Deformable image registration plays an essential role in various medical image tasks. Existing deep learning-based deformable registration frameworks primarily utilize convolutional neural networks (CNNs) or Transformers to learn features to predict the deformations. However, the lack of semantic information in the learned features limits the registration performance. Furthermore, the similarity metric of the loss function is often evaluated only in the pixel space, which ignores the matching of high-level anatomical features and can lead to deformation folding. To address these issues, in this work, we proposed LDM-Morph, an unsupervised deformable registration algorithm for medical image registration. LDM-Morph integrated features extracted from the latent diffusion model (LDM) to enrich the semantic information. Additionally, a latent and global feature-based cross-attention module (LGCA) was designed to enhance the interaction of semantic information from LDM and global information from multi-head self-attention operations. Finally, a hierarchical metric was proposed to evaluate the similarity of image pairs in both the original pixel space and latent-feature space, enhancing topology preservation while improving registration accuracy. Extensive experiments on four public 2D cardiac image datasets, two 3D image datasets, show that the proposed LDM-Morph framework outperformed existing state-of-the-art CNNs-and Transformers-based registration methods regarding accuracy with comparable topology preservation and computational efficiency. Our code is publicly available at: https://github.com/wujiong-hub/LDM-Morph.

Associations of spirituality and craving among individuals in substance use disorder treatment: A latent change score modeling approach.

Healthy ageing is a major public health challenge in rapidly ageing societies, where longer life expectancy does not always translate into sustained well-being. This study examines health trajectories among Chinese middle-aged and older adults, focusing on social participation and the mediating role of life satisfaction. Using four waves (2011-2018) of the China Health and Retirement Longitudinal Study, we applied latent growth curve modelling to assess overall change and individual heterogeneity and latent growth mixture modelling to identify five trajectory categories across six dimensions: environment, vitality, cognitive function, sensory and physical abilities, daily activities and psychological well-being. The five classes were high initial stable, high initial declining, moderate initial increasing, moderate initial declining and low initial increasing. Multinomial logistic regression showed that greater social participation was associated with a lower likelihood of less favourable trajectories; relative to high initial stable, the low initial increasing group exhibited a significantly reduced risk. Cultural and recreational activities showed the strongest protective association, whereas volunteer service had no significant impact. Life satisfaction partially mediated the association between social participation and trajectory membership. These results suggest that expanding meaningful social participation and improving life satisfaction are important pathways for promoting healthy ageing among Chinese middle-aged and older adults.

Trajectories of Pain Scores and Analgesic Administration Following Cardiac Surgery.

Hepatocellular carcinoma, the third leading cause of cancer-related deaths globally, presents a critical public health burden in China due to its high incidence and mortality. While targeted therapies and immunotherapies have improved survival in advanced HCC, drug resistance remains a major therapeutic challenge. Recent studies suggest that gefitinib, an EGFR inhibitor, overcomes lenvatinib resistance, yet its mechanistic underpinnings are incompletely understood. To investigate gefitinib's metabolic effects in HCC, we conducted untargeted metabolomic profiling using two separate platforms: gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) with both hydrophilic interaction liquid chromatography (HILIC) and reversed-phase modes. Raw data were processed by Mass Hunter, normalized with internal standards, and analyzed via SIMCA for pattern recognition. Principal component analysis (PCA) of quality control samples and experimental groups (n = 6 each) confirmed system stability and clear inter-group separation. Orthogonal projections to latent structures discriminant analysis models were validated by 200 permutation tests. Analysis identified 42 metabolites with VIP > 1, of which 25 showed significant alterations (p < 0.05) post-gefitinib treatment. KEGG/RaMP-DB enrichment revealed perturbations in four key pathways: arginine-proline metabolism, nitrogen metabolism, branched-chain amino acid biosynthesis, and taurine metabolism. These results delineate gefitinib-induced metabolic reprogramming in HCC cells, providing a foundation for targeting metabolic vulnerabilities to overcome therapy resistance.

Little is known about the relationship between sleep quality trajectories during pregnancy and preterm birth. To address this issue, we conducted a longitudinal assessment of maternal sleep quality to examine the relationship between sleep quality trajectories across all trimesters and the risk of preterm birth. A prospective birth cohort study was conducted in China, and a total of 15,042 women who had singleton births were included, including 647 who subsequently developed preterm birth. Sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI) scale at each study visit (8-14, 22-27 and 32-37 weeks of gestation). Preterm birth was defined as birth less than 37 weeks of gestation. Latent class trajectory models were applied to identify different sleep quality trajectories, and multivariate logistic regression models were applied to examine the associations between the determined trajectories and preterm birth. This study identified three distinct sleep quality trajectories: stable good group, stable poor group and increasing poor group. After adjusting for covariates, the odds ratio for preterm birth in the increasing poor group was 1.32 (95% confidence interval: 1.05-1.66) compared to the stable good group. Positive associations of the increasing poor group with preterm birth were exhibited only among women aged less than 30 years, those with normal BMI, or mothers of female infants. Our findings revealed that an increasing poor sleep quality trajectory was associated with an increased risk of preterm birth and emphasised the imperative to identify the high-risk groups as a priority target for intervention and treatment.

Latent behavioural modelling of multi-dimensional vehicle parking patterns for urban EV charging planning

Generative BlendPose-LDM: Utility-preserving face image anonymization via <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si229.svg" display="inline" id="d1e5635"> <mml:mi>k</mml:mi> </mml:math> -anonymous latent diffusion

TraceMark-LDM: Authenticatable watermarking for latent diffusion models via binary-guided rearrangement

Latent attention operator network with augmented representation for complex PDE systems in intricate geometries

Thyroid peroxidase antibody (TPOAb) is considered a highly sensitive marker of autoimmune thyroid diseases (AITD). Limited longitudinal data exist regarding its long-term natural history in the general population. This study aimed to assess risk factors for TPOAb positivity and its 18-year prevalence, incidence, trajectories, and projected prevalence for the year 2030 in a population-based cohort. A total of 5,438 adults were recruited at first visit (1999-2002) and followed across four subsequent visits up to 2018 in the iodine-sufficient population-based Tehran Thyroid Study (TTS). The age- and sex-standardized prevalence of TPOAb positivity was calculated, and the projected prevalence in 2030 was estimated using Poisson Generalized Linear Mixed Model (Poisson GLMM). Longitudinal trajectories of TPOAb were identified using latent class growth mixture model (LCGMM). Cox proportional hazards models were used to examine associations between potential risk factors and TPOAb positivity. The prevalence increased progressively from 11.7% at the first visit (1999-2002) to 16.3% at the fifth visit (2015-2018), and is projected to reach 21.04% in 2030. Four distinct TPOAb trajectories were identified: Low-stable (81.4%), Low-increasing (2.9%), High-decreasing (2.4%), and High-stable (13.3%). The overall incidence rate of TPOAb positivity was 5.6 per 1,000 person-years, higher among women and individuals aged <40 years. In multivariable analysis, female sex (Hazard Ratio (HR)=1.61; 95% CI: 1.15-2.27) and elevated TSH ≥5 mU/L (HR=2.69; 95% CI: 1.57-4.63) were significant positive predictors of TPOAb positivity, while age between 40 and 60 years was inversely associated with incident TPOAb positivity (HR=0.71; 95% CI: 0.55-0.90). This is the first and longest study worldwide that demonstrated a persistent rise in TPOAb positivity across five repeated measurements in an iodine-sufficient population, driven by female sex, age <40, and TSH ≥5 mU/L, which is projected to reach 21.04% in 2030. Trajectory patterns of TPOAb showed that the majority of participants had consistently low-stable levels of TPOAb.

Enhancing within-batch quality prediction by cyber-physical latent state models and incomplete first-principles

20-year trajectories of depressive symptoms and subsequent functional limitation among middle-aged and elderly population: an analysis of longitudinal cohort.

Links between hormonal and pubertal development, and adolescent females' risk for affective symptoms.

Longitudinal examination of a refined four-factor model of Protective Behavioural Strategies: Psychosocial barriers to their use and protective effects on students' alcohol consumption.

Analysis of duration between crashes among repeatedly crash-involved drivers using alternate unobserved-heterogeneity modeling approaches.

A physics-structured latent dynamics model for ion contamination diagnosis in proton exchange membrane water electrolyzers

Abstract Autoregressive next-step prediction models have become standard for building data-driven neural solvers to predict time-dependent partial differential equations (PDEs). The use of diffusion models has been shown to enhance the temporal stability of neural solvers, while its stochastic inference mechanism enables ensemble predictions and uncertainty quantification. However, a key drawback of diffusion models is the need to sample a series of discretized timesteps during both training and inference, which increases computational overhead. In addition, most diffusion models operate on structured, uniform grids, limiting their adaptability to irregular domains. To address these shortcomings, we propose a latent flow matching model for PDE simulation that embeds the PDE state in a lower-dimensional latent space, which reduces computational costs. In addition, we design an autoencoder to map different meshes onto a unified, structured latent grid, which allows predictions on complex geometries. Furthermore, we show that flow matching can result in faster and more accurate predictions than diffusion-based models, even with a coarser noise schedule. Numerical experiments show that the proposed model outperforms several deterministic and probabilistic baselines in both accuracy and long-term stability, highlighting the potential of flow matching-based approaches for data-driven PDE learning.

Developmental trajectories and predictive factors of foreign language classroom anxiety among university students: A latent growth mixture model approach.

Latinx engineering undergraduate students often face racial and gender barriers that shape their career expectations. Guided by Berry's acculturation framework, this study tested whether latent cultural orientation profiles moderate these relationships. Participants were 821 Latinx engineering undergraduate students (48% women; Mage = 20.1, SD = 1.7). Latent profile analysis identified four profiles: bicultural, moderately bicultural, assimilated, and separated. A latent mixture regression model tested whether profiles moderated prospective associations between Time 1 barriers and Time 2 engineering outcome expectations, controlling for baseline expectations and gender. Profiles significantly moderated the effect of gender barriers, Wald χ²(3) = 10.25, p = .017. Only the assimilated group (Class 3) showed a significant positive slope (b = 0.27, p = .042). In contrast, other groups showed nonsignificant slopes. No overall moderation emerged for racial barriers, Wald χ²(3) = 6.91, p = .075, though the separated group (Class 4) showed a significant negative slope (b = -0.66, p = .048), indicating vulnerability to racial barriers. Findings highlight the importance of considering cultural orientation profiles when examining how barriers influence Latinx students' career development. Whereas gender barriers may serve as a motivational challenge for assimilated students, racial barriers may undermine expectations for separated students. Psychologists and education professionals should tailor interventions to account for these cultural differences to better support Latinx engineering students' persistence and advancement in the field. (PsycInfo Database Record (c) 2026 APA, all rights reserved).