Qingyu Zhao

Adolescence involves pronounced changes in sleep timing and duration that have been linked to executive functioning, impulsivity, and risk behaviors, yet evidence for generalizable sleep - self-regulation associations remains limited. We used regularized canonical correlation analysis to examine multivariate relationships between sleep and self-regulation across two independent cohorts. In the Adolescent Brain Cognitive Development study (n = 11,143), later chronotype and shorter sleep duration were associated with higher impulsivity, particularly urgency and deficits in planning and perseverance (r = 0.21, permutation p < 0.001), with minimal contribution from delay discounting. Social jetlag showed weak loadings. In the National Consortium on Alcohol and Neurodevelopment in Adolescence cohort (n = 831), only one canonical function was statistically significant (r = 0.15, p = 0.011), characterized by later chronotype and greater social jetlag, with minimal contribution from sleep duration and primary loadings on impulsivity-related measures. Applying the ABCD-derived sleep - self-regulation pattern to NCANDA produced a small but nonzero association, driven primarily by social jetlag and delayed sleep timing, with minimal influence of sleep duration. These findings provide multivariate evidence linking delayed sleep timing and circadian misalignment to adolescent self-regulation, though effect sizes attenuate across cohorts. Longitudinal work is needed to clarify predictive pathways and inform sleep-focused interventions during this vulnerable developmental period.

The emergence of psychiatric symptoms during adolescence is increasingly hypothesized as arising from maturational imbalance across brain systems. However, this concept largely lacks quantitative grounding, which requires measuring fine-grained imbalance patterns that accurately capture acceleration or delay relative to normative developmental trajectories of brain regions. To address this gap, we leverage predictive normative modeling to learn models that predict chronological age from regional multivariate functional connectivity patterns. We demonstrate that these region-specific models are highly generalizable across independent cohorts and capture greater developmental effects than traditional functional connectivity metrics. From these models, we then derive a region-wise Relative Maturity (RM) index that quantifies individualized, region-specific deviations from normative development. Rigorous cross-cohort and longitudinal evaluations across four datasets show that RM maps are reproducible, subject-specific fingerprints of neurodevelopmental imbalance. These fingerprints are organized along continuous, low-dimensional axes aligned with intrinsic functional gradients and can predict dimensions of psychopathological vulnerability. Together, our findings establish RM as a robust, sensitive, and generalizable framework for quantifying individual vulnerability to psychopathology through system-level patterns of developmental imbalance.

Adolescence is a critical period of brain maturation, yet how functional dynamics relate to white-matter microstructure at the individual level remains poorly understood. We developed a machine learning workflow to predict fractional anisotropy (FA) from distributed resting-state functional connectivity (FC) in two large adolescent cohorts: NCANDA (n = 814, 12–22 years, longitudinal) and HCP-Development (n = 472, 12–22 years, cross-sectional). Whole-brain FA could be modestly predicted from FC (r = 0.16 in NCANDA; r = 0.27 in HCP-D). The accuracy of predicting regional FA significantly varied across 27 white-matter regions, with the highest structure-function coupling detected in fiber tracts subserving unimodal cortical regions. These regional accuracy scores were reproducible between datasets (r = 0.95). Region-specific analyses also revealed tract-clustered FC predictors, highlighting distinct large-scale functional circuits underlying regional microstructural integrity. We then defined a “structure–function gap” as the residual between predicted and observed FA in each white-matter region. These gap measures were significantly associated with a broad constellation of cognitive and behavioral measures, particularly involving memory, impulsivity, and executive function. Notably, significant behavioral coupling emerged in Corticospinal and Cingulum pathways. Together, these findings establish individualized structure–function coupling as a reproducible, anatomically specific, and behaviorally informative marker in youth, offering a new framework to link distributed FC patterns to white-matter development and behavioral variability.

Accurately modeling longitudinal brain MRI progression is crucial for understanding neurodegenerative diseases and predicting individualized structural changes. Existing state-of-the-art approaches, such as Brain Latent Progression (BrLP), often use multi-stage training pipelines with auxiliary conditioning modules but suffer from architectural complexity, suboptimal use of conditional clinical covariates, and limited guarantees of anatomical consistency. We propose Anatomically Guided Latent Diffusion Model (AG-LDM), a segmentation-guided framework that enforces anatomically consistent progression while substantially simplifying the training pipeline. AG-LDM conditions latent diffusion by directly fusing baseline anatomy, noisy follow-up states, and clinical covariates at the input level, a strategy that avoids auxiliary control networks by learning a unified, end-to-end model that represents both anatomy and progression. A lightweight 3D tissue segmentation model (WarpSeg) provides explicit anatomical supervision during both autoencoder fine-tuning and diffusion model training, ensuring consistent brain tissue boundaries and morphometric fidelity. Experiments on 31,713 ADNI longitudinal pairs and zero-shot evaluation on OASIS-3 demonstrate that AG-LDM matches or surpasses more complex diffusion models, achieving state-of-the-art image quality and 15-20\% reduction in volumetric errors in generated images. AG-LDM also exhibits markedly stronger utilization of temporal and clinical covariates (up to 31.5x higher sensitivity than BrLP) and generates biologically plausible counterfactual trajectories, accurately capturing hallmarks of Alzheimer's progression such as limbic atrophy and ventricular expansion. These results highlight AG-LDM as an efficient, anatomically grounded framework for reliable brain MRI progression modeling.

Key TakeawaysAlcohol-related publications using machine learning almost exclusively relied on conventional techniques, whereas current public discourse emphasizes state-of-the-art models.A majority of models predicted alcohol consumption or alcohol use disorder (AUD) diagnosis, which is generally easier to forecast than, for example, disorder or treatment outcome.Only 40% of models utilized multimoda

An individual’s risk of substance use disorder (SUD) is shaped by potent biosocial factors. Family history is one of the strongest predictors, yet its impact on the brain before substance exposure remains unclear. Here we apply network control theory to estimate transition energies (TEs)—the input required for the brain to shift between activity patterns—in youth from the Adolescent Brain Cognitive Development Study. Family history of SUD was associated with altered TE, expressed as sex-divergent effects across brain scales alongside elevated TE in specific regions in both sexes. Females with a family history showed higher TE in the default mode network, whereas males showed lower TE in dorsal and ventral attention networks. These findings demonstrate sex-specific influences of family history on brain dynamics and underscore the importance of considering sex as a biological variable when studying adolescent neurodevelopment and mechanisms of SUD risk.

Generative models enhance neuroimaging through data augmentation, quality improvement, and rare condition studies. Despite advances in realistic synthetic MRIs, evaluations focus on texture and perception, lacking sensitivity to crucial morphometric fidelity. This study proposes a new metric, called WASABI (Wasserstein-Based Anatomical Brain Index), to assess the morphometric plausibility of synthetic brain MRIs. WASABI leverages SynthSeg, a deep learning-based brain parcellation tool, to derive volumetric measures of brain regions in each MRI and uses the multivariate Wasserstein distance to compare distributions between real and synthetic anatomies. Based on controlled experiments on two real datasets and synthetic MRIs from five generative models, WASABI demonstrates higher sensitivity in quantifying morphometric discrepancies compared to traditional image-level metrics, even when synthetic images achieve near-perfect visual quality. Our findings advocate for shifting the evaluation paradigm beyond visual inspection and conventional metrics, emphasizing morphometric fidelity as a crucial benchmark for clinically meaningful brain MRI synthesis. Our code is available at https://github.com/BahramJafrasteh/wasabi-mri.

Importance: Maturational imbalance between the expression of heightened socioemotional drivers and immature executive control creates a period of vulnerability to risk-taking behaviors during adolescence. Developmental imbalance of the brain functional activity coupled with these behaviors may explain the propensity to develop heavy drinking. Objective: To compare the longitudinal trajectories of socioemotion and executive control in non–heavy drinkers with those of heavy drinkers before and after drinking onset. Design, Setting, and Participants: This cohort study was conducted from January 13, 2013, to January 15, 2022, among participants from the National Consortium on Alcohol and Neurodevelopment in Adolescence cohort with resting-state functional magnetic resonance imaging and neuropsychological assessment data. All participants were individuals who abstained from drinking or drank very little at the baseline visit. The statistical analysis was conducted between October 11, 2024, and July 3, 2025. Exposures: Categorical drinking levels that summarized self-reported alcohol consumption in the past year. Main Outcomes and Measures: Brain-behavioral system z scores that reflected the association of selective brain connectivity patterns with selected behaviors were calculated. Higher scores indicated higher reactivity strength of socioemotion or executive control. Results: A total of 3076 visits among 633 participants (mean [SD] age at baseline, 15.7 [2.6] years; 318 female [50.2%]) were analyzed. A total of 238 participants (37.6%) who initiated heavy drinking during the study showed significantly elevated brain-behavior scores (score increase, 0.18; 95% CI, 0.08-0.28) in the socioemotional system prior to drinking onset, but their developmental trajectories of the executive control system followed those of the control participants. After drinking onset, developmental deviation occurred in both the socioemotional system (score increase, 0.44; 95% CI, 0.33-0.54) and executive control system (score increase, 0.21; 95% CI 0.11-0.32), with a significantly larger effect size in the socioemotional system ( z = 3.51; P < .001) compared with before initiating drinking. Conclusions and Relevance: These findings suggest that asynchronized maturation of socioemotion- and executive function–coupled brain connectivity in adolescence may be a risk factor for heavy drinking onset in emerging adulthood. In turn, drinking was associated with deficits in the executive control system and exacerbated alteration in the socioemotional system. Recognition of these early-age developmental discrepancies could aid in averting alcohol use disorder in adulthood.