BACKGROUND:Formal thought disorder (FTD) is a highly disabling transdiagnostic feature that impedes communication and social ties. Progress in understanding and treating FTD has been hampered by the uncertainties in its assessment. AIMS:We examined if a short 3-5min assessment of transcribed speech can capture the latent dimensions and network structure of FTD and predict functional outcomes. METHOD:In a transdiagnostic sample (N = 666) with a single longitudinal follow-up over 3-12 months (n = 244), we administered the short form of the Thought and Language Index to measure eight individual features of FTD. We determined the baseline factor structure of FTD, its temporal invariance at follow-up, and the predictive validity of FTD dimensions on the global single-item Social and Occupational Functioning Assessment Scale scores at baseline and follow-up. We identified the most influential and putative primary phenomena within the FTD syndrome, using network analysis. RESULTS:Factor analyses revealed a stable three-factor model of FTD: impoverishment (poverty of speech, weakening of goal), loosening (looseness, illogicality) and peculiarities (peculiar words, peculiar sentences), with excellent fit (Comparative Fit Index: 0.997, root mean square error of approximation: 0.040) and metric invariance over time. Impoverishment and peculiarities predicted functioning at baseline and 3-12 months later (cross-sectional: β = -0.196, p < 0.001 and β = -0.298, p = 0.001, respectively; longitudinal: β = -0.201, p = 0.037 and β = -0.336, p = 0.042, respectively). Looseness and poverty of speech were putative primary features influencing other FTD phenomena. Weakening of goal and peculiar sentences were the most connected phenomena. CONCLUSIONS:By integrating latent variable and network approaches, we provide a unified, empirically grounded framework to interpret FTD assessed using a brief speech task. We report a replicable three-dimensional structure, identify central symptoms that may maintain the FTD syndrome, and the specific dimensions that influence functional disability. These findings clarify the prognostically valuable features of FTD for future mechanistic and interventional research.

BACKGROUND:Thought disorder (TD) is a core feature of severe mental illnesses such as schizophrenia, characterized by disruptions in speech, language, and communication. People with TD face unique barriers that hinder their involvement in research, both as participants and as partners. Their systematic underrepresentation in psychiatric research is driven by pervasive assumptions about their decisional capacity, willingness to participate, and ability to engage in research. This perpetuates a biased evidence base, likely hindering the therapeutic progress toward addressing this core problem. METHODS:This review, informed by professional (clinical and research) and lived (bottom-up and phenomenological) experience of TD, examines how flawed assumptions regarding capacity, engagement, and participatory abilities serve as active barriers to inclusion. RESULTS:We argue for a shift toward supported inclusion through tailored capacity assessments, enhanced informed consent procedures, targeted training of research personnel, and systemic institutional practices. Incorporating lived experiences of those with TD as research partners is integral to this approach, fostering co-production of research that is more valid, inclusive, and applicable. CONCLUSIONS:Without these inclusion-focused changes, the development of treatments for TD is likely to have very slow progress and a critical segment of the severely unwell population will continue to be underrepresented from the scientific process, undermining both the utility and generalizability of psychiatric research.

Background:Psychotic disorder represents a leading cause of disability worldwide, and relapse in psychosis is common. Artificial intelligence (AI) is increasingly recognized as a method that could aid clinical monitoring for individuals experiencing psychosis. Objective:This review aims to map the existing literature on AI-based approaches-including machine learning, deep learning, and natural language processing-used to detect relapse in individuals with psychotic disorders. Methods:A systematic search strategy was conducted on PubMed, PsycINFO, and Embase up to January 7, 2026. Observational studies, randomized controlled trials, and quasi-experimental studies that used AI methods to detect relapse in psychosis were eligible for inclusion. Screening and data extraction procedures were conducted by at least 2 reviewers working independently. Findings were extracted, charted, and described using narrative synthesis based on data extraction and consensus meetings with the research team. The scoping review was prospectively registered with the Open Science Framework. Results:Relevant studies identified (N=10) included the use of digital tools such as smartphone- and smartwatch-based monitoring, ecological momentary assessment tools, social media activity, and internet searches. Digital phenotyping via smartphones and wearables emerged as the most common method for data collection. The efficacy of AI models varied with sensitivity (or recall) ranging from 0.25 to 0.77 and specificity (or precision) ranging from 0.06 to 0.88. The reported area under the receiver operating characteristic curve for models ranged from 0.63 to 0.78. AI models were heterogeneous across studies, and most study findings were not replicated. Conclusions:This scoping review highlights both the promise and the current limitations of AI in psychosis relapse detection. Passive digital phenotyping research in the detection of psychosis relapse has progressed, and personalized approaches with individual-level modeling show promise; however, further studies need to include larger numbers of participants and should incorporate methods such as large language models. Future studies will require large collaborations aimed at delivering AI methods for use in real-world clinical practice.

BACKGROUND:Sex differences in recovery outcomes for psychosis have been observed. In patients with psychosis, rates of insecure attachment are significantly higher in patients of both sexes compared to the general population and have been linked to several aspects of recovery. However, the possible differential effect of attachment style on recovery between men and women with psychosis is currently unknown. METHODS:This study was performed in a subsample of 299 patients remitted from their first-episode psychosis (FEP) within the Handling Antipsychotic Medication Long-term Evaluation of Targeted Treatment (HAMLETT) study. First, t-tests were used to explore sex differences in baseline attachment style. Second, stepwise regression analyses were used to examine the association between baseline attachment style and symptomatic, social and personal recovery at three and 48 months follow-up, and the possible moderation effect of sex on these associations. Third, stepwise regression analyses were repeated with longitudinal change in symptomatic, social and personal recovery between three and 48 months follow-up as outcome measure. RESULTS:Male and female patients did not differ in baseline attachment style. Baseline attachment style was associated with recovery outcome at 3-months follow-up, whilst sex did not moderate this relationship. Baseline attachment style did not predict recovery outcome at 48-months follow-up, nor change in recovery outcome. CONCLUSIONS:Our findings suggest that attachment style is an important predictor of short-term recovery outcome in patients with FEP, whilst sex differences do not appear to significantly impact this relationship.

Introduction Substance use is common among patients with a first episode of psychosis (FEP). While long-term studies link substance use to poorer outcomes, short-term associations, particularly involving polysubstance use, remain understudied. This study examined substance use patterns 3 to 6 months after remission and their associations with symptoms, cognition, and functioning. Methods Data were derived from the HAMLETT randomized controlled trial, which compared continuation versus early discontinuation of antipsychotic medication following remission, using assessments at baseline, 3, and 6 months. Substance use patterns were categorized as no substance use, tobacco use, tobacco plus cannabis use, or polysubstance use (i.e. tobacco, cannabis and other illicit drugs). Associations with psychotic symptoms (PANSS), cognition (BACS, Stroop), and functioning (WHODAS 2.0) were examined using linear mixed-effects models adjusted for age, sex, education, alcohol use, and randomization condition. Results Of 379 FEP patients, 283 were included in the analyses: 161 reported no substance use, 83 tobacco use, 26 tobacco plus cannabis use, and 13 polysubstance use. Substance use patterns remained stable over six months. Polysubstance use was associated with more severe positive symptoms compared to other substance use groups and poorer self-care, compared to patients who do not use substances. Differences in positive symptoms were primarily observed in patients continuing antipsychotic medication, suggesting that positive symptoms reflect substance use-related clinical differences. Conclusion Polysubstance use represents a risk profile in the early post-remission phase of FEP, characterized by elevated positive symptoms and functional impairment in self-care, suggesting early disruption of daily functioning during the post-remission phase.

Capturing subtle speech disruptions across the psychosis spectrum is challenging because of the inherent variability in speech patterns. This variability reflects individual differences and the fluctuating nature of symptoms in both clinical and non-clinical populations. Accounting for uncertainty in speech data is essential for predicting symptom severity and improving diagnostic precision. Speech disruptions characteristic of psychosis appear across the spectrum, including in non-clinical individuals. We develop an uncertainty-aware model integrating acoustic and linguistic features to predict symptom severity and psychosis-related traits. Quantifying uncertainty in specific modalities allows the model to address speech variability, improving prediction accuracy. We analyzed speech data from 114 participants, including 32 individuals with early psychosis and 82 with low or high schizotypy, collected through structured interviews, semi-structured autobiographical tasks, and narrative-driven interactions in German. The model improved prediction accuracy, reducing RMSE and achieving an F1-score of 83 with ECE = 4.5e-2, showing robust performance across different interaction contexts. Uncertainty estimation improved model interpretability by identifying reliability differences in speech markers such as pitch variability, fluency disruptions, and spectral instability. The model dynamically adjusted to task structures, weighting acoustic features more in structured settings and linguistic features in unstructured contexts. This approach strengthens early detection, personalized assessment, and clinical decision-making in psychosis-spectrum research.

Psychotic disorder represents a leading cause of disability worldwide, and relapse in psychosis is common. Artificial intelligence (AI) is increasingly recognized as a method which could aid clinical monitoring in psychosis. This scoping review aims to identify studies which have used methods with an AI component to detect relapse in psychosis. A systematic search strategy was conducted on PubMed, PsycINFO and Embase from inception to January 2026. Observational studies, randomized controlled trials and quasi-experimental studies which used AI methods to detect relapse in psychosis were eligible for inclusion. Screening and data extraction procedures were conducted by at least two reviewers working independently. Findings were extracted, charted and described using narrative synthesis based on data extraction and consensus meetings with the research team. The scoping review was prospectively registered with Open Science Framework. Relevant studies identified (n = 10) included use of digital tools such as smartphone and smartwatch-based monitoring, ecological momentary assessment tools, social media activity and internet searches. Digital phenotyping via smartphones and wearables emerged as the most common method for data collection. Efficacy of AI models varied with sensitivity (or recall) ranging from 0.25 to 0.77 and specificity ranging from 0.06 to 0.88. Reported area under the receiver operating characteristic curve for models ranged from 0.63 to 0.78. AI models were heterogenous across studies, and most study findings were not replicated. This scoping review highlights both the promise and current limitations of AI in psychosis relapse prediction. Digital phenotyping research in detection of psychosis relapse has progressed, but future studies need to include larger numbers of participants and should incorporate other methods such as use of large language models. Future studies will require large collaborations aiming to deliver AI tools for use in real world clinical practice. N/A

BACKGROUND:Formal thought disorder (FTD) is a key determinant of social functioning in schizophrenia. However, existing syntheses have not differentiated positive versus negative FTD dimensions, which have distinct trajectories, mechanisms, and treatment implications. We examined the differential associations of positive and negative FTD with social functioning. METHODS:A comprehensive systematic review with stricter inclusion criteria to update prior meta-analyses, searching Scopus and PubMed for publications up to 31 January 2026. Bayesian random-effects meta-analyses were performed separately for negative FTD (k = 7 studies, N = 1226) and positive FTD (k = 24, N = 4072). Bias was assessed using a modified Newcastle-Ottawa Scale. Robust Bayesian methods assessed publication bias. Protocol pre-registered with OSF: 10.17605/OSF.IO/WMT3X. RESULTS:Both dimensions showed negative associations with social functioning. Positive FTD demonstrated a robust pooled effect (r = -0.31, 95% CrI: -0.40 to -0.20) with extreme evidence (BF10 > 100, posterior probability ~100%). Negative FTD showed a smaller effect (r = -0.21, 95% CrI: -0.38 to 0.000) with moderate evidence (BF10 = 7.97, posterior probability = 88.9%). Positive FTD accounts for more variance in functioning than negative FTD, but this difference is not robust due to paucity of negative FTD studies. High heterogeneity (I2 > 76%), unexplained by known moderators, indicates substantial variation across study settings. CONCLUSIONS:Both FTD dimensions link to poorer social functioning, with stronger evidence for positive FTD due to more studies and larger samples. Positive FTD may affect social functioning by disrupting communicative competence, highlighting a need for interventions targeting communication deficits.

Changes in spontaneous speech provide an early signal of cognitive dysfunction in Alzheimer's disease (AD) that large language models (LLMs) can detect. However, detection alone cannot establish whether the underlying model representations contribute functionally to behavior. We introduce an activation-guided intervention framework using Qwen3-8B. The framework identifies feed-forward neurons with higher activation rates for AD than control transcripts and modulates their output contributions during generation by scaling the corresponding down-projection weights. This yielded nine edited variants differing in intervention direction, magnitude, and scope. The original and edited models completed the same 12-turn neuropsychological battery, assessed through blinded human ratings and computational linguistic measures. Amplifying AD-associated neurons produced graded impairments in story recall, verbal fluency, working memory, procedural discourse, scene construction, and coreference resolution. Attenuation largely preserved performance and selectively improved several outcomes. Amplification also reduced lexical surprisal, idea density, syntactic complexity, and discourse quantity, broadly paralleling changes reported in human AD speech. These findings show that neurons identified solely from clinical language differences can influence behavior across multiple cognitive domains, providing proof of concept for an AD-related computational phenotype and a controlled framework for experimentally examining links between language and broader cognitive dysfunction.

Positive and negative schizotypy reflect distinct patterns of subclinical traits in the general population associated with neurodevelopmental and schizophrenia-spectrum pathologies. Yet, a comprehensive characterization of the unique and shared neuroanatomical signatures of these schizotypy dimensions is lacking. Leveraging 3D brain MRI data from 2730 unmedicated healthy individuals, we identified neuroanatomical profiles of positive and negative schizotypy and systematically compared them with disorder-specific, microarchitectural, neurotransmitter-level, and connectome measures. Positive and negative schizotypy were associated with distinct cortical signatures, of predominantly thinner frontal and thicker paralimbic cortical areas, respectively. These cortical signatures of positive and negative schizotypy were differentially linked to brain-wide cortical patterns of schizophrenia-spectrum (clinical high-risk for psychosis, schizophrenia) and neurodevelopmental conditions (ADHD, autism spectrum disorder and 22q11.2 deletion syndrome). Additionally, the positive and negative schizotypy-related cortical profiles mapped onto different local attributes of gene expression, cortical myelination, D1, and histamine receptor distributions. Network models further showed that positive and negative schizotypy cortical signatures were spatially associated with cortical hubs, suggesting that highly interconnected regions are more vulnerable to the morphological differences associated with both schizotypy dimensions. Finally, predominantly sensorimotor-to-association and paralimbic areas emerged as epicenters with connectivity profiles significantly linked to the schizotypy-related cortical patterns. Collectively, this study identified cortical signatures of positive and negative schizotypy traits that are embedded along multiple scales of cortical organization and neuropsychiatric pathologies. Our work yields novel insights into how neurobiology and brain architecture may guide neuroanatomical vulnerability and resilience to psychopathology in the general population.

Abstract Disorganized thought is a core feature of the psychosis spectrum. It is recognized, clinically and in daily life, through language, yet its proposed cognitive basis, shallow cognitive maps, the relational structures that organize knowledge, has been tested almost exclusively with task performance and neural measures. Here we show that the organization of a learned conceptual map can be read out from natural speech. Adults varying in schizotypal cognitive disorganization ( n = 107) learned the same two-dimensional conceptual structure from images or sentences, then described their memory strategy and explained the material to a novice. Explaining elicited two-dimensional spatial descriptions selectively after image-based learning. Higher cognitive disorganization was associated with answers that tracked the question asked less closely, on an automated, rater-free embedding measure robust to report language, again selectively after image-based learning. Natural speech thus provides a scalable behavioral readout of conceptual map organization and its disorganization.

Abstract Formal thought disorder (FTD) is a core psychosis feature. Disentangling its dimensions requires tasks simple enough for formal modeling yet sensitive enough to capture individual variation across the psychosis spectrum. The semantic verbal fluency task offers precisely this: a structured behavioral trace of semantic memory sampling, amenable to computational analysis using distributed word embeddings. We hypothesized that this sampling process is governed by two dissociable mechanisms mapping onto FTD dimensions: initial retrieval drive ( d 0 ), quantifying the motivational resource sustaining production, and semantic search precision ( α ), quantifying how strongly similarity to the preceding word constrains each retrieval step from near-random to highly structured. We hypothesized that reduced d 0 would track negative psychosis symptoms and alogia, while degraded α would track language disorganization and left inferior longitudinal fasciculus (ILF) fractional anisotropy. We tested these predictions in a primary ( N = 120) and an independent replication sample ( N = 249) of German-speaking individuals across the psychosis spectrum. Both parameters decreased with greater psychosis severity and, in the primary sample, they dissociated regarding their clinical correlates. d 0 correlated negatively with negative symptoms, general psychopathology, and poverty of speech, consistent with a computational signature of alogia. α correlated negatively with positive symptoms and cognitive flexibility, and, in individuals with psychosis, positively with left ILF fractional anisotropy. The association between d 0 and negative symptoms was replicated in the independent sample. These findings pave the way for mechanistic, automatically derived FTD markers capturing subclinical variation across the psychosis spectrum and mapping onto underlying cognitive and neural processes.

BACKGROUND AND HYPOTHESIS:Some women experience their first episode of psychosis (FEP) in mid-to-later life, including around menopause. Emerging evidence suggests reduced treatment response at this time. However, menopause is rarely considered in treatment guidelines. This study examined whether women experiencing FEP after menopause were prescribed higher olanzapine doses than younger women, using age as a proxy for menopausal status. STUDY DESIGN:A population-based cohort of individuals aged 18-64 presenting with FEP to a Dublin mental health service over 18 years was analyzed. Participants who received an adequate olanzapine trial (≥6 weeks) for FEP were included. Women were stratified by age: <51 years (assumed premenopausal) and ≥51 years (assumed postmenopausal). Mean maximum daily olanzapine doses were compared. Analyses were repeated in men. STUDY RESULTS:Of 396 individuals presenting with FEP, 218 met inclusion criteria. Women aged ≥51 (n = 16) received higher olanzapine doses compared to younger women (n = 78) (15.3 mg ± 6.4 vs 12.4 mg ± 5.4; mean difference 2.9 mg, 95% CI, -0.10 to 5.99, one-sided P = .029, d = 0.53). No age-related dose differences were observed in men (n = 124). CONCLUSIONS:Women presenting with FEP after assumed menopause were prescribed higher olanzapine doses than younger women. This finding should be interpreted cautiously due to the small sample size and inference of menopausal status from age. However, this aligns with emerging evidence suggesting antipsychotic response may reduce following menopause. If replicated, these findings may have important clinical implications for guiding psychosis treatment in mid-life women.

Importance:Dose reduction or discontinuation (DRD) early after remission from first-episode psychosis (FEP) increases short-term relapse risk. Controversy remains regarding potential benefits in functioning over the longer term because studies with long-term outcomes show conflicting findings. Objective:To compare short- and long-term effects between DRD and maintenance medication over a 4-year period in a large sample of patients with FEP. Design, Setting, and Participants:The Handling Antipsychotic Medication Long-Term Evaluation of Targeted Treatment (HAMLETT) study is a single-blind pragmatic randomized (1:1) clinical trial conducted in 26 specialized psychosis units in the Netherlands from September 2017 to March 2023. Patients remitted for FEP from in- and outpatient services were included. Interventions:DRD within 12 months after remission compared with 12 months maintenance treatment. Main Outcomes and Measures:The primary outcome was patient-rated functioning, measured by the World Health Organization Disability Assessment Schedule 2.0 (WHODAS-2). Secondary outcomes were researcher-rated global assessment of functioning (GAF), quality of life, relapse, symptom severity (measured by the Positive and Negative Syndrome Scale [PANSS]), serious adverse events, and adverse effects. Results:A total of 347 patients (241 male [69.5%]; mean [SD] age, 27.9 [8.7] years) were included, with 168 randomized to early DRD and 179 to maintenance. WHODAS-2 showed no time × condition interaction. In the first year, DRD was associated with higher risk of relapse (odds ratio, 2.84; 95% CI, 1.08 to 7.66; P = .04) and lower quality of life (β = -3.31; 95% CI, -6.34 to -0.29; P = .03). At 3 years (β = 3.61; 95% CI, 0.28 to 6.95; P = .03) and 4 years (β = 6.13; 95% CI, 2.03 to 10.22; P = .003), a nonlinear effect of time occurred, showing significantly better GAF for patients in the DRD condition, with a similar trend for PANSS at 4 years (P for trend = .06). Although SAEs and adverse effects were similar between groups, 3 confirmed deaths by suicide occurred in the DRD group, against 1 death by suicide in the maintenance group. Conclusions and Relevance:This randomized clinical trial found that DRD posed risks of relapse and worse quality of life over the first year but yielded better researcher-rated functioning at the third and fourth year, with a similar trend for symptom severity; because antipsychotic medication doses were comparable in the 2 groups from 1 year onwards, this finding is not a direct result of lower medication but may reflect a learning experience to use antipsychotics to better handle psychotic vulnerability. These findings suggest that the potential learning and empowering element of DRD needs to be weighed carefully against short-term risks. Trial registration:EudraCT number: 2017-002406-12.

INTRODUCTION:Prominent sex differences exist in severe mental illness (SMI), with increasing evidence pointing towards a pivotal role for sex hormones. Elucidation of these hormonal influences is crucial to tailor sex-specific prevention and treatment. METHODS:To investigate potential shared genetics and bi-directional causal effects between sex hormone traits and SMI (depressive disorder, bipolar disorder and schizophrenia-spectrum disorder), we computed genetic correlations using linkage disequilibrium score regression and bi-directional summary-level Mendelian Randomization (MR). A range of sensitivity methods was applied and potential mediators were investigated using multivariable MR. Sex-stratified data from genome-wide association studies were used, if available further stratified on menopausal status. We also incorporated other sex hormone traits (progesterone, sex hormone-binding globulin, prolactin, age of menarche, age of menopause) in exploratory analyses. RESULTS:We found a widespread pattern of statistically significant, modest genetic correlations between oestrogen/testosterone levels and depressive disorder/schizophrenia-spectrum disorder, in both positive and negative directions and in both sexes (ranging between -0.22 and 0.13). With MR, evidence for causal effects was largely lacking; apart from weak evidence for a causal, increasing effect of testosterone levels on schizophrenia-spectrum disorder risk in males, which was mediated by CRP. Conversely, there was very weak evidence for a causal, increasing effect of liability to schizophrenia-spectrum disorder on testosterone levels in both sexes. CONCLUSION:This study offers new insights into the complex aetiology of SMI by comprehensively mapping genetic associations with sex hormone traits, emphasizing the need to further investigate sex hormones' impact on SMI using larger and more precisely phenotyped samples to identify individuals particularly vulnerable to hormonal disturbances.

Lesion network mapping (LNM) is a neuroimaging framework that uses normative functional connectivity (FC) data to link heterogeneous brain lesions and functional alterations to brain networks implicated in neurological and psychiatric conditions. However, many of the networks identified by LNM and related methods appear to be highly similar across diverse conditions such as addiction, depression, psychosis and epilepsy. To understand this similarity, we re-examined the data from multiple LNM studies and assessed the methodological roots of the method. Our findings reveal a foundational limitation: at its core, LNM involves a repetitive sampling of one and the same FC matrix. As a result, it systematically maps sets of local brain changes-whether they are patient lesions, magnetic resonance imaging-derived alterations, synthetic or random-onto the same nonspecific properties of the used FC data, producing highly similar networks across conditions. This central limitation cautions the use of LNM as a method for studying distinct biological networks underlying brain disorders. Our work may aid the development of a new generation of network-mapping methods from first principles.

Importance:Revealing neurobiological markers of antipsychotic nonresponse in psychosis may aid outcome prediction and inform novel treatment targets. Objective:To examine differences in neurometabolites in antipsychotic nonresponsive compared to antipsychotic-responsive psychosis using individual participant data and meta-analysis. Data Sources:Web of Science was searched for studies published between January 1, 1980, and November 1, 2025. Authors of 21 eligible studies identified before August 2024 were invited to contribute individual participant data. Study Selection:Eighteen studies examining neurometabolites by treatment response in psychosis contributed individual participant data for the mega-analysis. These studies plus a further 5 studies were included in the meta-analyses of standardized mean differences and variability. Data Extraction and Synthesis:Individual participant data were analyzed using linear mixed models with study as a random effect. Subgroup analyses examined prospective designs and treatment-resistant samples. Published group means and standard deviations were extracted for meta-analyses. Main Outcomes and Measures:Group differences in glutamate, glutamate plus glutamine, choline, myo-inositol, N-acetylaspartate, γ-aminobutyric acid, and glutathione in the medial frontal cortex, dorsolateral prefrontal cortex, thalamus, and basal ganglia. Results:The mega-analysis included 1189 participants from 18 studies; of these, 476 were treatment nonresponders (mean [SD] age, 33.0 [12.5] years; 340 male), 427 were treatment responders (mean [SD] age, 30.3 [11.5] years; 299 male), and 286 were healthy control individuals (mean [SD] age, 31.0 [12.5] years; 170 male). Compared with the antipsychotic response group, nonresponders showed elevations in medial frontal glutamate (Glass Δ = 0.21; P = .02), glutamate plus glutamine (Glass Δ = 0.29; P = .002), choline (Glass Δ = 0.22; P = .03), and myo-inositol (Glass Δ = 0.35; P = .001); similar elevations were observed relative to control individuals. Elevated medial frontal glutamate plus glutamine in antipsychotic nonresponders compared with responders was also observed prospectively in first-episode psychosis (Glass Δ = 0.41; P = .002), whereas myo-inositol elevations were greatest in individuals meeting criteria for treatment-resistance (Glass Δ = 0.64; P = .001). The meta-analysis of 23 studies (1844 participants) also showed elevated medial frontal choline and myo-inositol in antipsychotic nonresponse compared with response. Conclusions and Relevance:These findings provide evidence of an association between antipsychotic nonresponse in psychosis with elevations in medial frontal glutamate, choline, and myo-inositol. The presence of elevations in these markers supports the continued investigation of glutamate-acting and inflammatory pathway-associated interventions for psychosis and schizophrenia.

Background Early intervention in first-episode psychosis (FEP) is critical for long-term outcomes with antipsychotic medicines among the primary treatment options. However, existing clinical practice guidelines (CPGs) do not provide sex-specific recommendations, despite females experiencing distinct vulnerabilities to antipsychotic side-effects. In particular, hyperprolactinemia and cardiometabolic side-effects are associated with substantial subjective distress and potential long-term physical health risks for females across the reproductive lifespan. We aimed therefore to develop a CPG on the preferred antipsychotic medicines for females experiencing FEP.Study Design An international multidisciplinary panel, including experts-by-experience, used the GRADE-ADOLOPMENT process and AGREE II framework to adapt existing FEP guidelines for adults and adolescents. Key health questions were developed through stakeholder consultation and literature review. Critically important patient outcomes were prioritized, and evidence was synthesized on side-effect profiles, with recommendations agreed by consensus. The guideline algorithm was field-tested and externally reviewed by experts.Study Results Prolactin-elevation and cardiometabolic side-effects were prioritized in antipsychotic medicine selection for females. Medicines with higher risks-first-generation antipsychotics, olanzapine, quetiapine, risperidone, paliperidone, and amisulpride-are not recommended first-line. Aripiprazole is recommended as the preferred first-choice due to its consistently favorable prolactin and cardiometabolic profile. Alternative options with low or low-to-medium risk profiles are recommended for adults and adolescents, supported by shared decision-making tools.Conclusions This is the first CPG addressing antipsychotic choice for females with FEP. By prioritizing critically important patient outcomes and lived experience, the guideline supports safer, sex-sensitive prescribing for females that may improve treatment acceptability, adherence, and equity in psychosis care.

Coherence in speech is clinically significant in mental disorders but remains difficult to quantify. We tested the widely-held assumption that semantic similarity metrics derived from large language models capture human-rated coherence. Across three large neurotypical datasets in different languages, semantic similarity failed to correlate with human ratings, while six other metrics, especially the probability-based metrics, showed significant but weak correlations. In an additional English dataset of 94 individuals, including healthy controls and patients with schizophrenia spectrum disorders (SSD), speech coherence was reduced in SSD. Incoherence in the drug-naïve first-episode samples related to altered whole-brain intrinsic functional gradients, and to the probabilistic metric of perplexity in speech. Together, these findings call into question semantic similarity as a proxy for coherence, motivate greater emphasis on probabilistic predictability measures for evaluating coherence, and substantiate the perspective of spontaneous speech as an overt readout of an alteration of hierarchical cortical organization in schizophrenia.