Nature Mental Health

The increased vulnerability to mental health problems in adolescence is frequently reported but poorly understood, hampered by a rigid diagnostic system which fails to capture intertwining symptoms and only loosely aligns with biological axes of variability. Here, we reconceptualised the mental health symptoms of young adolescents in the ABCD cohort (N=11862) as a latent topology of overlapping symptom dimensions, using an unsupervised machine learning algorithm to establish how transdiagnostic dimensions co-occur and overlap within individuals. Combining this with a novel classification approach, we delineated zones within this landscape, within which specific profiles of symptoms were robustly represented. These data-driven profiles were leveraged to establish associated resting-state functional connectivity and genetic characteristics. In doing so we recaptured the commonly reported p-factor axis as well as further symptom-subtype dimensions. Gene ontology analysis revealed that shared neurobiological and cellular mechanisms embedded in both the genome and transcriptome may confer risk for psychopathology.

Non-suicidal self-injury (NSSI) is common among adolescents with emotional disorders, yet biological indicators of current NSSI status remain limited. We developed a genome-aware multi-omics modeling framework in 107 adolescents with emotional disorders, including 53 without NSSI and 54 with current NSSI. The model integrated metabolomic, inflammatory, clinical blood and genome-derived features, with polygenic risk score and rare variant burden used as genetic-context variables. The fusion model achieved the strongest classification performance (mean AUC = 0.811) and outperformed single-omics alternatives, indicating that NSSI status was better represented by distributed multi-omics patterns than by a single biomarker layer. Repeated modeling prioritized 42 stable features, many of which were not significant in conventional univariate testing. Group-specific network reconstruction further revealed peripheral reorganization, including convergence of non-NSSI modules into an NSSI-associated module that linked inflammatory recruitment with weaker immune-communication, repair and support-related signals. Exploratory MRI, gut-related and stress-endocrine analyses provided additional biological anchors, while a compact sentinel marker panel translated the full model into clinically readable profiles. These findings support a distributed, genome-aware peripheral state associated with current NSSI and provide a framework for future validation of multi-omics state markers in adolescent emotional disorders.

Adolescent externalizing behavior is a major risk factor for later substance use and other psychiatric outcomes. Understanding its genetic architecture and its relationship with brain imaging phenotypes requires scalable genome-wide methods applied to youth cohorts. Using data from the Adolescent Brain Cognitive Development (ABCD) Study, we implemented a pipeline for genome-wide association studies (GWAS) of longitudinally measured externalizing traits and multimodal neuroimaging-derived phenotypes (IDPs). We performed quality-controlled genotype processing and constructed harmonized phenotype and covariate datasets. GWAS analyses were conducted using REGENIE in a two-step framework, with Step 1 ridge regression models trained on LD-pruned variants and Step 2 association testing performed genome-wide. Externalizing traits measured at baseline and summarized as longitudinal means and slopes, together with approximately 200 IDPs measured at baseline and summarized as longitudinal means and slopes, were analyzed. We further constructed a custom linkage disequilibrium (LD) reference panel using unrelated individuals and computed LD scores using LDSC. Genetic correlations between externalizing traits and imaging phenotypes were estimated using LD Score Regression. This exploratory study systematically evaluated genome-wide genetic correlations between regional cortical morphology and externalizing phenotypes in adolescence. Although several associations reached nominal significance, none remained significant after correction for multiple comparisons. These findings should not be interpreted as demonstrating an absence of shared genetic architecture. Rather, the precision of the estimates was constrained by the available imaging GWAS sample size, uncertainty in SNP-heritability estimates, and the large number of regional comparisons. Larger imaging-genetics samples and independent replication will be required to determine whether modest or regionally specific genetic correlations exist.

Background: Opioid use disorder (OUD) is heritable, yet most genome-wide association studies (GWAS) have focused on European populations, leaving the genetic architecture of OUD in non-European populations underexplored. Methods: We conducted GWAS of OUD across three ancestries using electronic health records and genomic data from 52,357 All of Us Research Program participants (8,912 cases; 43,445 matched opioid-exposed controls; 48.5% female). Participants were stratified into European (EUR), African (AFR), and Admixed American (AMR) ancestry groups for logistic regression GWAS, with independent replication in the Million Veteran Program. We then applied the deep-learning model AlphaGenome to predict the tissue-specific transcriptomic and splicing consequences of top risk variants across 13 reward-pathway brain regions. Results: We identified and replicated a novel DDX6 risk locus, alongside established OPRM1 and FURIN signals. AlphaGenome predicted the DDX6 regulatory allele downregulates the stress-resistance gene FOXR1 in the nucleus accumbens, while the protective OPRM1 variant (rs1799971) upregulates OPRM1 expression across reward networks. Other signals of interest included IL6R and SHISA9 (EUR); GHR (AFR); and ASTN2 (AMR). Conclusions: This study identifies DDX6 as a novel OUD risk locus, replicates associations with OPRM1 and FURIN, and highlights biologically plausible ancestry-specific signals in AFR and AMR populations. We also replicated top variants in an independent population. Finally, integrating GWAS with deep-learning annotations provides specific, localized biological hypotheses to guide future experimental validation and targeted therapeutics.

Sensory processing differences are a core feature of autism, affecting 60-95% of individuals, yet the associated neural mechanisms remain unclear. An excitation-inhibition (E/I) imbalance in brain circuits has been proposed, but in vivo evidence linking genetic variation in E/I pathways, regional neurochemistry, neural circuit function, and sensory behaviour has been lacking. Here we performed a multimodal investigation in 206 individuals (130 autistic), integrating gene-set polygenic scores for excitatory glutamatergic and inhibitory gamma-aminobutyric acid (GABA)-ergic pathways, magnetic resonance spectroscopy (MRS) measures of regional GABA and Glx (glutamate + glutamine) levels, vibrotactile psychophysical measures of tactile perception, and questionnaire measures of behavioural sensory reactivity. We found that glutamatergic polygenic scores predicted thalamic glutamate levels in neurotypical but not autistic individuals, suggesting altered genotype-neurochemistry coupling in autism. Thalamic Glx:GABA levels associated with tactile perception in both groups, but with opposing directions of effect, indicating that autistic and neurotypical individuals achieve similar perceptual outcomes with potentially differing thalamocortical circuit mechanisms. Within autistic individuals, tactile perceptual differences further related to behavioural sensory reactivity. Together, these findings suggest that autistic sensory processing potentially relies on distinct circuit mechanisms linking genetic variation, neurochemistry and perception. This work thus has important implications for how sensory differences are conceptualised, studied, and interpreted, and ultimately for how interventions and support are developed.

Background: Post-traumatic stress disorder (PTSD) remains highly prevalent affecting ~23% of World Trade Center (WTC) responders more than two decades after 9/11. While MRI studies have identified neural differences associated with PTSD, these findings have not translated into improved treatment. We introduce a novel multimodal MRI approach, DAta-driven Network Connectivity Estimate (DANCE), integrating structural and functional magnetic resonance imaging (MRI) to better capture PTSD mechanisms and inform biomarkers. Methods: In 96 WTC responders , including 45 with current WTC-related PTSD and 51 without PTSD. We applied graph theory to resting-state functional MRI to identify functional hubs via eigenvector centrality and identified divergence between groups using partial least squares discriminant analysis (PLS-DA). From diffusion MRI, we reconstructed five anatomical tracts (i.e., streamlines) in the temporal lobes. Using DANCE, we quantified the differential distribution of streamlines of the reconstructed tracts connecting the functional hubs. We then tested whether WTC exposure duration moderated associations between PTSD and DANCE indices. Results: Responders with PTSD showed altered centrality in nine functional hubs (AUC=0.75 (0.651-0.847)) including bilateral anterior inferior temporal gyrus, right superior parietal lobule, right anterior parahippocampal gyrus, right anterior/posterior superior temporal gyrus (STG), right caudate nucleus, left amygdala and brainstem. Connectivity differences emerged in four tracts: hippocampus, parahippocampus, inferior and superior temporal gyri (STG). DANCE differed in the inferior fronto-occipital fasciculus (IFOF), medial (IFLmed) and lateral (IFLlat) components of the inferior longitudinal fasciculus and in the middle longitudinal fascicle (MdLF). WTC exposure duration significantly moderated the association between PTSD and DANCE values in the IFLmed, right posterior STG (p= 0.035). Conclusion: Our novel DANCE approach revealed converging functional and anatomical connectivity alterations uniquely associated with PTSD in WTC responders and offers compelling evidence for distinct neurobiological signatures of the disorder. These findings significantly advance our understanding of PTSD pathophysiology and highlight potential biomarkers for diagnosis and targeted intervention.

Background: Depression is clinically heterogeneous, and immune-metabolic disturbances may not map uniformly onto categorical diagnosis or symptom severity. In populations exposed to substantial structural adversity, it remains unclear whether inflammatory and metabolic biomarker differences reflect adversity exposure itself or cluster around specific depressive phenotypes. Methods: Data were drawn from the Sao Paulo Megacity Mental Health Survey, a population-based study in which 5,037 household residents underwent structured psychiatric interviews. Among 770 participants assessed as having clinically significant symptoms (SCID-I), individuals with chronic physical illnesses, hs-CRP >20 mg/L, or missing data were excluded, yielding an analytic sample of 653. Latent class analysis of 16 DSM-IV depressive symptoms was used to identify symptom-derived phenotypes. Multinomial logistic regression tested associations between latent classes and immune-metabolic biomarkers, including hs-CRP, lipid fractions, fasting glucose, and triglycerides, adjusting for age, sex, education, smoking, and BMI. Results: A four-class solution identified asymptomatic (44.56%), mild-moderate (19.14%), atypical-like (16.69%), and melancholic-like (19.60%) classes. The two highseverity classes diverged by neurovegetative features: atypical-like depression was characterised by weight gain, hypersomnia, and psychomotor retardation, whereas melancholic-like depression was characterised by weight loss, insomnia, and psychomotor agitation. hs-CRP was highest in the atypical-like class and lowest in the melancholic-like class despite similar symptom severity. After BMI adjustment, elevated hs-CRP in the atypical-like class attenuated, whereas lower hs-CRP in the melancholic-like class persisted. Other metabolic markers did not robustly differentiate classes after adjustment. Conclusions: Immune-metabolic differences in depression do not simply follow symptom severity. In this Sao Paulo cohort, higher and lower hs-CRP profiles clustered around distinct latent depressive phenotypes, supporting biologically heterogeneous depressive presentations within a socially exposed urban population.

Background: Suicide attempts (SA) are elevated in individuals with autism spectrum disorder (ASD), but population-level data characterizing how SA prevalence varies across demographic and clinical subgroups - at the scale and granularity needed to inform evidence-based risk stratification - have been largely unavailable. This study examines SA prevalence across sex, age group, psychiatric comorbidity type, and substance use disorder subtype in the largest real-world ASD cohort to date. Methods: We conducted a retrospective cross-sectional analysis using Epic Cosmos electronic health record data from 2,311,171 individuals with ASD identified by International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes, spanning 2016-2025. SA prevalence was calculated with Wilson score 95% confidence intervals. Modified Poisson regression with robust variance estimation was used to estimate adjusted prevalence ratios (aPRs) for sex, age group, and psychiatric comorbidity. Unadjusted prevalence ratios (uPRs) were calculated separately for individual comorbidity types and substance use disorder subtypes. Results: Overall SA prevalence was 1.7% (38,160 individuals). Females showed higher SA prevalence than males (2.9% vs. 1.2%; aPR 1.61, 95% CI 1.35-1.92). SA prevalence peaked in the 15-24 age group overall (aPR 5.14, 95% CI 3.62-7.29), with sex-stratified analyses revealing that females peaked earlier (15-24 years; aPR 4.33, 95% CI 4.23-4.43) than males (25-34 years; aPR 6.08, 95% CI 5.05-7.31) - a sex-specific divergence in the timing of peak SA prevalence not previously documented in ASD. Having at least one psychiatric comorbidity was associated with a 30-fold higher SA prevalence (aPR 30.56, 95% CI 26.03-35.89), with the effect stronger in females (aPR 36.54, 95% CI 31.04-43.01) than males (aPR 27.77, 95% CI 22.65-34.06). Among comorbidity subtypes, substance-related disorders showed the highest crude SA prevalence (16.9%; uPR 134.20, 95% CI 67.68-266.10). Subtype-level characterization revealed SA prevalence ranging from 19.7% to 24.3% across all five substance use disorder subtypes examined, with stimulant use disorder showing the highest unadjusted prevalence ratio of any subtype (uPR 196.11, 95% CI 100.63-382.20). Conclusion: SA prevalence in ASD is markedly elevated relative to the general population and varies meaningfully by sex, age, and comorbidity profile in clinically important ways. Females carry a disproportionate SA burden relative to males, with peak vulnerability arriving earlier in adolescence; males peak later in young adulthood and remain at elevated risk into midlife. Psychiatric comorbidity - particularly substance use disorders - is associated with the largest relative elevations in SA prevalence. These population-level estimates are directly applicable to EHR-based risk stratification models and can inform the development of ASD-specific clinical decision support tools that concentrate surveillance and intervention on those at demonstrably elevated risk, rather than applying uniform approaches across a heterogeneous population.

Background Brain abnormalities related to childhood adversity (CA) have been reported across clinical presentations in psychotic disorder (PD) and bipolar disorder (BD). This systematic review and meta-analysis examined gray matter volume (GMV) alterations linked to CA in PD and BD. Methods A PRISMA-compliant systematic review was conducted (PROSPERO ID: CRD42022351133). The EMBASE, MEDLINE, and PsycINFO databases were searched from inception to June 2024 for studies investigating CA and structural brain imaging in PD and BD. Study quality was assessed with the Newcastle Ottawa Scale (NOS). Data were extracted and synthesized accounting for sex differences and CA subtypes with brain findings categorized by the presence and direction of associations. Meta-analyses were performed for hippocampal and amygdala volumes. Results In the systematic review (k = 29), 3,056 participants with PD and BD (mean age = 36.6; SD =16.1; 47% female), published between 2011 and 2023, were included. Study quality was fair, with high heterogeneity. Most studies reported significant negative associations between CA and GMV, especially in prefrontal regions, while findings for the hippocampus and amygdala were largely null or inconsistent. Meta-analyses of a study subset identified no significant association between CA and hemisphere-specific and combined volumes of the hippocampus (k = 5; p [≥] 8805; 0.66) or amygdala (k = 4; p [≥] 8805; 0.87). Conclusion CA was not consistently associated with hippocampal or amygdala volume alterations in PD and BD. More consistent evidence emerged for reduced GMV in prefrontal regions, suggesting that neurobiological impact of CA may be more robustly captured at the cortical level.

Importance: Patients who achieve remission from major depressive disorder (MDD) often face a preference-sensitive decision between continued antidepressant maintenance and discontinuation with active monitoring. Quantifying the tradeoff between depression burden and long-term medication exposure may support more individualized shared decision-making. Objective: To quantify tradeoffs between continuous antidepressant maintenance and active monitoring after MDD remission, and to identify preference thresholds favoring each strategy across relapse-risk strata. Design: Individual-level decision-analytic health-state transition model calibrated to randomized maintenance-discontinuation trials and a longitudinal first depressive episode cohort, with a 5-year time horizon. Setting: Outpatient clinical decision after completion of an 8-month continuation phase following remission from MDD. Participants: Adults in remission from MDD, represented across 4 clinically anchored relapse-risk strata ranging from very low risk after a first mild episode to high risk after highly recurrent depression. Exposures: Continuous antidepressant maintenance vs discontinuation with active monitoring and antidepressant restart after detected relapse. Main Outcomes and Measures: Severity-weighted depression-months, antidepressant medication-years, medication-years per depression-month averted, and net benefit across preference thresholds defined as the maximum additional medication-years a patient would be willing to accept to avert 1 depression-month. Results: Continuous maintenance reduced depression burden but required substantially more medication exposure, with efficiency strongly dependent on relapse risk. Medication-years per depression-month averted ranged from 11.8 (95% uncertainty interval [UI], 7.8-19.6) in the very low-risk group to 1.5 (95% UI, 0.8-3.0) in the high-risk group. At a preference threshold of 3 medication-years per depression-month averted, maintenance was preferred for moderate- and high-risk patients; at a threshold of 2, only for high-risk patients; and at a threshold of 1, for no risk group. Conclusions and Relevance: In this decision-analytic model, the value of continuous antidepressant maintenance depended strongly on baseline relapse risk and patient preferences regarding long-term medication exposure. These findings provide a quantitative framework for shared decision-making about antidepressant maintenance after remission from MDD.

Abstract Background Despite a rising global psychiatric burden, a treatment gap persists where the majority of symptomatic individuals remain unmedicated. Traditional epidemiological analyses treat this untreated population as a single, uniform block, obscuring specific barriers to care. This study uses an unsupervised machine learning pipeline to identify distinct socio-behavioural and biological sub-populations within the untreated cohort to guide targeted public health interventions. Methods Data were pooled from the 2015-2018 National Health and Nutrition Examination Survey (NHANES) cycles (N=11,848 total adult respondents). A symptomatic cohort of 3,075 individuals experiencing daily or weekly anxiety or depression symptoms was isolated, excluding severe liver pathology outliers ("GGT"[≥]80" U/L" ). A 22-feature matrix combining continuous clinical biomarkers (systolic blood pressure, waist circumference, HbA1c) and categorical social variables was projected using Factor Analysis of Mixed Data (FAMD). Latent sub-populations were identified via Gaussian Mixture Modelling (GMM), optimized by the Bayesian Information Criterion (BIC). Results The broad baseline population revealed a substantial mental health burden, with 30.4% reporting active psychiatric symptoms, of whom 71.6% were entirely unmedicated. The GMM pipeline successfully isolated three distinct sub-populations (k=3) separated by age, clinical strain, and treatment rates: Cluster 0 (Mature Adults, mean age 55.03): high psychiatric severity (34.1% severe untreated), central obesity, and hypertensive strain (135.82 mmHg), with 64.1% untreated despite frequent primary care contact; Cluster 1 (Working Professionals, mean age 38.38): highly educated, female-dominated (70.5%), with 77.7% untreated driven by moderate distress; Cluster 2 (Emerging Youth, mean age 18.49): a highly vulnerable late-adolescent group with a staggering 90.2% untreated rate. Conclusion The unmedicated symptomatic population is highly diverse and segmented by life stage. These profiles show that the treatment gap is driven by age-specific barriers, specifically workforce-age symptom masking and late-adolescent developmental transitions. Closing this deficit requires shifting from uniform public health approaches toward targeted interventions, such as digital peer support networks for youth and integrated primary care screenings for older adults.

Objective: Working memory (WM) deficits are a co-occurring feature to numerous neuropsychiatric disorders, particularly attention-deficit/hyperactivity disorder (ADHD), and there remain no treatments that directly target WM. The coupling between the phase of theta band activity and amplitude of gamma band activity (i.e., TGC) is an established neural correlate of WM. However, no studies have examined WM-related TGC in ADHD or whether neuromodulation can modulate these oscillatory dynamics in youth. This set of studies examined the effects of intermittent theta burst stimulation (iTBS) to the left dorsolateral prefrontal cortex (DLPFC) and left posterior parietal cortex (PPC) on TGC in youth with ADHD. Methods: In two randomized, double-blind, sham-controlled crossover trials, adolescents with ADHD and clinically significant parent-reported WM symptoms first completed a single-session study comparing DLPFC versus PPC iTBS targeting (n = 47) and then a multi-session clinical trial comparing 10 sessions of active versus sham left DLPFC iTBS (n = 29). Participants completed a computerized visuospatial Sternberg WM task with concurrent electroencephalography (EEG) before and after the single sessions, as well as at baseline, midway through treatment, and approximately 24 hours after the final session within the multi-session trial. Phase-amplitude coupling between theta phase and gamma amplitude was quantified using the Kullback Leibler modulation index at frontoparietal electrodes. Linear mixed-effects models examined treatment effects and associations between change in TGC and WM status (including accuracy, reaction time, and clinical symptoms). Results: Across participants, lower TGC was associated with lower symptoms and better WM performance, including higher accuracy, faster and more consistent RT. Active iTBS increased frontoparietal TGC relative to sham stimulation, with effects observed both acutely after a single session and ~24 hours after multiple sessions. DLPFC-targeted iTBS increased TGC, whereas PPC-iTBS had no measurable effect. Change in TGC was associated with change in WM, such that a decrease in TGC was associated with faster RT and decreased RT variability. Higher baseline TGC was associated with greater improvement in WM. Active iTBS decoupled the TGC-WM association observed during sham iTBS, and greater electric field intensity of iTBS was associated with greater improvement in WM accuracy and greater decrease in TGC. Conclusions: Active iTBS to the left DLPFC modulated WM-related TGC in youth with ADHD. These findings provide preliminary evidence that neuromodulation may improve WW by modifying oscillatory dynamics within frontoparietal networks. Larger clinical trials with higher stimulation doses are needed to determine whether targeting oscillatory coupling represents a potential therapeutic strategy for WM deficits.

Background: Actigraphy-derived rest-activity rhythm (RAR) features are widely used to characterize clinical states in bipolar disorder (BD). Both mean levels and temporal variability of these features have been associated with mood episodes; however, variability measures are often statistically coupled with the mean, particularly in skewed distributions. This raises a question as to whether variability reflects a separate characteristic of the data or whether the observed association arises from statistical properties of the data. Objective: In this study, we aim to determine whether temporal variability of actigraphy-derived RAR features provides standalone information on mood episodes in BD beyond mean activity levels after accounting for mean-variance dependence. Methods: We analyzed actigraphy data from a subset of 72 participants with BD drawn from a larger longitudinal study, extracting 22 daily RAR features aggregated weekly as sample mean (MEAN) and within-week temporal variability computed as sample standard deviation (VAR). Variance-stabilizing transformations (Box-Cox or Yeo-Johnson) were applied to the entire study cohort to reduce mean-variance dependence. Associations with mood episodes and remission (mania: n=34; depression: n=58 annotated participants) were evaluated using generalized linear mixed-effects models with a logistic link function, including univariate (MEAN or VAR) and multivariate (MEAN+VAR) specifications, assessed by likelihood-based metrics and the area under the receiver operating characteristic curve (AUC). Results: Transformations reduced mean-absolute correlations from 0.43 to below 0.06. Temporal variability remained significantly associated with clinical state for 11/22 RAR features in mania and 16/22 features in depression, with all significant associations remaining after false discovery rate correction (p<0.05). Joint models showed modest incremental gains (AUC 3%-4% overall; up to 12% in mania, 7% in depression), with absolute performance remaining limited (AUC 0.50-0.66). In both mania and depression, nearly all significant variability-based regressors contributed incremental information beyond mean-based models. Only sleep duration and activity changes around wake time (+-1 hour), did not improve discrimination between mania and remission. Conclusions: Temporal variability in RAR features can be considered a standalone state marker of mood episodes not captured by mean activity. We found it to be more consistently associated with depression than mania. Its incremental discriminative contribution is modest, suggesting greater utility within multivariate or multimodal frameworks.

Background The global prevalence of autism spectrum disorder (ASD) has significantly increased over the past two decades. Despite substantial research advances, critical aspects, including etiology, diagnostic biomarkers, and pharmacological interventions, remain incompletely elucidated. This persistent knowledge gap warrants systematic mapping of the field's evolution to inform future research priorities. Methods A bibliometric analysis of ASD-related publications indexed in Web of Science was conducted from January 2020 to May 2025. Following a systematic deduplication process, original articles, reviews, case reports, and clinical trials were included in the analysis. The analytical framework comprised co-authorship networks, institutional collaboration patterns, national research contributions, and keyword co-occurrence structures, all of which were examined using CiteSpace (version 5.8.R3) and VOSviewer. Results After deduplication, 8,162 publications (January 2020-May 2025) were analyzed. The annual output grew steadily, confirming ASD as a sustained priority in neuroscience. Research remains academia-driven, led by the United States, with China as the second-largest contributor. Chinese institutions place greater emphasis on mechanistic and developmental phenotyping, which aligns with national priorities. These studies maintain strong methodological rigor, and their growing volume underscores the central role of ASD in translational neuroscience. Conclusion Future research on ASD should focus on strengthening case identification, refining clinical phenotyping, and expanding large-scale cohort studies to advance our understanding of its etiology and identify reliable diagnostic biomarkers. It is equally important to develop and evaluate targeted interventions for core symptoms and integrate telemedicine into service delivery models. A critical yet understudied priority is improving the quality of life for autistic individuals and their families, an area in which research globally, including in China, requires greater depth and consistency. With China's growing investment in autism research, it is well-positioned to contribute to these pressing international challenges.

Background: Childhood adversity (CA) has been identified as one of the most robust risk factors for psychotic disorders; several treatable mediating mechanisms have been proposed. Aims: To conduct a systematic review and meta-analysis examining mediating pathways linking CA and psychosis. Method: This PRISMA-compliant systematic review (PROSPERO: CRD42024542972). consisted of a search conducted in January 2026 on Ovid (PsycINFO, Medline, and Embase) using search terms related to psychosis, CA, and mediation analyses. Evidence was appraised by calculating the percentage of the total effect mediated in each study, grouping mediators into meaningful groups. When possible, meta-analyses using two-stage meta-analytic structural equation modelling (METASEM) were conducted. Results: 117 studies were included (54 in clinical samples, 59 in non-clinical samples, and four studies in both clinical and non-clinical samples). 107 studies examined psychological mediators and 12 examined biological. The median percentages of total effect mediated across all analyses per mediator family were: 49% for dissociation (k = 24), 45% for psychosocial stressors (k = 6), 37.9% for negative schemas (k = 23), 35.2% for post-traumatic symptoms (k = 10), 31.5% for depressive symptoms (k = 14), 27.8% for anxiety (k = 11), 27.1% for attachment styles (k = 12), and 8.7% for mentalization domains (k = 5). Meta-analyses confirmed a robust mediating effect of dissociation (k = 7; N = 2143; indirect effect (I.E) =0.42 [0.17, 0.66] on psychosis; 50.49%), on delusions (k =7; N = 1053; I.E = 0.36, [0.27, 0.46]; 46.44%]) and on hallucinations (k =10; N = 5705; I.E = 0.28 [0.20, 0.36]; 57.59%). Robust mediation via depression (k =5; N= 5028; indirect effect= 0.33 [0.31, 0.35]; 31.05%) and negative schemas of the association between trauma and psychosis broadly defined (k = 7; N=10791; I.E= 0.26 [0.17, 0.35]; 26.36%) was also observed. High heterogeneity was observed across all meta-analyses. Fewer studies examined biological mediators, preventing quantitative synthesis. Conclusions: Childhood adversity impacts psychosis through psychosocial mediators, particularly dissociation. Further work is required to on the potential role of biological mechanisms and its interplay with psychological mechanisms.

The emotional content of suicide notes is typically examined using categorical coding, where each labeled passage is treated in isolation from its surrounding language. In contrast, dimensional models of psychopathology propose that affective content varies along continuous gradients. We evaluated this proposition directly. Excerpts from 884 annotated suicide notes were embedded in a semantic space defined solely by their linguistic properties, and we investigated whether human-assigned emotion labels changed smoothly across this space. They did: affective tone showed clear spatial autocorrelation (Moran's $I = 0.18$, $z = 19.68$, $p < 0.001$), an effect that replicated across three different encoders and remained after removing all within-note dependencies. Emotions occupied recognizable yet overlapping regions rather than forming distinct clusters and varied substantially in how tightly they were concentrated: love and hopelessness appeared with similar frequency, but love was far more localized ($z = 15.7$ versus $10.8$). Among all emotions, hopelessness was the most linguistically diffuse, implying that a single categorical label is capturing multiple, qualitatively different manifestations of suicidal distress.

Background: Emotion dysregulation is a core feature of bipolar disorder (BD), yet its behavioral expression during depressive episodes, and potential differences between its types, BD-I and BD-II, remain unclear. This study used automated facial-expression analysis during naturalistic affective film viewing to examine subtype-specific and context-dependent emotional responding in bipolar depression. Methods: The sample included 135 participants: 69 healthy controls and 66 patients with BD (BD-I, 23; BD-II, 43). Participants viewed nine emotionally evocative film clips spanning negative, positive, neutral, and socially threatening contexts, while their facial expressions were continuously recorded and quantified using computer vision-based facial-expression analysis. Results: Patients with BD-I showed a distinct, context-dependent facial-expression profile, characterized by greater negative responses across multiple contexts than other groups. Specifically, they showed increased sadness during sad, reward, and amusing clips, and elevated anger during sad and neutral clips. In socially threatening contexts, BD-I participants showed a multivalent pattern of elevated anger, fear, and joy, suggesting poorly coordinated or context-incongruent affective expression. In contrast, BD-II participants did not differ significantly from healthy controls on any emotion, despite depressive symptom severity comparable to BD-I participants. Conclusions: These findings suggest that facial-expression patterns in bipolar depression differ across subtypes. BD-I may be characterized by heightened negative reactivity and altered context-appropriate modulation of emotional expression, whereas BD-II may not show comparable alterations in overt facial output. Automated facial-expression analysis during naturalistic stimulation may provide a useful behavioral marker for characterizing subtype-specific affective disturbance in bipolar depression and related psychopathology.

Large Language Models are increasingly used in consumer-facing mental health tools, many of which claim that prompt engineering alone can ensure safe therapeutic behavior. This study evaluates that assumption by testing 20 proprietary and open-source LLMs on high-risk psychiatric scenarios, using prompts grounded in behavioral therapy principles. Prompt engineering reduced some predictable risks, such as explicit endorsement of self-harm, but consistently failed in ambiguous or clinically nuanced situations. Models frequently validated harmful statements, colluded with hallucinations, minimized symptoms, or used stigmatizing language, including in the newest and largest models. These failures reflect structural limitations such as lack of memory, insufficient contextual reasoning, and training-related biases. Prompt engineering alone is therefore insufficient for safe AI-mediated psychotherapy; clinician-guided fine-tuning, integrated safety mechanisms, and system-level oversight will be required. This work provides early evidence motivating deeper clinician-led evaluation and safety-oriented model development.

Schizophrenia and bipolar disorder are diagnostically distinct categories that overlap substantially in clinical features and genetic aetiology. Understanding genetic variants that contribute liability specifically to each disorder can offer insights into biological processes that differentiate them. Here we used Case-Case GWAS (CC-GWAS) to identify common genetic variants differentially associated with schizophrenia and bipolar disorder, analysing 67,390 schizophrenia cases and 41,917 bipolar disorder cases. We identified 19 genome-wide significant loci, of which 16 (84%) demonstrated divergent genetic effects with risk alleles showing opposite directions of association between disorders. The CC-GWAS summary statistics had detectable disorder-differentiating heritability (10.27%, SE=0.01) and showed genetic correlations indicating that SCZ-differentiating alleles were associated with lower educational attainment, lower cognitive performance, and increased risk of ADHD, anorexia, autism, BD1 (though not BD2), cannabis use disorder, and OCD. Four loci showed divergent effects despite not reaching genome-wide significance in either individual disorder GWAS, demonstrating enhanced power to detect opposite-direction effects. Functional annotation identified 102 mapped genes significantly enriched for expression across all 13 tested brain regions, with no significant enrichment in peripheral tissues, and gene set enrichment analysis implicated neuronal projection and synaptic compartments as the strongest biological themes differentiating the two disorders. Polygenic risk scores derived from these disorder-differentiating variants were associated with earlier age at onset and more severe negative symptoms in schizophrenia, consistent with these variants marking neurodevelopmental dimensions of illness. Our findings provide targets for understanding pathogenic differences between schizophrenia and bipolar disorder and demonstrate that genuine divergent genetic effects exist beyond the substantial shared liability.

Objective Psychosis spectrum symptoms (PSS) are common among children and youth. These symptoms may be clinically significant as studies indicate a heightened risk of mental health disorders, in general, as well as psychotic disorders, specifically, in youth that endorse PSS. This systematic review and meta-analysis investigates the longitudinal association between PSS in children and youth and subsequent mental health diagnosis. Methods A comprehensive search of Ovid Medline, PsycINFO, and EMBASE databases was conducted to identify longitudinal studies that: (i) assess PSS at a baseline timepoint, (ii) in individuals under 25 years, and (iii) assess mental health disorder diagnosis using a structured assessment at a later time point in the same sample. We conducted a meta-analysis and calculated pooled odds ratios (ORs) for mental health and psychotic disorders using random-effects models. Post-hoc meta-regressions were performed to examine the influence of a number of moderators on the relationship between earlier recorded PSS and subsequent mental health disorders or psychotic disorders. Results The search yielded 41 eligible studies of which 25 were included in the meta-analysis. Most included studies assessed PSS using brief self-report measures and recruited their samples from clinical or community settings. Among children and youth without an identified mental health diagnosis at baseline assessment, baseline PSS were associated with a 2-fold (OR = 2.07, CI = 1.61 - 2.66, I2 = 86.92%, p < 0.0001) increased risk of meeting diagnostic criteria for subsequent mental health disorder diagnosis and a 3-fold increased risk (OR = 3.11, CI = 2.11 - 4.58, (I2 = 60.93%, p < 0.0090) of meeting diagnostic criteria for a subsequent psychotic disorder diagnosis with a minimum 1 year follow-up time from baseline assessment. Meta-regression analysis indicated that study quality and sample size explained a substantial proportion of between-study heterogeneity for psychotic disorder outcomes. Conclusions Our results suggest that administration of simple self-report measures of PSS in both clinical and community settings may be helpful to identify children and youth at higher risk of subsequently meeting criteria for a mental disorder generally, and for a severe mental illness (i.e., psychotic disorder), specifically. Future longitudinal studies should focus on improving study design characteristics to increase confidence in identified longitudinal associations. The results of our work suggests that integration of self-report measures of PSS may be useful in a variety of settings to identify youth at increased risk of subsequent mental illness.