes, and the literature has been enriched by research that situates victimization in the life context of key social groups such as children, youth, women, and veterans. Yet, less is known about violent victimization in the context of prisoner reentry. Using longitudinal data documenting reentry experiences, the current study examines the prevalence, triggers and impact of victimization among returning citizens with specific attention given to mental health outcomes. Longitudinal multilevel modeling is employed to estimate the heterogenous victimization experiences among respondents as well as the temporal change in victimization over the follow-up period of a respondent. Results underscore an alarmingly high rate of victimization against returning citizens. Risky neighborhood and family environments are significant predictors of their victimization. Upon victimization, the respondents’ mental health deteriorates. Policy implications and directions for the future research are provided.
@article{Liu2022, author = {Liu, Lin and Mowen, Thomas J. and Visher, Christy A. and Sun, Dayu}, journal = {Justice Quarterly}, title = {Violent victimization during reentry: prevalence, triggers, and impact on mental health}, year = {2022}, pages = {534-558}, volume = {40}, author+an = {4=highlight}, doi = {10.1080/07418825.2022.2104747}, keywords = {colab}, publisher = {{Routledge}}, shorttitle = {Violent {{Victimization During Reentry}}}, }

Inference of a time-varying coefficient regression model for multivariate panel count data

Yuanyuan Guo, Dayu Sun, and Jianguo Sun

Journal of Multivariate Analysis, 2022

DOI Bib

@article{Guo_2022,
  author = {Guo, Yuanyuan and Sun, Dayu and Sun, Jianguo},
  journal = {Journal of Multivariate Analysis},
  title = {Inference of a time-varying coefficient regression model for multivariate panel count data},
  year = {2022},
  pages = {105047},
  volume = {192},
  author+an = {2=highlight,corresponding},
  doi = {10.1016/j.jmva.2022.105047},
  publisher = {Elsevier {BV}},
}

2021

Regression analysis of asynchronous longitudinal data with informative observation processes

Dayu Sun, Hui Zhao, and Jianguo Sun

Computational Statistics & Data Analysis, 2021

DOI Bib

@article{Sun2021,
  author = {Sun, Dayu and Zhao, Hui and Sun, Jianguo},
  journal = {Computational Statistics {\&} Data Analysis},
  title = {Regression analysis of asynchronous longitudinal data with informative observation processes},
  year = {2021},
  pages = {107161},
  volume = {157},
  author+an = {1=highlight},
  doi = {10.1016/j.csda.2020.107161},
  publisher = {Elsevier {BV}},
}

Do both petty and serious female offenders have shorter incarcerations than their male counterparts? Testing the universality of chivalrous treatment

Lin Liu, Ronet Bachman, Jing Qiu, and Dayu Sun

Women & Criminal Justice, 2021

Abs DOI Bib

Abstract While numerous studies have unpacked gender-based disparities in judges’ sentencing decisions, few studies have examined the gender gap in correction settings. This study examines inmates’ gender gap in actual incarceration length with the effects of criminal propensities adjusted. Based on a sample including both petty and serious offenders, we use survival analysis to examine whether female inmates had shorter incarcerations to similarly situated males. We use Wilcoxon signed-rank tests to examine whether males and females who had indistinguishable lengths of incarcerations also demonstrated comparable levels of criminal propensities. Findings illustrate that the lenient treatment for females is conditional rather than universal. Females had shorter incarceration lengths only when they committed less severe crimes. Additionally, among offenders who committed less severe crimes, females demonstrated significantly lower levels of criminal propensity than males. However, among serious offenders, neither a gender gap in incarceration length nor gender disparity in criminal propensities was found.
@article{Liu2021, author = {Liu, Lin and Bachman, Ronet and Qiu, Jing and Sun, Dayu}, journal = {Women {\&} Criminal Justice}, title = {Do both petty and serious female offenders have shorter incarcerations than their male counterparts? Testing the universality of chivalrous treatment}, year = {2021}, pages = {1--15}, author+an = {4=highlight}, doi = {10.1080/08974454.2021.1962479}, keywords = {colab}, publisher = {Informa {UK} Limited}, }
When post-incarcerated individuals return to high-risk neighborhoods: staying out of trouble, social withdrawal, and mental health

Lin Liu, Christy A. Visher, Daniel J. O’Connell, and Dayu Sun

Crime & Delinquency, 2021

Abs DOI Bib

Studies show that residents from urban, high-risk neighborhoods fair worse on multiple behavioral and health outcomes than their counterparts from more socially and economically advantaged neighborhoods. However, few research efforts have been devoted to examining how formerly incarcerated individuals’ concerns over neighborhood environment are associated with reentry outcomes. Using longitudinal data that captured the reentry experiences of individuals released from prison, the present study quantifies how returning citizens’ concerns over neighborhood environment predict their social withdrawal and mental health deterioration. Findings suggest that when respondents’ post-release family bonds, financial difficulty, drug use, and past mental health histories are all taken into account, their concerns over neighborhood environment exert a significant and positive effect on social withdrawal, depression, and hostility. Returning citizens who believe it is hard to stay out of trouble and prison in their neighborhoods tend to avoid social interactions with others and experience depression and increased hostility and vigilance. Implications for reentry programing and interventions are discussed.
@article{Liu2021a, author = {Liu, Lin and Visher, Christy A. and O'Connell, Daniel J. and Sun, Dayu}, journal = {Crime {\&} Delinquency}, title = {When post-incarcerated individuals return to high-risk neighborhoods: staying out of trouble, social withdrawal, and mental health}, year = {2021}, pages = {001112872110647}, author+an = {4=highlight}, doi = {10.1177/00111287211064785}, keywords = {colab}, publisher = {{SAGE} Publications}, }

Examining the Association Between Angiopoietin-2 Levels and Acute Respiratory Distress Syndrome Subtypes in Critically Ill Patients with Sepsis

P. Yang, D. Sun, A. Manatunga, F. Harris, L. Wells, S. Bhavani, G.S. Martin, and A.M. Esper

In American Journal of Respiratory and Critical Care Medicine, 2021

DOI Bib

@inproceedings{Yang2021,
  author = {Yang, P. and Sun, D. and Manatunga, A. and Harris, F. and Wells, L. and Bhavani, S. and Martin, G.S. and Esper, A.M.},
  booktitle = {American Journal of Respiratory and Critical Care Medicine},
  title = {Examining the Association Between Angiopoietin-2 Levels and Acute Respiratory Distress Syndrome Subtypes in Critically Ill Patients with Sepsis},
  year = {2021},
  pages = {A2669-A2669},
  volume = {203},
  author+an = {2=highlight},
  doi = {10/h8rm},
  keywords = {colab},
}

2020

An early adverse experience goes a long, criminogenic, gendered way: the nexus of early adversities, adult offending, and gender

Lin Liu, Susan L. Miller, Jing Qiu, and Dayu Sun

Women & Criminal Justice, 2020

Abs DOI Bib

Abstract Early adverse experiences have been identified as a salient risk factor for crime and delinquency. However, past empirical studies predominantly used youth and young adult samples; much less is known about this risk factor’s effect on adult offending. This study examines early adverse experiences and adult pro-social bonds simultaneously using a mixed-gender sample of serious adult offenders with an average age of 35. Findings from survival analysis suggest that early adversities have an enduring detrimental effect on people’s lives well into adulthood yet in an intricate way. They have no direct effect on recidivism among adult offenders. However, they significantly influence recidivism by interacting with gender: Female respondents with early adversities demonstrated a significantly higher risk of recidivism than other female respondents, whereas no such effect was observed among male respondents. Implications for future research and policymaking are discussed.
@article{Liu2020, author = {Liu, Lin and Miller, Susan L. and Qiu, Jing and Sun, Dayu}, journal = {Women {\&} Criminal Justice}, title = {An early adverse experience goes a long, criminogenic, gendered way: the nexus of early adversities, adult offending, and gender}, year = {2020}, number = {1}, pages = {24--39}, volume = {31}, author+an = {4=highlight}, doi = {10.1080/08974454.2020.1805395}, keywords = {colab}, publisher = {Informa {UK} Limited}, }
Do released prisoners’ perceptions of neighborhood condition affect reentry outcomes?

Lin Liu, Christy A. Visher, and Dayu Sun

Criminal Justice Policy Review, 2020

Abs DOI Bib

As the United States enters a decarceration era, the factors predicting reentry success have received a rapidly growing body of research attention. Numerous studies expand beyond individual-level attributes to assess the contextual effect of neighborhoods to which released prisoners return. However, past studies predominantly used neighborhood structural/economic characteristics as the proxies of neighborhood context, leaving the roles of community cohesion and disorder understudied in the context of reentry. Using longitudinal data, this study examines the influence of neighborhood cohesion and disorder on reentry outcomes, represented by released prisoners’ determination to desist and social isolation. The results of linear regression analyses show that net of the effects of individual-level risk factors, released prisoners’ perception of neighborhood disorder exhibit profound influence on reentry outcomes. Implications for reentry programming and interventions are presented.
@article{Liu2020a, author = {Liu, Lin and Visher, Christy A. and Sun, Dayu}, journal = {Criminal Justice Policy Review}, title = {Do released prisoners' perceptions of neighborhood condition affect reentry outcomes?}, year = {2020}, number = {7}, pages = {764--789}, volume = {32}, author+an = {3=highlight}, doi = {10.1177/0887403420980806}, keywords = {colab}, publisher = {{SAGE} Publications}, }

2019

Simultaneous estimation and variable selection for incomplete event history studies

Hui Zhao, Dayu Sun, Gang Li, and Jianguo Sun

Journal of Multivariate Analysis, 2019

Abs DOI Bib

This paper discusses regression analysis of incomplete event history studies with a focus on simultaneous estimation and variable selection. Such studies are commonly performed in areas such as medical studies and social sciences, and a great deal of literature has been devoted to their analysis except for the problem considered here (Sun and Zhao, 2013). We develop a new method, which will be referred to as a broken adaptive ridge regression approach. We establish its asymptotic properties, including the oracle property and clustering effect. We also report simulation results which indicate that the proposed method performs well, and better than the existing methods, in practice. In addition, an application is provided.
@article{Zhao2019, author = {Zhao, Hui and Sun, Dayu and Li, Gang and Sun, Jianguo}, journal = {Journal of Multivariate Analysis}, title = {Simultaneous estimation and variable selection for incomplete event history studies}, year = {2019}, pages = {350--361}, volume = {171}, author+an = {2=highlight}, doi = {10.1016/j.jmva.2019.01.005}, publisher = {Elsevier {BV}}, }

2018

Variable selection for recurrent event data with broken adaptive ridge regression

Hui Zhao, Dayu Sun, Gang Li, and Jianguo Sun

Canadian Journal of Statistics, 2018

Abs DOI Bib

Recurrent event data occur in many areas such as medical studies and social sciences and a great deal of literature has been established for their analysis. On the other hand, only limited research exists on the variable selection for recurrent event data, and the existing methods can be seen as direct generalizations of the available penalized procedures for linear models and may not perform as well as expected. This article discusses simultaneous parameter estimation and variable selection and presents a new method with a new penalty function, which will be referred to as the broken adaptive ridge regression approach. In addition to the establishment of the oracle property, we also show that the proposed method has the clustering or grouping effect when covariates are highly correlated. Furthermore, a numerical study is performed and indicates that the method works well for practical situations and can outperform existing methods. An application is provided.The Canadian Journal of Statistics46: 416–428; 2018 © 2018 Statistical Society of Canada
@article{Zhao2018, author = {Zhao, Hui and Sun, Dayu and Li, Gang and Sun, Jianguo}, journal = {Canadian Journal of Statistics}, title = {Variable selection for recurrent event data with broken adaptive ridge regression}, year = {2018}, number = {3}, pages = {416--428}, volume = {46}, author+an = {2=highlight}, doi = {10.1002/cjs.11459}, publisher = {Wiley}, }

2017

Inference on an adaptive accelerated life test with application to smart-grid data-acquisition-devices

Lijuan Shen, Dayu Sun, Zhisheng Ye, and Xingqiu Zhao

Journal of Quality Technology, 2017

Abs DOI Bib

An accelerated life test (ALT) is often well planned to yield the most statistical information given limited test resources. Nevertheless, ALT planning requires rough estimates of the model parameters as an input, called planning values. The discrepancy between the planning values and the true values may result in insufficient or even no failures at the low-stress level, making the subsequent data analysis difficult. Motivated by the need in the ALTs of data acquisition devices used in smart grids, an adaptive ALT scheme is proposed. The key idea is based on the observation that, when the product reliability is underestimated during the ALT design phase, it is unlikely to observe failures at the early stage of the test. Therefore, the low-stress level should be elevated to protect against insufficient failures. Under this adaptive ALT framework, order statistics techniques are used to derive the likelihood function by assuming a general log-location-scale distribution for the product lifetime. Confidence intervals for the parameters are constructed based on the large-sample approximation as well as the accelerated bootstrap method. A simulation study is conducted to demonstrate the advantages of the adaptive ALT compared with the simple constant-stress ALT. Its application is illustrated using the motivating example from smart grids.
@article{Shen2017, author = {Shen, Lijuan and Sun, Dayu and Ye, Zhisheng and Zhao, Xingqiu}, journal = {Journal of Quality Technology}, title = {Inference on an adaptive accelerated life test with application to smart-grid data-acquisition-devices}, year = {2017}, number = {3}, pages = {191--212}, volume = {49}, author+an = {2=highlight}, doi = {10.1080/00224065.2017.11917990}, publisher = {Informa {UK} Limited}, pubstate = {Alphabetical authorship}, }