Precision mental health aims to enable personalized care for mental disorders via the identification of brain-behavior associations at the individual level, yet current frameworks often face challenges in generalizability, interpretability, and clinical translation. Contrastive machine learning (CML) has emerged as a promising paradigm for characterizing individual brain variations by extracting brain dimensions that capture both disease-relevant aberrations and inter-subject heterogeneity. In this Review, we synthesize the conceptual foundations and recent methodological advances of CML. We compare CML with related frameworks and illustrate how it integrates with subtyping and predictive modeling pipelines, situating it within the broader landscape of precision mental health. We then review emerging applications that use CML to link brain structure and function to cognition, emotion, and treatment response. Finally, we outline future directions of applications and methodological innovations where CML could further advance personalized diagnosis and intervention in mental health.

Evidence suggests that patients with inflammatory bowel disease (IBD) are at higher risk of psychiatric disorders, yet temporal pattern of risk elevation before/after IBD diagnosis and role of familial factors remain unclear.

Swedish IBD patients diagnosed 2007-2023 were compared with reference individuals in a matched population-based cohort and compared with IBD-free full siblings in a sibling-controlled cohort. Primary outcome was any psychiatric disorder, while secondary outcomes included psychotic disorders, mood disorders, anxiety disorders, eating disorders, substance misuse, personality disorders, attention-deficit/hyperactivity disorder, and autism spectrum disorders. We estimated temporal patterns of hazard ratios (HRs) from five years before to ten years after IBD diagnosis by using flexible parametric survival models.

The study included 43,862 IBD patients (mean age at diagnosis: 41.6 years; 47.3% female) and 178,821 reference individuals. The HR of any psychiatric disorder started to increase from 2-3 years before IBD diagnosis (e.g., HR=1.15 [95%CI: 1.09-1.21] at -2 years), peaked immediately after IBD diagnosis (HR=1.50 [1.41-1.59] at 0.5 year), and decreased rapidly thereafter but remained increased even at 10 years after diagnosis (HR=1.19 [1.13-1.24]). The increased HR was mainly driven by major depressive disorder, anxiety disorders, and substance misuse. No increase was observed for psychotic disorders, personality disorders, and attention-deficit hyperactivity disorder, although IBD patients were more likely to have autism spectrum disorders shortly after diagnosis and eating disorders from ≥5 years onwards. Similar temporal HR patterns were observed across all subtypes of IBD: Crohn's disease, ulcerative colitis, and IBD-unclassified. Similar risk increases for psychiatric disorders were observed in the sibling-controlled cohort.

Patients with IBD were at an increased risk of several psychiatric disorders from 2-3 years before diagnosis through up to 10 years after, and this risk increase cannot be fully explained by shared familial factors.

Personality disorders (PDs) are increasingly recognized and highly prevalent, yet remarkably poorly understood, diagnosed and treated mental conditions. Common PDs include antisocial, avoidant, borderline, dependent, histrionic, narcissistic, obsessive-compulsive, and paranoid disorders. Experimental (animal) models are a valuable tool to study various brain disorders, including PDs. However, characterized by complex and frequently overlapping aberrant behaviors, PDs pose major challenges for their preclinical modeling. Here, we discuss the potential utility of zebrafish (Danio rerio) as complementary translational models relevant to studying PDs. Paralleling clinical symptoms and traits of PDs to specific behavioral phenotypes, biomarkers and paradigms in these fish, we critically evaluate the existing challenges and limitations of such models, and provide a conceptual framework for further research in this field. Despite these conceptual and practical challenges, we argue that zebrafish models not only provide a powerful platform for elucidating neurobiological mechanisms of a wide range of key PD traits (e.g., impulsivity, aggression, emotional dysregulation), but may markedly accelerate both translational modeling of these disorders and therapeutic discovery.

Clinical decision-making in psychiatry has traditionally relied on rating scales and clinical impressions documented in the electronic health record (EHR). Yet, clinical interviews contain rich behavioral signals that remain underutilized in psychiatric care. Recent advances in artificial intelligence (AI) now enable quantification of these signals, and prior work demonstrates that computational measures of speech, language, and facial expression can inform diagnosis and estimate symptom severity. Despite this progress, most prediction efforts remain confined to single modalities and individual diagnoses and focus on diagnostic classification rather than clinically actionable outcomes such as treatment discontinuation or the need for crisis care. Here, we contextualize advances in behavioral quantification and multimodal data fusion, and present the Phenotypes REimagined to Define Clinical Treatment and Outcome Research (PREDiCTOR) study, a prospective cohort study of 2100 patients entering outpatient mental health care. PREDiCTOR is designed to develop and validate dynamic, multimodal prediction signatures that predict treatment discontinuation, emergency department visits, and hospitalizations over a one-year follow-up period. The study audiovisual recordings of clinical encounters, EHR data, cognitive assessments, smartphone passive sensing, therapeutic alliance measures, and audio/text diaries within a Contextual Bandit framework that continuously updates individualized outcome estimates as new data become available. Both interpretable features and learned embeddings are leveraged, with large language models serving as feature extractors rather than clinical decision-makers. We describe the study design, data collection, and processing pipelines, hybrid predictive modeling approach, and prospective validation strategy, and discuss the potential for translating dynamic behavioral quantification into individualized clinical prognostics.

Aberrations in neuro-immune, metabolic, and oxidative stress (NIMETOX) pathways are implicated in major depressive disorder (MDD). First-episode simple dysmood disorder (FE-SDMD) without metabolic syndrome offers a unique model to investigate early lipid alterations using a systems biology approach.

Plasma samples were collected from 88 university students (44 FE-SDMD, 44 healthy controls). Participants underwent psychiatric assessments, including adverse childhood experiences (ACEs), negative life events (NLEs), depression, anxiety, suicidal behaviors, and insomnia. Untargeted lipid profiling was performed using LC-QTOF-MS, while indices of oxidative and nitrosative stress (ONS) and lecithin-cholesterol acyltransferase (LCAT) activity were assessed. Data was analyzed using machine learning approaches with recursive feature elimination and cross-validation.

FE-SDMD was characterized by increased ceramides (CER), diacylglycerides (DAG), triacylglycerides (TG), sphingomyelins (SM), bis-monoacylglycerol phosphates (BMP), cholestone, and fatty-acyl amino acids (FAAA). DAG, CER, and BMP were the strongest predictors of depression severity and physiosomatic symptoms, whereas cholestone, CER, and SM predicted suicidal behaviors. These lipid modules, together with lowered LCAT and increased ONS, explained substantial variance in depression severity (46.4%), physiosomatic symptoms (42.4%), cognitive-affective symptoms (37.9%), suicidal behaviors (30.1%), insomnia (32%), and anxiety (19.5%). ACEs and NLEs were strongly associated with CER (p < 0.001), DAG (p < 0.01), and cholestone (p < 0.01), revealing coordinated lipidomic networks within a systems biology framework.

Early-stage MDD is characterized by lipid dysregulations, which are linked to psychosocial stress exposure, oxidative stress and attenuated reverse cholesterol transport. These lipid modules identify biologically functional pathways associated with early disease and may inform the development of future therapeutic strategies.

Depressive and anxiety symptoms frequently persist during stable phases of bipolar disorder (BD), yet their interconnected architecture remains unclear. This study examined the prevalence, clinical correlates, and network structure of these symptoms in patients with BD.

We evaluated 9455 clinically stable patients with BD in Hangzhou, China. Symptoms were assessed using the Patient Health Questionnaire-9 and Generalized Anxiety Disorder-7. Multivariable logistic regression identified correlates, and a regularized Gaussian Graphical Model mapped symptom interactions, centralities, and bridge symptoms.

Screen-positive rates of depressive symptoms, anxiety symptoms, and their co-occurrence were 18.3%, 15.1%, and 12.1%, respectively, with a strong correlation between the two domains (r = 0.71, p < 0.01). Poor social relationships and sleep disturbances were consistently associated with higher symptom burden (ORs: 1.37-2.65). Network analysis revealed a prominent "concentration"-"motor" association (weight = 0.38). Centrality analysis identified "guilty," "uncontrollable worry," and "motor" as the most influential nodes. "Motor," "irritability," and "feeling afraid" served as critical bridge symptoms linking both domains. In the flow network, "suicidal ideation" was directly anchored by "guilty" (weight = 0.21) and "motor" symptoms (weight = 0.20). Three distinct symptom communities were identified, reflecting anxiety, cognitive-behavioral, and somatic-affective clusters.

Even during stable periods of BD, residual depressive and anxiety symptoms form a highly structured network. These findings suggest that clinical management should transcend aggregate scale scores to target key central and bridge features, specifically psychomotor alterations, guilt, and uncontrollable worry, to disrupt symptom reinforcement and mitigate suicide risk.

Cognitive symptoms represent a significant and persistent challenge in patients with treatment-resistant depression (TRD), substantially impacting their functional outcomes. Deep brain stimulation (DBS) targeting the subcallosal cingulate gyrus (SCG) has shown clinical efficacy. Nonetheless, its long-term effects on cognition remains insufficiently researched. This study systematically evaluates cognitive outcomes following SCG-DBS over an extended follow-up period.

This retrospective, exploratory secondary analysis used longitudinal data from 12 TRD patients treated with SCG-DBS without a control group. Cognitive and clinical assessments were conducted pre-surgery and at long-term follow-up. Potential changes were examined using the Wilcoxon signed-rank test and its corresponding effect size (r), while Reliable Change Indices (RCIs) were applied to evaluate the statistical reliability of intra-subject cognitive changes. Spearman correlations explored associations between cognitive outcomes with clinical and functional ones.

After a median follow-up of 5 years following SCG-DBS, TRD patients showed statistically significant improvements in immediate and delayed verbal memory (Wilcoxon's z = -2.28, p = .023, r = 0.66; Wilcoxon's z = -2.83, p = .005, r = 0.82, respectively), both reflecting large effect sizes. RCIs indicated reliable improvement in immediate (41.7% of patients) and in delayed (75%) verbal memory. Changes in verbal memory were not significantly correlated with clinical or functional improvements.

These findings support the cognitive safety and benefits of SCG-DBS in individuals with TRD, particularly in verbal memory. However, further research is required to investigate the broader functional effects of SCG-DBS beyond the improvement of core clinical symptoms.

Patient-reported outcomes are essential for monitoring recovery after traumatic brain injury (TBI), yet commonly used instruments differ in scope. The Rivermead Post-Concussion Symptoms Questionnaire (RPQ) is symptom-specific, whereas Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health is a generic measure of physical and mental health. The degree to which these tools provide overlapping versus non-interchangeable information in outpatient TBI care is unclear.

We conducted a retrospective observational study at a Level I academic trauma center (June 2018-February 2025). Adults (≥18 years) evaluated for TBI who completed paired RPQ and PROMIS Global Health assessments within ±3 days were included (one paired assessment per patient). Clinical and injury variables were abstracted from the electronic medical record and REDCap; head CT was classified as positive versus negative based on neuroradiology report, and polytrauma was defined as non-neurosurgical injuries requiring additional service management. Associations between PROMIS domains (physical, mental, total; higher scores indicating better health) and RPQ burden (RPQ-3, RPQ-13, total; higher scores indicating greater symptom burden) were assessed using Spearman correlation. Agreement was evaluated using two-way random-effects intraclass correlation coefficients (ICC; absolute agreement), Bland-Altman analysis after linear transformation to a 0-100 scale with standardized directionality, and kappa for selected construct-matched item pairs.

Of 372 screened patients, 216 met inclusion criteria. Median age was 58 years (IQR 39-73); 67% were male; 84% had mild TBI by admission Glasgow Coma Scale; 51% had polytrauma. PROMIS domains showed consistent inverse correlations with RPQ burden (ρ -0.41 to -0.68; all p < 0.001), strongest for PROMIS total versus RPQ-13 (ρ = -0.68) and RPQ total (ρ = -0.67). Despite these associations, patient-level agreement was poor: ICCs were negative across PROMIS-RPQ pairings (ICC -0.04 to -0.49) and Bland-Altman limits of agreement were wide. Construct-matched item agreement was moderate to substantial (κ 0.46-0.64), highest for combined affect ("angry and sad") (κ = 0.64).

PROMIS and RPQ are associated at the group level but demonstrate poor agreement and should not be used interchangeably for individual outpatient TBI monitoring. Findings support combined use of global and symptom-specific instruments across care settings.

Background: Military sexual trauma (MST) disproportionately impacts women veterans globally. UK research highlights that these in-service experiences may contribute to unique mental and physical health needs. Yet, service providers are often perceived to be unaware of and unable to meet these needs.Objective: To evaluate a national training programme that synthesised evidence to support UK women veterans. Situated within a wider knowledge mobilization (KMb) initiative and guided by the Knowledge-to-Action (KTA) framework, a three-part programme was developed for different professional audiences to enhance understanding of and support for women veterans.Method: Co-produced and grounded in KMb theory. Course 1 aimed to enhance knowledge of women veterans' health needs. Course 2 targeted healthcare professionals and veteran specialist providers, exploring evidence-based considerations for promoting service engagement. Course 3 aimed to increase UK clinical capacity in the best-evidenced treatment for post-traumatic stress disorder associated with MST in women veterans, through training mental health professionals in cognitive processing therapy (CPT). Mixed-methods data were collected from all programme attendees. CPT attendees completed pre- and 3-month post-training measures; quantitative data were analysed using a Wilcoxon signed-rank test and qualitative data using manifest content analysis.Results: Course 1 was completed by 526 individuals, 363 completed Course 2, and 98 completed Course 3. Course 3 attendees were split across two cohorts (n = 57; n = 41). Compared with pre-training, 3-month post-training self-efficacy in delivering CPT across cohort one increased significantly (p < .001, r = 0.915). Content analysis identified three themes: (1) Clinician confidence delivering CPT; (2) Clinicians' perceptions of CPT; and (3) CPT in a service system.Conclusions: The education programme addressed gaps in UK service provider knowledge and increased CPT clinical capacity. The wider KMb initiative demonstrates how research can be translated into impact for military populations and highlights the value of co-produced KMb approaches.

Eating attitude can be defined as beliefs, thoughts, feelings, behaviour and relationship with food. It can impact a person's food choices and health status. A deviant eating attitude is a maladaptive pattern of thoughts and behaviours related to food consumption, which can include periods of inconsistent eating, ranging from overeating to undereating, dietary restriction, emotional eating and preoccupation with food and body weight. These concerns may be visible in the form of emotional dysregulation, anxiety, depression, low self-esteem and interpersonal difficulties. The deviant eating attitude can lead to psychological distress, as individuals may use restrictive eating patterns to cope with overwhelming emotions and life stresses, further deteriorating their overall mental health. Dialectical behaviour therapy (DBT) plays a key role in addressing this vicious cycle by targeting the core emotional and behavioural patterns which maintain disturbed eating disorder. DBT reduces the shame and self-stigma that affect the eating cycle and fosters genuine engagement with the therapeutic process. Through DBT, an individual can learn about mindfulness, distress tolerance, emotion regulation and interpersonal effectiveness. It facilitates individuals to recognise and manage uncomfortable emotions without resorting to dysfunctional eating patterns.

The study aimed to explore the therapeutic effect of DBT on deviant eating attitude, along with mental health in adolescents.

A single case report method was used in the current study. The tools used in the study were case history, Mental Status Examination, Eating Attitudes Test-26, developed by David M. Garner, and Mental Health Inventory, developed by A. K. Srivastava. A total number of 14 sessions of DBT were given, and the pre- and post-therapy assessments were conducted. A case of a 15-year-old female presented with the chief complaints of concerns with her weight and physical looks, high consumption of junk food, disturbed sleep routine, disturbed interpersonal relationship, anger outburst, low self-esteem, irritability and reduced academic performance since the last 2 years, which increased significantly in the last 6 months.

The pre- and post-assessments were conducted, and they indicated relief in the symptoms reported by the client. She reported improved focus and increased motivation for academic goals. She also reported improved social relationships with improved mood and regulated sleep than before. She reported a reduction in the consumption of junk food. Her interpersonal skills improved significantly, allowing her to express her thoughts and feelings more effectively.

Findings support DBT as an effective mode of intervention that targets deviant eating attitudes and improves mental health by strengthening emotion regulation, increasing distress tolerance, and improves interpersonal skills.