JLJinyuan Liu

Jinyuan Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Jinyuan Liu has a total of 167 co-authors. Here are the top co-authors by number of shared publications:

  • Xinlian Zhang

    Division of Biostatistics and Bioinformatics, Herbert Wertheim School of Public Health and Human Longevity Science, University of California, San Diego, La Jolla, California, USA

    5 shared publications
  • Tanya T. Nguyen

    Center for Microbiome Innovation UC San Diego La Jolla California USA

    4 shared publications
  • Hanchang Cai

    Division of Biostatistics and Bioinformatics, Herbert Wertheim School of Public Health and Human Longevity Science, University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093, United States

    3 shared publications
  • Ke Xu

    Department of Biostatistics, Vanderbilt University Medical Center, Nashville, TN, US

    3 shared publications
  • Kristan Armstrong

    Vanderbilt University Medical Center , Nashville, TN

    3 shared publications
  • Shengjia Tu

    Division of Biostatistics and Bioinformatics UC San Diego La Jolla California USA

    3 shared publications

Latest Publications

  • Open Access

    U-statistics-based Pairwise Discordance Regression for Joint Mean-Variability Inference in Longitudinal Studies with Missing Data

    by Jinyuan Liu, Amir Javid, Sarah Osmundson, Xin M. Tu, Jonathan S. Schildcrout

    Longitudinal studies are central to scientific inquiry across disciplines; yet, statistical analyses have predominantly focused on mean trajectories while treating variability as a secondary or nuisance component. Increasing evidence suggests that temporal variability and heterogeneity themselves encode critical scientific information, particularly in studies with repeated measurements, heterogeneous populations, or rich covariate structures. The classical mixed-effects and location-scale model family represents variability through latent variable components but relies on strong parametric assumptions and often requires numerical approximation or computationally intensive estimation. Here we introduce the semiparametric pairwise discordance regression, a unified framework for longitudinal data that directly models pairwise contrasts to simultaneously characterize mean trajectories, between-subject heterogeneity, and within-subject temporal variability. By targeting pairwise discordances rather than latent variance components, the proposed approach enables robust inference without specifying full parametric distributions for latent components or error terms, provided the specified conditional moment restrictions hold. Pair-level covariate encodings further accommodate flexible covariate structures and allow variability parameters to be interpreted through pairwise contrasts or dissimilarities. Estimation and inference are carried out via (weighted) U-statistic-based generalized estimating equations (UGEE) with a valid sandwich variance estimator, and we extend the fitting procedure to accommodate missing-at-random dropout. We establish the asymptotic distributions of the resulting estimators under mild regularity conditions. Simulation studies demonstrate reliable finite-sample performance in the settings considered, including non-Gaussian errors and misspecified working correlation structures, whereas likelihood-based parametric fits can be sensitive to distributional assumptions or numerically unstable. An application to longitudinal blood pressure during pregnancy illustrates the framework on real data.

  • Open Access

    Evaluating Clustering and Imputation Approaches for Multivariate Longitudinal Clinical Data Using a Real-World eCDF-Copula Resampling Framework

    by Hanchang Cai, Ke Xu, Sarrah Osmundson, Olusola Ajilore, Carmen Andreescu, Warren D Taylor, Jinyuan Liu, Xinlian Zhang

    The rapid growth of technologies such as wearable devices and high-throughput sequencing has generated unprecedented volumes of longitudinal data, often with multiple correlated features. This trend highlights the need for robust clustering methods tailored to multi-feature longitudinal datasets. Although previous reviews have benchmarked clustering methods in this context, most evaluations relied heavily on simulated and overly simplified datasets, which substantially limit their interpretability and applicability to real-world data. Missing data is also ubiquitous in longitudinal data, yet understudied in the context of clustering. To address these gaps, we systematically evaluate clustering methods across three distinct types of real-world longitudinal datasets: (1) a high-dimensional, low-timepoint metagenomic study, (2) a low-dimensional, high-timepoint multi-site repeated-measure study, and (3) a high-dimensional, high-timepoint repeated-measure study. We further investigate how missing data and different imputation strategies influence clustering performance. In addition, we introduce a novel resampling framework based on the empirical cumulative distribution function (eCDF) and a copula. This framework enables us to simulate multi-feature longitudinal data that preserves the correlation structure of real datasets, while also allowing for controlled variations such as sample size. Our simulation framework naturally accommodates missing data, thereby providing a flexible tool for benchmarking and comparing clustering methods in more realistic longitudinal settings.

  • Open Access

    Catatonic Features That May Be Mistaken for Worsening Autism Show Treatment-Associated Improvement: Implications for Later Regression

    by Joshua Ryan Smith, Maria Bonnee, Sarah Marler, Seri Lim, D. Catherine Fuchs, Rafael Tamargo, Isaac Baldwin, Christopher Maley, Ashley VanHaverbeck, Courtney Hamilton, et al.

    Catatonia in autistic individuals may resemble worsening autism yet represent a potentially treatable departure from baseline functioning. We compared item-level catatonic signs in autistic and non-autistic patients and evaluated treatment-associated change during electroconvulsive therapy (ECT). This single-center observational cohort included patients with catatonia who received ECT from May 2022 through April 2026. The 23-item Bush-Francis Catatonia Rating Scale (BFCRS) was assessed longitudinally. Pretreatment BFCRS data were available for 104 patients, including 41 autistic and 63 non-autistic patients, with 993 near-complete repeated assessments available. Pretreatment item presence was modeled using a three-level clinical-group variable with adjustment for age, biologic sex, and calendar year. Compared with non-autistic patients, patients with autism and intellectual disability had higher adjusted odds of eight activated, repetitive, or behaviorally dysregulated signs and lower odds of immobility/stupor. Among 96 patients with paired first and last eligible BFCRS assessments, multiple items improved in both cohorts. Generalized estimating equations demonstrated longitudinal declines in total BFCRS scores and multiple psychomotor-domain scores, with no clinical-group-by-time interaction surviving false-discovery rate correction in unrestricted or 30-, 60-, 90-, and 180-day models. Autism-specific Kanner analyses identified significant fixed-endpoint improvement in seven severity items after false-discovery rate correction, six of which were reproduced longitudinally. Catatonia in autistic patients with intellectual disability was characterized by a phenotype that may resemble worsening autism during later regression, and many constituent signs demonstrated treatment-associated improvement during ECT. Lay Summary: Some autistic individuals experience later behavioral, motor, or functional regression, which may have potentially treatable causes such as catatonia. In this study, patients with autism and intellectual disability were more likely to show activated, repetitive, and behaviorally dysregulated catatonic features than non-autistic patients, and these features improved during treatment. Catatonia should be considered when an autistic person develops new or substantially worsening repetitive behavior, impulsivity, aggression, communication loss, reduced self-care, mobility changes, food refusal, or other departures from prior functioning.

  • Open Access

    Electroconvulsive Therapy for Catatonia in Autistic and Non‐Autistic Patients: An Observational Study on Course, Efficacy, Aggression, and Self‐Injury Outcomes

    by Joshua Ryan Smith, Maria Bonnee, Sarah Marler, Seri Lim, D. Catherine Fuchs, Rafael Tamargo, Isaac Baldwin, Christopher Maley, Ashley VanHaverbeck, Courtney Hamilton, et al.

    Catatonia occurs disproportionately in autistic individuals and may respond to electroconvulsive therapy (ECT), yet prior reports suggest greater treatment burden in this population. We compared longitudinal ECT utilization, safety, and clinical outcomes between autistic and non‐autistic patients with catatonia. This single‐center observational cohort included 110 patients treated from May 2022 through April 2026, comprising 43 autistic and 67 non‐autistic patients. Bush‐Francis Catatonia Rating Scale (BFCRS), Clinical Global Impression‐Improvement (CGI‐I), Clinical Global Impression‐Severity (CGI‐S), and Kanner Catatonia Severity and Examination scores were recorded across acute and maintenance treatment. Because the cohorts differed in age, biologic sex, and intellectual disability, treatment burden was evaluated using covariate‐adjusted, censoring‐aware models with a three‐level clinical‐group variable comprising non‐autism, autism with intellectual disability, and autism without intellectual disability. Autistic patients received more ECT sessions than non‐autistic patients (median, 35 vs. 14) over longer observed treatment courses (median, 342 vs. 97 days). After adjustment, this difference reflected longer retention in maintenance ECT rather than a higher treatment rate and was concentrated among autistic patients with intellectual disability (Cox hazard ratio, 0.21). Clinical improvement was observed in both cohorts: CGI‐I response occurred in 100.0% of autistic and 89.1% of non‐autistic patients, BFCRS reduction of at least 50% occurred in 76.9% and 75.4%, and CGI‐S improvement of at least 1 point occurred in 85.0% and 70.8%, respectively. Kanner total scores decreased significantly, and self‐injury prevalence declined from 42.9% to 17.1% among autistic patients. Combativeness improved in both cohorts, and documented adverse events were uncommon. ECT was associated with substantial clinical improvement in autistic and non‐autistic patients with catatonia. Autistic patients, particularly those with intellectual disability, experienced longer maintenance courses rather than more intensive treatment, underscoring the importance of sustained treatment access.


  • Open Access

    Synthetic Longitudinal Tabular Data Generation via Copula

    by Hanchang Cai, Wenshan Yu, Ruijin Lu, Ishanu Chattopadhyay, Xinlian Zhang, Jinyuan Liu

    Synthetic data generation is increasingly used to enable data sharing and secondary analysis while protecting participant privacy, particularly for longitudinal tabular health data, where repeated measures per subject create within-subject dependence that most synthetic data methods are not designed to preserve. Existing generative methods, particularly generative adversarial network (GAN)-based approaches, can model complex distributions, but their estimated dependence structures are often difficult to interpret and their performance may be unstable or prone to overfitting in modestly sized datasets. Here we show that eCDF-copula, a statistically rooted approach using the empirical cumulative distribution function (eCDF) and copula modeling, preserves within- and between-visit dependence structure. To handle pervasive missing data, we propose a two-stage strategy combining multiple imputation with copula-based synthesis, enabling a variance decomposition that quantifies replication variability across methods. We benchmarked the proposed approach against four established methods on two longitudinal clinical datasets spanning markedly different sample sizes ( n = 120 vs. n = 3, 612). eCDF-copula achieved resemblance and utility exceeding those of state-of-the-art synthetic data methods, while maintaining comparable privacy.

  • Open Access

    Novel Distance Regression for Repeated Outcomes With Missing Data: Applications to Longitudinal and Crossover Studies of Microbiome Beta‐Diversity

    by Jinyuan Liu, Ke Xu, Jane F. Ferguson, Kaidi Kang, Yue Wang, Yuqi Qiu, Lucy Shao, Shengjia Tu, Tanya T. Nguyen, Tuo Lin, et al.

    The human microbiome plays a crucial role in health, but understanding its dynamic relationship with the host requires regular monitoring. Beyond challenges such as high dimensionality and sparsity, additional complexities arise, particularly within‐cluster correlation from repeated measures and pervasive missing data. To address these issues, we develop Edger, a novel distance regression method for modeling community‐level beta‐diversity dynamics and their interactions with treatment or host physiology. By focusing on beta‐diversity, a distance metric between microbial profiles, Edger ( E nsembled semiparametric d istance‐based g eneralized e stimation for r epeated outcomes) directly models these distances as repeated outcomes, yielding interpretable coefficients and enabling a covariate batching strategy to mitigate omitted variable bias. Our semiparametric inference framework eliminates the need for time‐consuming permutation tests, distinguishes between‐cluster heterogeneity from within‐cluster fluctuations, and allows flexible specification of working correlation structures. To handle missing data, we assume a missing‐at‐random (MAR) mechanism and incorporate a between‐subject propensity score in the repeated distance regression to provide seamless joint inference, ensuring robust variance estimation without casewise deletion. Additionally, we introduce an algorithm to generate synthetic data from real‐world microbial counts while preserving their zero‐inflated and correlated nature. Edger demonstrates superior inferential power and computational efficiency through our numerical studies and real‐world applications, making it a valuable tool for uncovering microbiome‐host interactions and advancing multi‐omics data integration.


  • Open Access

    Effects of cognitive behavioral therapy for worry versus befriending therapy in individuals with persecutory delusions: a randomized trial

    by Julia M Sheffield, Ali Sloan, Jinyuan Liu, Kendall Beals, Lauren Hall, Taylor Gautier, Alexandra B Moussa-Tooks, Lénie J Torregrossa, Margaret Achee, Kristan Armstrong, et al.

    Background: Worry is common and distressing in psychotic disorders, contributing to the onset and persistence of persecutory delusions. A previously developed 8-week manualized CBT-worry intervention has shown efficacy in treating persecutory delusions, compared with standard care. We aimed to test CBT-worry’s impact on persecutory delusions at 8-weeks, as previously reported, delusions at 24-weeks, and secondary clinical outcomes, relative to an active comparison therapy. Study Design: A two-arm, assessor-blinded, randomized controlled trial was conducted. 62 adults with a non-affective psychotic disorder and current persecutory delusion were randomized to CBT-worry (n=32) or befriending (n=30). The preregistered primary clinical outcome was persecutory delusion severity. Outcomes were assessed at baseline, post-treatment (8-weeks) and at follow-up (24-weeks). Results: Regarding primary outcomes, CBT-worry was not superior to befriending in treating persecutory delusions at post-treatment or follow-up (p’s>.05). For secondary exploratory outcomes, CBT-worry demonstrated superiority over befriending for self-reported worry (d=.26, p=.006), depression (d=.23, p=.018), perseveration (d=.21, p=.024), insomnia (d=.25, p=.01), and asocial beliefs (d=.23, p=.022) and assessor-rated affective symptoms (d=.36, p<.001) post-treatment. At follow-up, no significant between-group treatment effects were found after correction for multiple comparisons. Conclusions: A brief CBT-worry intervention did not outperform befriending in the treatment of persecutory delusions. CBT-worry did, however, demonstrate significant benefit for affective symptoms post-treatment, compared to befriending. At follow-up, gains were maintained in CBT-worry but were also observed in befriending, minimizing group differences. Limitations include a small sample size, lack of a non-clinical control, and 20% drop-out. ClinicalTrials.gov (NCT04748679)

  • Open Access

    Distress as a bridge to suicidality in schizophrenia spectrum disorders: a network-based intervention simulation study

    by Jinyuan Liu, Fay Womer, Julia Sheffield, Kristan Armstrong, Trey McGonigle, Jennifer Blackford, Neil Woodward, Stephan Heckers, Brandee Feola

    Schizophrenia spectrum disorders (SSD) are associated with a markedly elevated risk of suicide, yet the symptom-level mechanisms linking psychopathology to suicidality remain incompletely understood. Network-based models provide a data-driven framework for characterizing symptom interrelations, identifying candidate intervention targets, and generating hypotheses about potential downstream effects. However, the Positive and Negative Syndrome Scale (PANSS), the gold-standard assessment of schizophrenia symptoms, has primarily been used as a static rating tool rather than as a dynamic simulation framework for intervention prioritization. Here, we developed a network-based simulation framework using PANSS to identify symptom domains most strongly associated with suicidality and to evaluate how targeted modulation of these symptoms may relate to suicidal ideation. Symptom networks reproduced a robust five-domain PANSS structure (Positive, Negative, Cognitive Impairment, Impulsive-Hostile, and Distress). Across item- and cluster-level analyses, distress-related symptoms (anxiety, depression, guilt, and tension) showed the most consistent associations with suicidality and emerged as key bridge symptoms. Analyses stratified by lifetime suicide attempt history revealed distinct networks and simulation responses across risk strata. Simulation-based modulation of the Distress domain was associated with attenuated network connectivity and lower predicted suicidal ideation. While derived from simulations, these findings highlight affective distress as a clinically relevant symptom pathway linked to suicide risk in SSD and demonstrate how network-informed digital simulations may support the prioritization of symptom targets. Future studies, particularly randomized controlled trials, will be critical to determine whether targeting distress leads to measurable reductions in suicidality, with simulation-based analyses complementing observational evidence by informing hypothesis generation and trial design.

  • Open Access

    Between- and Within-subject Variability in Actigraphy Data: A Case Study on Sleep Patterns in Schizophrenia

    by Jinyuan Liu, Hanchang Cai, Zhichao Zhang, Jiawei Wang, Ellen Lee, Xinlian Zhang

    Variability in longitudinal health outcomes provides critical insight into disease dynamics, yet most statistical analyses continue to prioritize mean trajectories. This limitation is particularly consequential in psychiatric and mobile health (mHealth) research, where intensive repeated measurements increasingly reveal pronounced heterogeneity and temporal instability that cannot be captured by average responses alone. Standard mixed-effects models typically impose homoscedastic variance assumptions and relegate variability to a nuisance component, despite mounting evidence that both variability information, at both between- and within-subject level, carry independent clinical significance. Here, we describe an improved mixed-effects location–scale model for longitudinal outcomes that simultaneously characterizes mean structure, between-subject heterogeneity, and within-subject fluctuations. The proposed framework explicitly links variance components to both time-invariant and time-varying covariates, extending classical mixed-effects models to accommodate heteroscedastic and dynamically evolving variability. By jointly modeling location and scale parameters, the method enables direct inference on variability (or stability) of the data alongside mean trends. We evaluated the approach using an actigraphy-based mHealth case study of sleep patterns in schizophrenia. Our results demonstrate that explicitly modeling variability uncovers clinically relevant structure obscured by mean-based analyses, underscoring the importance of variability-aware modeling for modern longitudinal and mHealth research.

  • Open Access

    Multi-modal and multi-region distance model for neuroimaging: Application to ABCD study

    by Xinyu Zhang, Simon Vandekar, Andrew A. Chen, Kaidi Kang, Jakob Seidlitz, Aaron Alexander-Bloch, Jinyuan Liu

    Large-scale neuroimaging studies often collect multiple modalities, such as task and resting-state functional MRI, diffusion MRI, and structural MRI. Joint inference across these modalities uses shared variation to improve statistical efficiency, increase replicability, and provide a more integrated view of brain-phenotype associations. In practice, however, such analyses are limited because complex cross-modality covariance cannot be flexibly modeled, which makes the resulting joint effects difficult to interpret. A recent distance-based ANOVA extension allows multimodal analysis and increases power for detecting group differences, but it cannot easily distinguish location from scale effects in distance space, offers only an omnibus pseudo- F test without interpretable parameters, and requires computationally intensive permutation inference. We propose a novel semiparametric, U-statistics-based Generalized Estimating Equation (UGEE) framework that unifies univariate and multivariate distance models. By regressing pairwise dissimilarities on covariates, this method yields interpretable regression coefficients that disentangle location and scale effects and quantify inter-modality differences, while flexibly accounting for correlations among modality distances. The estimator is based on efficient influence functions, ensuring asymptotic efficiency, robustness to misspecification, and computational scalability for large-scale data analysis. We evaluate the proposed method through extensive simulations and analyses of the Adolescent Brain Cognitive Development dataset. Results show that UGEE accurately estimates modality, group, and interaction effects and achieves a 100-fold speed-up compared with permutation-based approaches. This framework provides a general and computationally efficient tool for semiparametric inference on multimodal data, particularly suited for large neuroimaging applications.