Accumulating data show that probiotics may be beneficial for reducing depressive, anxiety, and stress symptoms. However, the best combinations and species of probiotics have not been identified. The objective of our study was to assess the most effective combinations and components of different probiotics through network meta-analysis.

A systematic search of four databases, PubMed, Web of Science, Cochrane, and Embase, was conducted from inception to 11 January 2024. The GRADE framework was used to assess the quality of evidence contributing to each network estimate.

We deemed 45 trials eligible, these included 4053 participants and 10 types of interventions. The quality of evidence was rated as high or moderate. The NMA revealed that Bifidobacterium exhibited a greater probability of being the optimal probiotic species for improving anxiety symptoms (SMD = −0.80; 95% CI −1.49 to −0.11), followed by Lactobacillus (SMD = −0.49; 95% CI −0.85 to −0.12). In addition, for multiple strains, compared with the other interventions, Lactobacillus + Bifidobacterium (SMD = −0.41; 95% CI −0.73 to −0.10) had a positive effect on depression.

The NMA revealed that Lactobacillus and Bifidobacterium had prominent efficacy in the treatment of individuals with anxiety, depression, and combination of Lactobacillus + Bifidobacterium had a similar effect. With few direct comparisons available between probiotic species, this NMA may be instrumental in shaping the guidelines for probiotic treatment of psychological disorders.

Studies investigating parenthood and how it affects long-term outcomes are lacking among individuals with schizophrenia spectrum disorders. This study aimed to examine the life of participants 20 years after their first diagnosis with a special focus on parenthood, clinical illness course, and family-related outcomes.

Among 578 individuals diagnosed with first-episode schizophrenia spectrum disorder between 1998 and 2000, a sample of 174 participants was reassessed at the 20-year follow-up. We compared symptom severity, remission, clinical recovery, and global functioning between 75 parents and 99 non-parents. Also, family functioning scored on the family assessment device, and the children's mental health was reported. We collected longitudinal data on psychiatric admission, supported housing, and work status via the Danish registers.

Participants with offspring had significantly lower psychotic (mean (s.d.) of 0.89 (1.46) v. 1.37 (1.44), p = 0.031) negative (mean [s.d.] of 1.13 [1.16] v. 1.91 [1.07], p < 0.001) and disorganized symptom scores (mean [s.d.] of 0.46 [0.80] v. 0.85 [0.95], p = 0.005) and more were in remission (59.5% v. 22.4%, p < 0.001) and in clinical recovery (29.7% v. 11.1%, p = 0.002) compared to non-parents. When investigating global functioning over 20 years, individuals becoming parents after their first diagnosis scored higher than individuals becoming parents before their first diagnosis and non-parents. Regarding family-related outcomes, 28.6% reported unhealthy family functioning, and 10% of the children experienced daily life difficulties.

Overall, parents have more favorable long-term outcomes than non-parents. Still, parents experience possible challenges regarding family functioning, and a minority of their children face difficulties in daily life.

Wellbeing is relatively stable over the life span. However, individuals differ in this stability and change. One explanation for these differences could be the influence of different genetic or environmental factors on wellbeing over time.

To investigate causes of stability and change of wellbeing across the lifespan, we used cohort-sequential data on wellbeing from twins and their siblings of the Netherlands Twin Register (NTR) (total N = 46.885, 56% females). We organized wellbeing data in multiple age groups, from childhood (age 5), to adolescence, up to old age (age 61+). Applying a longitudinal genetic simplex model, we investigated the phenotypic stability of wellbeing and continuity and change in genetic and environmental influences.

Wellbeing peaked in childhood, decreased during adolescence, and stabilized during adulthood. In childhood and adolescence, around 40% of the individual differences was explained by genetic effects. The heritability decreased toward old adulthood (35–24%) and the contribution of unique environmental effects increased to 76%. Environmental innovation was found at every age, whereas genetic innovation was only observed during adolescence (10–18 years). In childhood and adulthood, the absence of genetic innovation indicates a stable underlying set of genes influencing wellbeing during these life phases.

These findings provide insights into the stability and change of wellbeing and the genetic and environmental influences across the lifespan. Genetic effects were mostly stable, except in adolescence, whereas the environmental innovation at every age suggests that changing environmental factors are a source of changes in individual differences in wellbeing over time.

Epigenetic modifications, such as DNA methylation, contribute to the pathophysiology of major depressive disorder (MDD). This study aimed to identify novel MDD-associated epigenetic loci using DNA methylation profiles and explore the correlations between epigenetic loci and cortical thickness changes in patients with MDD.

A total of 350 patients with MDD and 161 healthy controls (HCs) were included in the epigenome-wide association studies (EWAS). We analyzed methylation, copy number alteration (CNA), and gene network profiles in the MDD group. A total of 234 patients with MDD and 135 HCs were included in neuroimaging methylation analysis. Pearson's partial correlation analysis was used to estimate the correlation between cortical thickness of brain regions and DNA methylation levels of the loci.

In total, 2018 differentially methylated probes (DMPs) and 351 differentially methylated regions (DMRs) were identified. DMP-related genes were enriched in two networks involved in the central nervous system. In neuroimaging analysis, patients with MDD showed cortical thinning in the prefrontal regions and cortical thickening in several occipital regions. Cortical thickness of the left ventrolateral prefrontal cortex (VLPFC, i.e. pars triangularis) was negatively correlated with eight DMPs associated with six genes (EML6, ZFP64, CLSTN3, KCNMA1, TAOK2, and NT5E).

Through combining DNA methylation and neuroimaging analyses, negative correlations were identified between the cortical thickness of the left VLPFC and DNA methylation levels of eight DMPs. Our findings could improve our understanding of the pathophysiology of MDD.

Several factors shape the neurodevelopmental trajectory. A key area of focus in neurodevelopmental research is to estimate the factors that have maximal influence on the brain and can tip the balance from typical to atypical development.

Utilizing a dissimilarity maximization algorithm on the dynamic mode decomposition (DMD) of the resting state functional MRI data, we classified subjects from the cVEDA neurodevelopmental cohort (n = 987, aged 6–23 years) into homogeneously patterned DMD (representing typical development in 809 subjects) and heterogeneously patterned DMD (indicative of atypical development in 178 subjects).

Significant DMD differences were primarily identified in the default mode network (DMN) regions across these groups (p < 0.05, Bonferroni corrected). While the groups were comparable in cognitive performance, the atypical group had more frequent exposure to adversities and faced higher abuses (p < 0.05, Bonferroni corrected). Upon evaluating brain-behavior correlations, we found that correlation patterns between adversity and DMN dynamic modes exhibited age-dependent variations for atypical subjects, hinting at differential utilization of the DMN due to chronic adversities.

Adversities (particularly abuse) maximally influence the DMN during neurodevelopment and lead to the failure in the development of a coherent DMN system. While DMN's integrity is preserved in typical development, the age-dependent variability in atypically developing individuals is contrasting. The flexibility of DMN might be a compensatory mechanism to protect an individual in an abusive environment. However, such adaptability might deprive the neural system of the faculties of normal functioning and may incur long-term effects on the psyche.

Cardiometabolic disease risk factors are disproportionately prevalent in bipolar disorder (BD) and are associated with cognitive impairment. It is, however, unknown which health risk factors for cardiometabolic disease are relevant to cognition in BD. This study aimed to identify the cardiometabolic disease risk factors that are the most important correlates of cognitive impairment in BD; and to examine whether the nature of the relationships vary between mid and later life.

Data from the UK Biobank were available for 966 participants with BD, aged between 40 and 69 years. Individual cardiometabolic disease risk factors were initially regressed onto a global cognition score in separate models for the following risk factor domains; (1) health risk behaviors (physical activity, sedentary behavior, smoking, and sleep) and (2) physiological risk factors, stratified into (2a) anthropometric and clinical risk (handgrip strength, body composition, and blood pressure), and (2b) cardiometabolic disease risk biomarkers (CRP, lipid profile, and HbA1c). A final combined multivariate regression model for global cognition was then fitted, including only the predictor variables that were significantly associated with cognition in the previous models.

In the final combined model, lower mentally active and higher passive sedentary behavior, higher levels of physical activity, inadequate sleep duration, higher systolic and lower diastolic blood pressure, and lower handgrip strength were associated with worse global cognition.

Health risk behaviors, as well as blood pressure and muscular strength, are associated with cognitive function in BD, whereas other traditional physiological cardiometabolic disease risk factors are not.

Suicidal thoughts and behaviors are elevated among active-duty service members (ADSM) and veterans compared to the general population. Hence, it is a priority to examine maintenance factors underlying suicidal ideation among ADSM and veterans to develop effective, targeted interventions. In particular, interpersonal risk factors, hopelessness, and overarousal have been robustly connected to suicidal ideation and intent.

To identify the suicidal ideation risk factors that are most relevant, we employed network analysis to examine between-subjects (cross-sectional), contemporaneous (within seconds), and temporal (across four hours) group-level networks of suicidal ideation and related risk factors in a sample of ADSM and veterans (participant n = 92, observations n = 10 650). Participants completed ecological momentary assessment (EMA) surveys four times a day for 30 days, where they answered questions related to suicidal ideation, interpersonal risk factors, hopelessness, and overarousal.

The between-subjects and contemporaneous networks identified agitation, not feeling close to others, and ineffectiveness as the most central symptoms. The temporal network revealed that feeling ineffective was most likely to influence other symptoms in the network over time.

Our findings suggest that ineffectiveness, low belongingness, and agitation are important drivers of moment-to-moment and longitudinal relations between risk factors for suicidal ideation in ADSM and veterans. Targeting these symptoms may disrupt suicidal ideation.

Evidence suggests that cannabis may be a causal factor for development of schizophrenia. We aimed to investigate whether use of antipsychotic medication, benzodiazepines, and psychiatric service use differs among patients with schizophrenia depending on whether psychosis was precipitated by a diagnosis of cannabis use disorder (CUD).

We utilized the nationwide Danish registries to identify all individuals with an incident diagnosis of schizophrenia from 1995 to 2016. We also collected information on whether first CUD diagnosis preceded schizophrenia and thus defined a group of potentially cannabis-related schizophrenia. We compared the cannabis-related schizophrenia group both with all non-cannabis-related patients with schizophrenia and with non-cannabis-related patients with schizophrenia that were propensity-score matched to cases using a range of potentially confounding variables.

We included 35 714 people with incident schizophrenia, including 4116 (11.5%) that were cannabis-related. In the unmatched-comparison analyses, there were no clear differences over time in use of antipsychotics and benzodiazepines related to whether the diagnosis of schizophrenia was cannabis-related. After propensity-score matching, use of antipsychotics and benzodiazepines was significantly lower among cannabis-related cases of schizophrenia. In the unmatched comparison, the cannabis-related group had significantly more days admitted than the non-cannabis-related group. This was markedly attenuated after propensity-score matching.

Our findings indicate the importance of considering cannabis-related cases of schizophrenia as a potentially distinct disorder in terms of prognosis. It is unclear, however, if these differences are due to different biological types of schizophrenia being compared or if they rather indicate behavioral differences such as reduced adherence and treatment-seeking.

Although the link between alcohol involvement and behavioral phenotypes (e.g. impulsivity, negative affect, executive function [EF]) is well-established, the directionality of these associations, specificity to stages of alcohol involvement, and extent of shared genetic liability remain unclear. We estimate longitudinal associations between transitions among alcohol milestones, behavioral phenotypes, and indices of genetic risk.

Data came from the Collaborative Study on the Genetics of Alcoholism (n = 3681; ages 11–36). Alcohol transitions (first: drink, intoxication, alcohol use disorder [AUD] symptom, AUD diagnosis), internalizing, and externalizing phenotypes came from the Semi-Structured Assessment for the Genetics of Alcoholism. EF was measured with the Tower of London and Visual Span Tasks. Polygenic scores (PGS) were computed for alcohol-related and behavioral phenotypes. Cox models estimated associations among PGS, behavior, and alcohol milestones.

Externalizing phenotypes (e.g. conduct disorder symptoms) were associated with future initiation and drinking problems (hazard ratio (HR)⩾1.16). Internalizing (e.g. social anxiety) was associated with hazards for progression from first drink to severe AUD (HR⩾1.55). Initiation and AUD were associated with increased hazards for later depressive symptoms and suicidal ideation (HR⩾1.38), and initiation was associated with increased hazards for future conduct symptoms (HR = 1.60). EF was not associated with alcohol transitions. Drinks per week PGS was linked with increased hazards for alcohol transitions (HR⩾1.06). Problematic alcohol use PGS increased hazards for suicidal ideation (HR = 1.20).

Behavioral markers of addiction vulnerability precede and follow alcohol transitions, highlighting dynamic, bidirectional relationships between behavior and emerging addiction.

Auditory system plasticity is a promising target for neuromodulation, cognitive rehabilitation and therapeutic development in schizophrenia (SZ). Auditory-based targeted cognitive training (TCT) is a ‘bottom up’ intervention designed to enhance the speed and accuracy of auditory information processing, which has been shown to improve neurocognition in certain SZ patients. However, the dynamics of TCT learning as a function of training exercises and their impact on neurocognitive functioning and therapeutic outcomes are unknown.

Forty subjects (SZ, n = 21; healthy subjects (HS), n = 19) underwent comprehensive clinical, cognitive, and auditory assessments, including measurements of auditory processing speed (APS) at baseline and after 1-h of TCT. SZ patients additionally completed 30-hours of TCT and repeated assessments ~10–12 weeks later.

SZ patients were deficient in APS at baseline (d = 0.96, p < 0.005) relative to HS. After 1-h of TCT, analyses revealed significant main effects of diagnosis (d = 1.75, p = 0.002) and time (d = 1.04, p < 0.001), and a diagnosis × time interaction (d = 0.85, p = 0.02) on APS. APS learning effects were robust after 1-h in SZ patients (d = 1.47, p < 0.001) and persisted throughout the 30-h of training. Baseline APS was associated with verbal learning gains after 30-h of TCT (r = 0.51, p = 0.02) in SZ.

TCT learning metrics may have prognostic utility and aid in the prospective identification of individuals likely to benefit from TCT. Future experimental medicine studies may advance predictive algorithms that enhance TCT-related clinical, cognitive and functional outcomes.

Youth adversity is associated with persistence of depression and anxiety symptoms. This association may be greater for disadvantaged societal groups (such as females) compared with advantaged groups (e.g. males). Given that persistent symptoms are observed across a range of disadvantaged, minoritized, and neurodivergent groups (e.g. low compared with high socio-economic status [SES]), the intersection of individual characteristics may be an important moderator of inequality.

Data from HeadStart Cornwall (N = 4441) was used to assess the effect of youth adversity on combined symptoms of depression and anxiety (Strengths and Difficulties Questionnaire emotional problems subscale) measured at three time-points in 11–14-year-olds. Latent trajectories and regressions were estimated for eight intersectionality profiles (based on gender, SES, and hyperactivity/inattention), and moderating effects of the individual characteristics and their intersections were estimated.

Youth adversity was associated with higher average depression/anxiety symptoms at baseline (11–12-years) across all intersectionality profiles. The magnitude of effects differed across profiles, with suggestive evidence for a moderating effect of youth adversity on change over time in depression/anxiety symptoms attributable to the intersection between (i) gender and SES; and (ii) gender, SES, and hyperactivity/inattention.

The detrimental effects of youth adversity pervade across intersectionality profiles. The extent to which these effects are moderated by intersectionality is discussed in terms of operational factors. The current results provide a platform for further research, which is needed to determine the importance of intersectionality as a moderator of youth adversity on the development of depression and anxiety symptoms in adolescence.

Extensive research has explored altered structural and functional networks in major depressive disorder (MDD). However, studies examining the relationships between structure and function yielded heterogeneous and inconclusive results. Recent work has suggested that the structure-function relationship is not uniform throughout the brain but varies across different levels of functional hierarchy. This study aims to investigate changes in structure-function couplings (SFC) and their relevance to antidepressant response in MDD from a functional hierarchical perspective.

We compared regional SFC between individuals with MDD (n = 258) and healthy controls (HC, n = 99) using resting-state functional magnetic resonance imaging and diffusion tensor imaging. We also compared antidepressant non-responders (n = 55) and responders (n = 68, defined by a reduction in depressive severity of >50%). To evaluate variations in altered and response-associated SFC across the functional hierarchy, we ranked significantly different regions by their principal gradient values and assessed patterns of increase or decrease along the gradient axis. The principal gradient value, calculated from 219 healthy individuals in the Human Connectome Project, represents a region's position along the principal gradient axis.

Compared to HC, MDD patients exhibited increased SFC in unimodal regions (lower principal gradient) and decreased SFC in transmodal regions (higher principal gradient) (p < 0.001). Responders primarily had higher SFC in unimodal regions and lower SFC in attentional networks (median principal gradient) (p < 0.001).

Our findings reveal opposing SFC alterations in low-level unimodal and high-level transmodal networks, underscoring spatial variability in MDD pathology. Moreover, hierarchy-specific antidepressant effects provide valuable insights into predicting treatment outcomes.

Eating disorder (ED) research has embraced a network perspective of psychopathology, which proposes that psychiatric disorders can be conceptualized as a complex system of interacting symptoms. However, existing intervention studies using the network perspective have failed to find that symptom reductions coincide with reductions in strength of associations among these symptoms. We propose that this may reflect failure of alignment between network theory and study design and analysis. We offer hypotheses for specific symptom associations expected to be disrupted by an app-based intervention, and test sensitivity of a range of statistical metrics for identifying this intervention-induced disruption.

Data were analyzed from individuals with recurrent binge eating who participated in a randomized controlled trial of a cognitive-behavioral smartphone application. Participants were categorized into one of three groups: waitlist (n = 155), intervention responder (n = 49), and intervention non-responder (n = 77). Several statistical tests (bivariate associations, network-derived strength statistics, network invariance tests) were compared in ability to identify change in network structure.

Hypothesized disruption to specific symptom associations was observed through change in bivariate correlations from baseline to post-intervention among the responder group but were not evident from symptom and whole-of-network based network analysis statistics. Effects were masked when the intervention group was assessed together, ignoring heterogeneity in treatment responsiveness.

Findings are consistent with our contention that study design and analytic approach influence the ability to test network theory predictions with fidelity. We conclude by offering key recommendations for future network theory-driven interventional studies.

Many autistic people in mental health are suicidal. This study evaluated the effectiveness of dialectical behavior therapy (DBT) v. treatment as usual (TAU) in reducing suicidal ideation and suicide attempts.

At six Dutch mental health centers, 123 outpatients (18–65 years) with DSM-5 diagnosed autism spectrum disorder (ASD) and suicidal behavior were randomly assigned to the DBT intervention group (n = 63) or TAU control group (n = 60). Assessments were conducted at baseline, post-treatment at 6 months and 12-month follow-up. The primary outcomes were severity of suicidal ideation and frequency of suicide attempts. The severity of depression and social anxiety were secondary outcomes.

At end-of-treatment, DBT significantly reduced both suicidal ideation (z = −2.24; p = 0.025; b = −4.41; s.e. = 197.0) and suicide attempts (z = −3.15; p = 0.002; IRR = 0.046; s.e. = 0.045) compared to TAU, but lost statistical significance at the 12-month follow-up. Depression severity significantly decreased with DBT (z = −1.99; p = 0.046: b = −2.74; s.e. = 1.37) remaining so at 12 months (z = −2.46; p = 0.014; b = −3.37; s.e. = 1.37). No effects were observed on social anxiety. Severe adverse events included two suicides in the TAU condition.

DBT is an acceptable, safe, and short-term effective intervention to reduce suicidal ideation and suicide attempts in autistic adults with suicidal behavior.

Selective serotonin reuptake inhibitors (SSRIs) are first-line pharmacological treatments for depression and anxiety. However, little is known about how pharmacological action is related to cognitive and affective processes. Here, we examine whether specific reinforcement learning processes mediate the treatment effects of SSRIs.

The PANDA trial was a multicentre, double-blind, randomized clinical trial in UK primary care comparing the SSRI sertraline with placebo for depression and anxiety. Participants (N = 655) performed an affective Go/NoGo task three times during the trial and computational models were used to infer reinforcement learning processes.

There was poor task performance: only 54% of the task runs were informative, with more informative task runs in the placebo than in the active group. There was no evidence for the preregistered hypothesis that Pavlovian inhibition was affected by sertraline. Exploratory analyses revealed that in the sertraline group, early increases in Pavlovian inhibition were associated with improvements in depression after 12 weeks. Furthermore, sertraline increased how fast participants learned from losses and faster learning from losses was associated with more severe generalized anxiety symptoms.

The study findings indicate a relationship between aversive reinforcement learning mechanisms and aspects of depression, anxiety, and SSRI treatment, but these relationships did not align with the initial hypotheses. Poor task performance limits the interpretability and likely generalizability of the findings, and highlights the critical importance of developing acceptable and reliable tasks for use in clinical studies.

This article presents research supported by NIHR Program Grants for Applied Research (RP-PG-0610-10048), the NIHR BRC, and UCL, with additional support from IMPRS COMP2PSYCH (JM, QH) and a Wellcome Trust grant (QH).

The DSM Level 1 Cross-Cutting Symptom Measure (DSM-XC) allows for assessing multiple psychopathological domains. However, its capability to screen for mental disorders in a population-based sample and the impact of adverbial framings (intensity and frequency) on its performance are unknown.

The study was based on cross-sectional data from the 1993 Pelotas birth cohort in Brazil. Participants with completed DSM-XC and structured diagnostic interviews (n = 3578, aged 22, 53.6% females) were included. Sensitivity, specificity, positive (LR+), and negative (LR−) likelihood ratios for each of the 13 DSM-XC domains were estimated for detecting five internalizing disorders (bipolar, generalized anxiety, major depressive, post-traumatic stress, and social anxiety disorders) and three externalizing disorders (antisocial personality, attention-deficit/hyperactivity, and alcohol use disorders). Sensitivities and specificities >0.75, LR+ > 2 and LR− < 0.5 were considered meaningful. Values were calculated for the DSM-XC's original scoring and for adverbial framings.

Several DSM-XC domains demonstrated meaningful screening properties. The anxiety domain exhibited acceptable sensitivity and LR− values for all internalizing disorders. The suicidal ideation, psychosis, memory, repetitive thoughts and behaviors, and dissociation domains displayed acceptable specificity for all disorders. Domains also yielded small but meaningful LR+ values for internalizing disorders. However, LR+ and LR− values were not generally meaningful for externalizing disorders. Frequency-framed questions improved screening properties.

The DSM-XC domains showed transdiagnostic screening properties, providing small but meaningful changes in the likelihood of internalizing disorders in the community, which can be improved by asking frequency of symptoms compared to intensity. The DSM-XC is currently lacking meaningful domains for externalizing disorders.

Although behavioral mechanisms in the association among depression, anxiety, and cancer are plausible, few studies have empirically studied mediation by health behaviors. We aimed to examine the mediating role of several health behaviors in the associations among depression, anxiety, and the incidence of various cancer types (overall, breast, prostate, lung, colorectal, smoking-related, and alcohol-related cancers).

Two-stage individual participant data meta-analyses were performed based on 18 cohorts within the Psychosocial Factors and Cancer Incidence consortium that had a measure of depression or anxiety (N = 319 613, cancer incidence = 25 803). Health behaviors included smoking, physical inactivity, alcohol use, body mass index (BMI), sedentary behavior, and sleep duration and quality. In stage one, path-specific regression estimates were obtained in each cohort. In stage two, cohort-specific estimates were pooled using random-effects multivariate meta-analysis, and natural indirect effects (i.e. mediating effects) were calculated as hazard ratios (HRs).

Smoking (HRs range 1.04–1.10) and physical inactivity (HRs range 1.01–1.02) significantly mediated the associations among depression, anxiety, and lung cancer. Smoking was also a mediator for smoking-related cancers (HRs range 1.03–1.06). There was mediation by health behaviors, especially smoking, physical inactivity, alcohol use, and a higher BMI, in the associations among depression, anxiety, and overall cancer or other types of cancer, but effects were small (HRs generally below 1.01).

Smoking constitutes a mediating pathway linking depression and anxiety to lung cancer and smoking-related cancers. Our findings underline the importance of smoking cessation interventions for persons with depression or anxiety.

Machine learning could predict binge behavior and help develop treatments for bulimia nervosa (BN) and alcohol use disorder (AUD). Therefore, this study evaluates person-specific and pooled prediction models for binge eating (BE), alcohol use, and binge drinking (BD) in daily life, and identifies the most important predictors.

A total of 120 patients (BN: 50; AUD: 51; BN/AUD: 19) participated in an experience sampling study, where over a period of 12 months they reported on their eating and drinking behaviors as well as on several other emotional, behavioral, and contextual factors in daily life. The study had a burst-measurement design, where assessments occurred eight times a day on Thursdays, Fridays, and Saturdays in seven bursts of three weeks. Afterwards, person-specific and pooled models were fit with elastic net regularized regression and evaluated with cross-validation. From these models, the variables with the 10% highest estimates were identified.

The person-specific models had a median AUC of 0.61, 0.80, and 0.85 for BE, alcohol use, and BD respectively, while the pooled models had a median AUC of 0.70, 0.90, and 0.93. The most important predictors across the behaviors were craving and time of day. However, predictors concerning social context and affect differed among BE, alcohol use, and BD.

Pooled models outperformed person-specific models and the models for alcohol use and BD outperformed those for BE. Future studies should explore how the performance of these models can be improved and how they can be used to deliver interventions in daily life.

Mood disorders are characterized by great heterogeneity in clinical manifestation. Uncovering such heterogeneity using neuroimaging-based individual biomarkers, clinical behaviors, and genetic risks, might contribute to elucidating the etiology of these diseases and support precision medicine.

We recruited 174 drug-naïve and drug-free patients with major depressive disorder and bipolar disorder, as well as 404 healthy controls. T1 MRI imaging data, clinical symptoms, and neurocognitive assessments, and genetics were obtained and analyzed. We applied regional gray matter volumes (GMV) and quantile normative modeling to create maturation curves, and then calculated individual deviations to identify subtypes within the patients using hierarchical clustering. We compared the between-subtype differences in GMV deviations, clinical behaviors, cell-specific transcriptomic associations, and polygenic risk scores. We also validated the GMV deviations based subtyping analysis in a replication cohort.

Two subtypes emerged: subtype 1, characterized by increased GMV deviations in the frontal cortex, cognitive impairment, a higher genetic risk for Alzheimer's disease, and transcriptionally associated with Alzheimer's disease pathways, oligodendrocytes, and endothelial cells; and subtype 2, displaying globally decreased GMV deviations, more severe depressive symptoms, increased genetic vulnerability to major depressive disorder and transcriptionally related to microglia and inhibitory neurons. The distinct patterns of GMV deviations in the frontal, cingulate, and primary motor cortices between subtypes were shown to be replicable.

Our current results provide vital links between MRI-derived phenotypes, spatial transcriptome, genetic vulnerability, and clinical manifestation, and uncover the heterogeneity of mood disorders in biological and behavioral terms.