Brain maturation varies between individuals, particularly during dynamic developmental periods such as adolescence. Directly assessing the differences in longitudinal trajectories can reveal deviations from normative patterns. To ascertain the association of longitudinal change in brain volumes with birth weight, gestational age, and longitudinal changes in psychopathology. In this cohort study, cross-sectional and longitudinal normative models were developed for brain volumes from the first 2 neuroimaging data collection time points (baseline: 2016-2018; follow-up: 2019-2021) of the Adolescent Brain Cognitive Development (ABCD) Study, an ongoing community-based longitudinal cohort study at 21 US sites. Longitudinal models indexed an individual's expected brain volume at follow-up, conditioned on their baseline measurement, and thus were conditional-longitudinal models. Split-half subsets on demographically matched samples were used to fit the models. ABCD Study participants were recruited through the US school systems. Exclusion criteria included severe medical conditions that interfered with the study protocols. The present analysis further excluded the sample based on imaging quality flags and missing data. Data analysis was performed between May 2024 and August 2025. Birth weight and gestational age derived from parent-reported questionnaires. General psychopathology and subfactor scores were calculated using a bifactor model. Brain volumes and change in volume between time points 1 (baseline) and 2 (follow-up). Cross-sectional and longitudinal centiles were used to quantify deviations in volumes. The sample included 10 830 ABCD Study participants with neuroimaging data collected at baseline (mean [SD] age, 9.9 [0.62] years; 5609 males [51.8%]) and 7262 with data collected at follow-up (mean [SD] age, 12.0 [0.65] years; 3875 males [53.4%]). Longitudinal centiles were sensitive to individual-specific changes in brain volumes. Lower birth weight was associated with lower longitudinal centiles, suggesting larger decreases in brain volumes over time (n = 27 regions, β range = 0.030-0.083). Lower longitudinal centiles were associated with greater increases in psychopathology, suggesting decreasing brain volumes with increasing psychopathology scores (n = 37 regions, β range = -0.061 to -0.031). Changes in psychopathology were not associated with brain volumes at either time point when indexed by cross-sectional centiles. In this cohort study, conditional-longitudinal models captured individual-level deviations from expected growth trajectories, providing information beyond static positions on a growth curve to assess differences in maturation. Robust associations were observed between individual trajectory deviations, birth weight, and longitudinally assessed mental health symptoms. Conditional-longitudinal models hold promise for applications across psychiatric neuroscience, from development to aging.
The emergence of high-throughput sequencing technologies has generated unprecedented amounts of molecular data, posing significant challenges for analysis and interpretation. In the past 15 years, deep learning (DL) has revolutionized many data analysis fields. However, a key limitation of most DL approaches is their reliance on massive amounts of data that often require labeling. Self-supervised learning (SSL) uses large-scale unlabeled data to learn meaningful representations for specific tasks, allowing training of the models without labels. While SSL has primarily been applied to fields like natural language processing and medical image analysis, SSL can also be applied to molecular data to learn meaningful representations of molecular sequences that can be used for downstream tasks. Despite the growing interest in SSL applications using molecular data over the recent years, no comprehensive review has been published on this topic, so far, making it timely to address. This paper aims to provide researchers in DL and bioinformatics with a clear view of SSL omics applications to foster future work in the domain. This review examines the principles of SSL, such as foundation models, and it discusses the application of SSL to various omics data types, summarizes information from 17 studies, and categorizes applications by data type, detailing common tasks, model architectures, and repositories. Key applications such as DNABERT and Nucleotide Transformer are highlighted, demonstrating the contributions of SSL in understanding gene regulation. Future directions for SSL in omics are outlined, emphasizing the potential for integrating multi-omics data and developing more sophisticated pretext tasks.
Sleep disturbances are common as we age and have been linked to poor cognition and increased cognitive decline. We aimed to examine cross-sectional and longitudinal associations between objective sleep measures and cognition in middle-aged and older adults, including cognitively healthy (CH) individuals and those with mild cognitive impairment (MCI). Participants from the Aiginition Longitudinal Biomarker Investigation Of Neurodegeneration (ALBION) study (age > 40) underwent 7-day wrist actigraphy (Actiwatch 2). Sleep exposures included sleep duration, sleep efficiency, sleep variability, sleep onset latency, wake after sleep onset (WASO), and number of awakenings. A neuropsychological battery was administered examining memory, executive function, visuospatial ability, language, attention speed, and a global composite score. Cross-sectional associations were tested using generalized linear models (adjusted for age, sex, education). Longitudinal associations with cognitive trajectories were examined with linear mixed-effect models. In total (N = 184; 65% women; mean age 65 years), average sleep duration was 7.2 h and mean sleep efficiency was at 80%. Cross-sectionally, more nightly awakenings were associated with poor memory and attention speed. In a 1.5-year follow-up, (n = 93), higher baseline sleep efficiency was associated with better memory and language performance, while longer WASO, more awakenings, and longer sleep onset latency showed nominal associations with less favorable cognitive trajectories, although these associations did not remain statistically significant after FDR correction. Time-varying analyses indicated that sleep variability showed robust non-linear associations with poorer memory trajectories over follow-up and remained significant after FDR adjustment; significant mean change in awakenings and variability appeared to intensify in later follow-up phases. The association between sleep characteristics and cognitive decline varied across follow-up time, with stronger adverse changes observed during later follow-up phases. Objective indicators of sleep continuity, especially sleep variability, were most consistently related to domain-specific cognitive outcomes, with strongest evidence for memory over time. Sleep fragmentation and irregular sleep patterns may represent potentially modifiable targets for future strategies aimed at preserving cognitive health during aging.
Sex/gender-related measurement bias in autism measures has been proposed as a factor contributing to the underdiagnosis of autistic women, yet research in this area remains limited. This study builds on a small but growing body of work addressing this gap by using a modified Delphi methodology to reach consensus among autistic women and academics/clinicians on the most relevant items from five measures of autistic traits: Autism Spectrum Quotient (AQ-50), Broad Autism Phenotype Questionnaire (BAPQ), Comprehensive Autistic Trait Inventory (CATI), Girls Questionnaire for Autism Spectrum Condition (GQ-ASC), and Ritvo Autism and Asperger Diagnostic Scale-14 (RAADS-14), and to evaluate each measure's suitability for autistic women in their current form. Thirty-three autistic women and thirty-three academics/clinicians participated in the study. Each participant was asked to rate the relevance of individual items to the experiences of autistic women and to indicate whether each measure comprehensively captured experiences relevant to this group. Agreement was calculated separately for each group using item-level content validity indices, from which scale-level indices were derived for each measure. The RAADS-14 was rated as the most relevant measure for assessing autistic traits in women, followed by the CATI, while the remaining questionnaires were associated with lower perceived relevance. However, none fully captured experiences relevant to autistic women, indicating construct underrepresentation. Items assessing masking were identified as valuable for inclusion to enhance the relevance of measures for women, whereas items related to attention to detail and imagination appeared less useful for assessing autism more broadly or for capturing more nuanced presentations of autistic traits. The sample was predominantly Western, White, and highly educated, limiting the generalisability of findings. Furthermore, the expert groups lacked sufficient homogeneity, as academics and clinicians were combined and participants evaluated different subsets of measures. Although the RAADS-14 and CATI were identified as the most relevant measures for autistic women, all instruments demonstrated incomplete coverage of the autism construct in this population. Items reaching consensus as most relevant across the five measures provide a foundation for refining existing scales to better capture the experiences of individuals with more nuanced presentations of autistic traits.
Consciousness science is fragmented, and empirical investigations are largely confined within the framework of each theory of consciousness (ToC). Proliferating ToCs accumulate anomalies without progress, operating orthogonally. The principle of recurrency, functional and architectural, specifies a unifying, mechanistic scaffold for consciousness to arise. We show that all theories tacitly invoke feedback loops and can be re-expressed along a single axis of recursion. Four nested levels, viz., cellular, local inter-areal, global, and lateral, map onto state consciousness, phenomenal (P) content, conscious access (A), and phenomenal character, respectively, thus tying evolved anatomy to phenomenality. The classical P-A distinction dissolves into a graded cascade wherein deeper recursion expands the set of reportable, behaviour-driving variables. Recurrency, defined as structural, causal, and/or functional feedforward and feedback connections between elements within a system, is favoured in biological architectures, and supports the closed-loop interactions between brain, body, and environment that underlie adaptive behaviour and self-related processing. Disparities in the explanatory mechanisms across theories of consciousness can thus be resolved by converging onto this shared principle on the implementation axis, rather than deal with phenomenal-first or functional-first ontologies. The field urgently requires this implementation-first distillation that is biologically plausible, mechanistic, and empirically testable.
Diabetes is associated with an increased risk of cognitive dysfunction. We are aimed at examining cognitive performance in people with Type 2 diabetes mellitus (T2DM) and their correlation with physical activity (PA)/fitness level and mood. A cross-sectional study involving 47 individuals with T2DM, 30 diabetes-free individuals (controls): ages of 40-65 years and without dementia. Subjects were evaluated using a neuropsychological test battery that included Stroop, VLMT, CORSI, TMT and neuropsychiatric scales. The Global Physical Activity Questionnaire (GPAQ) was assessed. Physical fitness was evaluated using a maximal cardiopulmonary exercise testing (CPX). Compared with controls, people with T2DM had lower cardiorespiratory fitness (CRF) and more frequent cognitive performance below 1 SD of the normative value. Those individuals with T2DM, who displayed low fitness, showed significantly poorer cognitive inhibition (Stroop, p < 0.001), processing speed (TMT-A, p < 0.001), cognitive flexibility (TMT-B, p < 0.001), visuospatial memory (CORSI, p < 0.001) and memory consolidation (VLMT, p < 0.001). Higher ventilatory and metabolic CRF markers and higher PA levels correlated significantly with higher cognitive performance. None of the neuropsychological or psychiatric background variables correlated significantly with any of the cognitive scales. Individual CRF markers and PA levels correlated positively with cognitive performance in people with T2DM and in nondiabetic individuals. However, individuals with T2DM showed at least two to three times more frequent cognitive performances more than 1 SD below norm as compared with controls. Among people with T2DM, those with low aerobic fitness/PA levels showed significantly lower cognitive performance, especially in the attention-concentration, executive functioning, episodic memory and visuospatial processing domains.
Understanding how the brain gives rise to social cognition has been a key goal of neuroimaging research. Both changes in regional activation as well as functional connectivity have been implicated as potential mechanisms underlying social cognition, but the two have rarely been examined concurrently. Moreover, because the neural processes underlying social cognition are dynamic, developing approaches to capture dynamic changes in regional activity and functional connectivity are critical. Here, we describe a novel analysis approach that captures both regional activity and dynamic functional connectivity simultaneously during a naturalistic, socially focused movie-watching task. We found that both regional activation and functional connectivity were uniquely related to awkwardness, a judgment associated with social faux pas detection and theory of mind. Regional activation within sensorimotor networks was positively associated with awkwardness, whereas activation in the default network was negatively associated. Models including functional connectivity accounted for unique variance beyond models with activity alone. Specifically, dynamic functional connectivity between networks, primarily the frontoparietal control network, was positively associated with awkwardness. Together, these findings suggest that both dynamic regional brain activity and functional connectivity each uniquely contribute to complex and dynamic social judgments. We assessed the relationship between regional activity, functional connectivity, and awkwardness judgments during a video-watching task using fMRI in a cohort of healthy young adults. We used a novel analysis framework to simultaneously disentangle the unique contributions of dynamic regional activity and dynamic functional connectivity in their predictions of awkwardness. We found regional activity associated with awkwardness primarily in sensorimotor, frontoparietal control, and default networks, and functional connectivity associated with awkwardness within and between the visual, dorsal attention, frontoparietal control, and default networks. Finally, hierarchical models demonstrated that modeling functional connectivity significantly explained additional variance beyond modeling the activity of two regions alone, indicating the unique explanatory power offered by dynamic functional connectivity in relation to behavior.
Amid a global shift toward older populations, understanding the mechanisms of cognitive aging is a public health priority. Processing speed shows age-related decline and predicts dementia risk. Neuroimaging links dopaminergic system integrity to cognitive performance in aging, but the contribution of common genetic variation remains unclear. This study tested whether common dopaminergic variants influence 12-year processing speed decline, performance at age 70, and other cognitive domains, with exploratory analyses of post-mortem pathology. A total of 89 linkage disequilibrium-independent variants (derived from 957 SNPs) across nine dopamine pathway genes (TH, DDC, DRD1-3, SLC6A3, COMT, DBH, PPP1R1B) were analysed in 1,539 participants from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age. Across single-variant, gene-based, and unweighted pathway allele score analyses, no associations survived multiple testing correction (Bonferroni p < 5.62 × 10⁻4). For processing speed decline, the strongest nominal signals were DRD2 rs10789943 (p = 0.0066) and DBH rs2005663 (p = 0.0074), followed by DRD2 rs12805897 (p = 0.013). For performance at age 70, the leading signal was DRD2 rs11214607 (p = 0.0025). Gene-based tests were non-significant (strongest: DRD2 for slopes p = 0.063; DRD2 for intercepts p = 0.019), and the dopamine pathway allele score was unassociated with decline (β = 0.001, p = 0.969) and performance (β = 0.009, p = 0.721). Null findings extended to fluid reasoning, episodic memory, and vocabulary, and to post-mortem analyses (neuropathology n = 116; synaptic density n = 50), including SNP-marker and marker-trajectory tests. With 80% power to detect single variants explaining at least 1.19% of variance and allele score effects explaining at least 0.51% of variance, no moderate-to-large effects of common dopaminergic variation on cognitive aging trajectories were detected. Smaller effects, or mechanisms not captured by common variant analyses such as rare variants, epigenetic regulation, or gene-environment interactions, may contribute to individual differences in cognitive aging.
Oxidative stress and genotoxic damage activate NF-κB signaling through intracellular pathways distinct from those initiated by membrane receptors. DNA damage selectively induces post-translational modifications at lysine residues 277 and 309 of the human ubiquitin-binding protein NEMO to promote NF-κB signaling, but the physiological importance of these modifications in vivo remains unclear. Here, we show that a newly developed mouse model (NEMODK) carrying germline arginine substitutions of the corresponding NEMO lysine residues exhibits B-cell-intrinsic defects in germinal center formation and anti-viral humoral responses. Mechanistically, we identify in NEMODK B-cells a CD40-specific NF-κB signaling defect that is not linked to the well-characterized canonical or noncanonical NF-κB pathways. These B-cells fail to secure sustained NEMO monoubiquitination following CD40-induced ROS generation, which specifically reduces downstream RelA (p65) signaling required for the transcriptomic and epigenetic remodeling underlying homotypic B-cell aggregation, cell proliferation, class-switch recombination, and antibody-secreting cell generation. Our results establish a physiological role of murine NEMO K270 and K302 in linking CD40 engagement to B-cell responses through enabling sustained NEMO modification and RelA transcriptional activity.
Primary dysmenorrhea (PDM) is a chronic pelvic pain condition characterized by recurrent painful phases. While abnormal prostaglandin activity is central to its pain mechanism, and previous studies link PDM to central nervous system alterations associated with prostaglandin F2αlevels and pain intensity, the dynamic evolution of brain networks across the menstrual cycle remains unknown. This study employed the co-activation pattern analysis to investigate dynamic brain network characteristics across the menstrual, periovulatory, and luteal phases. Correlation analyses were performed between CAP metrics, pain scores, and PGF2α levels. Our results revealed that dynamic alterations of brain networks in patients with PDM exhibited trending changes throughout the menstrual cycle. Notably, the default mode network, salience network, sensorimotor network, and central executive network demonstrated significant periodic changes, which correlated with fluctuations in pain and PGF2α levels. This longitudinal study elucidates the dynamic neural mechanisms of patients with PDM, offering insights for early intervention strategies.
The role of macrophages in teeth, beyond their function in innate immunity, remains unexplored. This study demonstrates that macrophages populate dental tissues during early development and increase in number during pre-eruptive stages postnatally. In continuously growing teeth, they are associated with epithelial and mesenchymal stem cell niches throughout lifetime. To investigate their role in development, we genetically disrupt macrophage migration using neural-crest specific Wnt1Cre/Csf1fl/fl and general Csf1R knockout mouse models. This results in abnormal dentin and enamel deposition, eruption defects, early tooth mispatterning, and impaired osteogenesis. Notably, the phenotype is partially rescued via bone marrow transplantation. In addition, the short-term pharmacological depletion of macrophages in wildtype adult mice causes striking disintegration of both dentin and enamel in the apical part of continuously-growing teeth, which is restored after withdrawal of clodronate. Following treatment, macrophages rapidly repopulate dental tissue and the presence of M2 reparative state phenotype is observed by single-cell RNA-seq analysis. Overall, our findings reveal essential role of macrophages in the dental development and patterning of both mesenchymal and epithelial tooth compartments. This previously unrecognized role of macrophages in teeth is reminiscent of their function in complex tissue regeneration and requires future studies to dissect precise cellular and molecular mechanisms.
Road injuries are a leading cause of mortality and morbidity worldwide. Years of international efforts have aimed to strengthen policy engagement, including the 2020 UN General Assembly's proclamation of the Second Decade of Action for Road Safety (2021-30), targeting a 50% reduction in road traffic deaths and serious injuries by 2030. The aim of this study is to provide estimates to monitor progress and identify intervention gaps. As part of the Global Burden of Diseases, Injuries, and Risk Factors Study 2023, we estimated incidence, mortality, and morbidity of road injuries for 204 countries and territories from 1990 to 2023. Four road injury types and 47 nature-of-injury categories were examined. Morbidity and mortality data from clinical records, vital registration, and police reports were harmonised using meta-analytic techniques to ensure consistency and correct for systematic bias. Incidence was modelled with the meta-regression tool Disease Modelling-Meta-Regression version 2.1 and cause-specific mortality with the Cause of Death Ensemble model, both incorporating location-specific covariates to support interpolation. Years of life lived with disability (YLDs) were estimated from the prevalence and severity of the nature of road injury, and years of life lost (YLLs) from the number of cause-specific deaths multiplied by the standard life expectancy at the age of death. Disability-adjusted life-years (DALYs) were the sum of YLLs and YLDs. All metrics were calculated with 95% uncertainty intervals (UIs). In 2023, there were 50·9 million (95% UI 46·1-56·1) road injury incident cases, 1·34 million (1·04-1·58) deaths, and 75·3 million (59·8-89·2) DALYs globally. Road injuries were the leading global cause of death among males aged 10-39 years. Between 1990 and 2023, age-standardised incidence decreased by 38·3% (95% UI 36·9-39·7) and mortality decreased by 32·3% (6·1-49·0), but progress varied widely by World Bank income group. Mortality in low-income countries (43·8 [95% UI 31·7-56·0] deaths per 100 000 population) was approximately six times higher than in high-income countries (7·5 [7·1-7·9] deaths per 100 000), despite the high-income countries showing the highest age-standardised incidence rates (858·1 [95% UI 781·9-947·1] cases per 100 000). In the past decade, many countries achieved notable reductions in road injuries, but others, including Ghana and the USA, saw increases. More severe injuries tended to occur in low-income and middle-income countries. Although global incidence, mortality, and DALY rates from road injuries have declined, progress remains uneven, with pronounced disparities across income groups reflecting systemic inadequacies in infrastructure, vehicle standards, enforcement, and post-crash care. Strengthening emergency response, improving road design, enforcing safety measures, and adapting policies to the evolving demographics remain essential. Gates Foundation.
Emotion dysregulation (ED) is a core transdiagnostic feature of several psychiatric disorders, including borderline personality disorder, bipolar disorder, and attention-deficit/hyperactivity disorder. These ED disorders (EDD) exhibit overlapping clinical presentations, shared heritability, and common neurobiological substrates. This study used a transdiagnostic framework to identify early and multimodal markers of vulnerability, particularly in high-risk populations such as the offspring of EDD patients (EDDoff). A total of 237 participants (97 EDD patients, 67 EDDoff, 73 healthy controls; 139 females in total) completed a multimodal assessment including clinical evaluations, diffusion and functional MRI, and immune and neurotrophic serum biomarkers. Dimensionality reduction was performed using principal component analysis (PCA), and exploratory random forest (RF) models were trained for group classification and symptoms prediction. PCA on the full multimodal dataset yielded eight components, two of which significantly differed between groups, one reflecting high ED and altered hippocampal dynamic functional connectivity (dFC), for which EDDoff showed an intermediate phenotype, and another driven by systemic inflammation, increased in EDD patients only. Modality-specific PCA identified significant inter-modality correlations, including reduced white matter integrity with increasing immune dysregulation, and positive correlations between hippocampal dFC and both ED symptoms and inflammation (padj = < 0.01 for all correlations). A RF classifier achieved a balanced accuracy of 80.3% in distinguishing controls from EDD/EDDoff individuals. Multimodal non-clinical features predicted ED symptoms (p < 0.01). This study identifies a specific, clinically relevant, transdiagnostic and multimodal signature of vulnerability to ED, spanning behavioral, neural, and immune systems. This multimodal profile may inform future early intervention strategies targeting at-risk populations, such as EDDoff.
There is a global need for bioimage analysis (BIA) as advances in life sciences increasingly rely on cutting-edge imaging systems that have dramatically expanded the complexity and dimensionality of biological images. Turning these data into scientific discoveries requires scientists with effective data management skills and knowledge of state-of-the-art image processing and data analysis, in other words, bioimage analysts. The Global BioImage Analysts' Society (GloBIAS) aims to enhance the profile of bioimage analysts as a key role in science and research. To better understand the needs and geographical representation of the BIA community, a worldwide survey was conducted, and 290 responses were collected across people from all career stages and continents. The survey underscores a strong interest of the BIA community in the activities proposed by GloBIAS to address shortcomings in work environment, funding, and scientific activities, and the enthusiasm of the community to actively contribute to the growth and sustainability of GloBIAS as a scientific society.
Modern artificial intelligence (AI) systems have achieved remarkable capabilities, but at an extraordinary energy cost. Training and running large-scale models can consume vast resources, posing environmental, economic, and societal challenges. In contrast, biological brains perform lifelong learning, adaptive control, and flexible reasoning using orders of magnitude less energy for learning and adaptation over a lifetime. What accounts for this difference - and how can it guide future AI development? In this review, we identify key biological principles that support energy-efficient capacities in biological brains, and consider how they might inform the design of more sustainable artificial systems. We organize our analysis around three domains: architectural constraints, signaling strategies, and learning algorithms. In each domain, we discuss concrete observations from biology, from cell to circuit to cognitive level, and describe how current and emerging AI systems mirror or diverge from these motifs. One striking feature of biological energy optimization is often overlooked: that brains are remarkably stable in their energy usage across heterogeneous modes, suggesting they may minimize energy needs during active environmental processing through maximizing the utility of 'rest-like' background processes. Overall, rather than advocating for biomimicry for its own sake, we argue for biologically informed engineering. Understanding how natural systems minimize energetic cost while maximizing flexibility may help us build AI that is not only powerful, but also efficient, equitable, and environmentally responsible.
Studying the transcriptional changes in the brain following sleep deprivation has provided insight into the molecular mechanisms that differ between sleep and wake. Individual studies are limited in their ability to detect differentially expressed genes due to small sample size. Here we performed a meta-analysis of published brain expression data, totalling 173 microarrays across 245 mice. 498 genes were identified as significantly changing with sleep-deprivation at q < 0.01, 96 of which were previously identified by the original studies. Of the remaining 402 novel candidate sleep genes, 14 were associated with human sleep traits and 3 with sleep phenotypes in knockout mice. Candidate gene validation showed significant upregulation of Rasd1 (Dexras1) following sleep deprivation, and phenotyping of Rasd1 KO mice revealed changes in the amount and distribution of behavioural sleep duration and sleep bout structure. These results provide a greater understanding of the molecular correlates of sleep and provide a resource for the sleep research community.
Psychosis prevention relies on early detection of individuals at clinical high risk for psychosis (CHR-P). The effectiveness of the CHR-P state is constrained, in part, due to clinical assessments requiring specialist interpretation of narrative interviews, limiting scalability. Here, we evaluate whether large language models (LLMs; deep learning models trained on large text corpora to process and generate language) can extract clinically meaningful information from such interviews to support psychosis risk assessment. We assessed 11 open-weight LLMs on 678 partial PSYCHS interview transcripts from 373 participants (77.7% CHR-P). Models inferred CHR-P status and estimated severity and frequency across 15 symptom domains, benchmarked against researcher-rated scores. Larger models achieved the strongest classification performance (Llama-3.3-70B: accuracy = 0.80, sensitivity = 0.93, specificity = 0.58). LLM-generated symptom scores showed good correlations with researcher-rated scores (ICCsev = 0.74, ICCfreq = 0.75). Performance disparities were minimal across most demographic groups but varied across sites. Generated summaries were largely faithful to source transcripts, with low rates of clinically relevant confabulation (3%). Errors primarily reflected over-pathologisation of non-clinical experiences. While accuracy scaled with model size, smaller models achieved competitive performance with substantially lower computational cost. These findings demonstrate that open-weight LLMs have the potential to assess psychosis risk from psychometric interview transcripts, supporting scalable, human-in-the-loop approaches to early detection.
Prior research has examined the role of interpersonal relationships in adolescents' problematic use of the internet (PUI). However, studies that examine the distinct contributions of various interpersonal relationships (i.e., with mothers, fathers, teachers, and peers) to adolescent PUI at the within-person level remain lacking. Thus, this longitudinal study was conducted to address this research gap by incorporating both between- and within-person perspectives and potential sex differences. The study recruited 598 Chinese adolescents (Mage = 15.47 years at Wave 1, SD = 0.53; 54.52% girls) who participated in five waves of data collection across 2.5-years. Multilevel model was constructed to examine the between- and within-person associations between various sources of interpersonal relationships and adolescent PUI. The findings demonstrated that improved relationships with teachers and peers were related to lower levels of PUI at the between- and within-person levels, whereas no such association was found for parent-adolescent relationships. In addition, the associations between interpersonal relationships and adolescent PUI were not moderated by sex at either the between- or within-person level. These findings underscore the need to examine the distinct associations between multiple sources of interpersonal relationships and PUI among Chinese adolescents. The results revealed that positive interpersonal relationships, especially those with teachers and peers, may be an effective compensatory strategy for preventing and intervening in adolescent PUI.
Peripheral nerve blocks (PNBs) represent an accessible, minimally invasive therapeutic approach for headache disorders; however, real-world data on procedural patterns and outcomes in different headache subtypes remain limited. This multicenter study evaluated 2,219 patients (receiving 2250 PNBs) with migraine, tension-type headache (TTH), trigeminal autonomic cephalalgias (TACs), and cranial neuralgias. Demographic features, preventive and acute medicine, target nerves, anaesthetic agents, injection volumes, application intervals, number of sessions, clinical outcomes, and adverse events were analysed. Longitudinal outcomes were assessed with three-level mixed-effects models accounting for repeated visits within patients and patients within centers, using linear mixed models for VAS and cumulative-link mixed models for ordinal headache frequency and analgesic use. Migraine was the predominant diagnosis (77.5%), comprising chronic (55.6%) and episodic (44.4%) forms. Greater occipital nerve block was the dominant procedure in migraine, TTH, and TACs, but was less frequent in neuralgias, both in occipital and trigeminal. Injection volume, application interval, and local anesthetic choice varied by headache type and target nerve. GON volumes were most often 1-2 mL across headache groups, except in neuralgias, where 2-3 mL was most common. PNBs were associated with statistically significant improvements in headache frequency, severity, and duration as well as decrease in analgesic use in all analysed groups and subgroups. Adverse events were uncommon and mainly mild. Intercept-only models indicated modest center-level clustering, with ICCs of 0.093 for VAS scores, 0.134 for headache frequency, and 0.130 for analgesic use. Overall, these findings support PNBs as useful and well-tolerated options while highlighting substantial real-world variation in practice.
Research and clinical practice on childhood maltreatment largely rely on retrospective self-reports. However, self-reports are often considered unreliable due to concerns about memory biases and shifting subjective appraisals. Here, in a meta-analysis of 49 studies including n = 38,332 individuals followed for an average of 2.4 years (range = 2 months to 12 years), we found that retrospective self-reports of maltreatment are overall highly stable (r = 0.79). However, stability was lower in population-representative samples than in clinical or convenience samples, for neglect compared with abuse, and in children compared with adults. In children, but not adults, stability declined with longer follow-up. These findings challenge the view that retrospective self-reports are inherently unstable, although further research is needed to investigate long-term stability. Reshaping trauma-related appraisals may require deliberate intervention and might be most effective during childhood when memories appear more malleable.