Bayesian Estimation for Gaussian Graphical Models: Structure Learning, Predictability, and Network Comparisons

Network Analysis of Behavioral Activation/Inhibition Systems and Brain Volume in Individuals With and Without Major Depressive Disorder or Social Anxiety Disorder

BACKGROUND

Social anxiety disorder (SAD) and major depressive disorder (MDD) are characterized by behavioral abnormalities in motivational systems, namely the behavioral inhibition system (BIS) and behavioral activation system (BAS). Limited studies indicate brain volume in regions that support emotion, learning/memory, reward, and cognitive functions relate to BIS/BAS. To increase understanding of BIS/BAS, the current study used a network approach.

METHODS

Patients with SAD (n = 59), patients with MDD (n = 64), and healthy control participants (n = 36) completed a BIS/BAS questionnaire and structural magnetic resonance imaging scans; volumetric regions of interest comprised cortical and limbic structures based on previous BIS/BAS studies. A Bayesian Gaussian graphical model was used for each diagnostic group, and groups were compared. Among network metrics, bridge centrality was of primary interest. Analysis of variance evaluated BIS/BAS behaviors between groups.

RESULTS

Bridge centrality showed hippocampus positively related to BAS, but not to BIS, in the MDD group; no findings were observed in the SAD or control groups. Yet, network density (i.e., overall strength of relationships between variables) and degree centrality (i.e., overall relationship between one variable to all other variables) showed that cortical (e.g., precuneus, medial orbitofrontal) and subcortical (e.g., amygdala, hippocampus) regions differed between diagnostic groups. Analysis of variance results showed BAS was lower in the MDD/SAD groups compared with the control group, while BIS was higher in the SAD group relative to the MDD group, which in turn was higher than the control group.

CONCLUSIONS

Preliminary findings indicate that network-level aberrations may underlie motivational abnormalities in MDD and SAD. Evidence of BIS/BAS differences builds on previous work that points to shared and distinct motivational differences in internalizing psychopathologies.

Testing Conditional Independence in Psychometric Networks: An Analysis of Three Bayesian Methods

Network psychometrics uses graphical models to assess the network structure of psychological variables. An important task in their analysis is determining which variables are unrelated in the network, i.e., are independent given the rest of the network variables. This conditional independence structure is a gateway to understanding the causal structure underlying psychological processes. Thus, it is crucial to have an appropriate method for evaluating conditional independence and dependence hypotheses. Bayesian approaches to testing such hypotheses allow researchers to differentiate between absence of evidence and evidence of absence of connections (edges) between pairs of variables in a network. Three Bayesian approaches to assessing conditional independence have been proposed in the network psychometrics literature. We believe that their theoretical foundations are not widely known, and therefore we provide a conceptual review of the proposed methods and highlight their strengths and limitations through a simulation study. We also illustrate the methods using an empirical example with data on Dark Triad Personality. Finally, we provide recommendations on how to choose the optimal method and discuss the current gaps in the literature on this important topic.

The Mediating Role of Anxiety/Depression Between Auditory Verbal Hallucinations and the Level of Insight in Schizophrenia

OBJECTIVE

Auditory verbal hallucination (AVH) is a prominent symptom of schizophrenia causing profound distress. The influence of AVHs on insight appears to be intricate and contingent on other accompanying symptoms. This study investigated the relationship and possible mediators between AVHs and the degree of insight.

METHODS

One hundred patients with schizophrenia participated in the study. Scales were used to evaluate the hallucinatory experience, the level of insight and other psychopathology. Complex relationships between variables were envisaged as a path model, whose initial structure was constructed via Gaussian Graphical Model. The validity of the final model was verified by Structural Equation Modeling. Separate analyses were performed for self-reported and clinician-rated data to enhance the model's robustness.

RESULTS

The greater the severity of the physical aspects of AVHs, the lower the level of insight observed. Conversely, higher emotional distress was associated with increased insight. These relationships were only evident in the self-reported results and were not reflected in the clinician-rated results. The path model suggested that the Positive and Negative Syndrome Scale (PANSS) anxiety/depression factor was an important mediator that linked the found association. Notably, the PANSS negative symptom had the opposite effect on the PANSS anxiety/depression factor and insight, making it difficult to define its overall effect.

CONCLUSION

The findings of this study provided one possible route for the positive influence of AVH experience in gaining insight. The mediating role of anxiety/depression modified by negative symptoms emerged as a valuable concept for clarifying this intricate relationship.

Psychometric Properties of the Chinese Version of 20-Item Zimbardo Time Perspective Inventory (C-ZTPI-20) in Chinese Adolescent Population

Background

This study assesses the psychometric properties of the Chinese version of the Zimbardo Time Perspective Inventory (C-ZTPI-20) in an adolescent population.

Methods

The investigation encompasses a sample of 2634 middle school students from China and aims to evaluate the instrument's reliability, structural validity, measurement invariance, criterion validity, and network structure attributes.

Results

First, descriptive analysis revealed satisfactory reliabilities for four out of five C-ZTPI-20 dimensions, with Present Fatalistic (PF) exhibiting relatively low reliability. Moreover, Confirmatory Factor Analysis (CFA) supported the 5-dimensional structure across all samples and sexes, albeit with a modest Tucker-Lewis Index (TLI) for girls. Furthermore, measurement invariance analysis underscores unbiased assessment across sexes. Sex differences emerge in the Present Hedonistic (PH) dimension, where boys showed higher scores. Furthermore, criteria validity analysis revealed that Past Positive (PP) and Future (F) were positively associated with extraversion, agreeableness, conscientiousness, openness, grit, and mental health, while they were negatively associated with neuroticism. Past Negative (PN) and PF showed inverse trends, while PH perspective demonstrated complex, varied correlations with these psychological traits, underscoring the multifaceted nature of time perspectives. Finally, network analysis revealed positive inter-correlations within dimensions and significant edge differences between sexes, particularly in inter-dimension connections. Despite differing rankings, the most central and marginal items remained consistent between boys and girls in network models.

Conclusion

These findings contribute to understanding the C-ZTPI-20's effectiveness in assessing adolescent time perspectives and inform interventions promoting psychological well-being and coping strategies.

The clinical presentation of major depressive disorder in youth with co-occurring obsessive-compulsive disorder

BACKGROUND

Major depressive disorder (MDD) is common in youth and among the most frequent comorbid disorders in pediatric obsessive-compulsive disorder (OCD), but it is unclear whether the presence of OCD affects the symptom presentation of MDD in youth.

METHODS

A sample of youth with OCD and MDD (n = 124) and a sample of youth with MDD but no OCD (n = 673) completed the Patient Health Questionnaire for Adolescents (PHQ-A). The overall and symptom-level presentation of MDD were examined using group comparisons and network analysis.

RESULTS

Youth with MDD and OCD, compared to those with MDD and no OCD, had more severe MDD (Cohen's d = 0.39) and more reported moderate to severe depression (75 % vs 61 %). When accounting for demographic variables and the overall severity of MDD, those with comorbid OCD reported lower levels of anhedonia and more severe difficulties with psychomotor retardation/agitation. No significant differences in the interconnections among symptoms emerged.

LIMITATIONS

Data were cross-sectional and self-reported, gold standard diagnostic tools were not used to assess OCD, and the sample size for the group with MDD and OCD was relatively small yielding low statistical power for network analysis.

CONCLUSIONS

Youth with MDD and OCD have more severe MDD than those with MDD and no OCD and they experience more psychomotor issues and less anhedonia, which may relate to the behavioral activation characteristic of OCD.

Comparing community detection algorithms in psychometric networks: A Monte Carlo simulation

Identifying the correct number of factors in multivariate data is fundamental to psychological measurement. Factor analysis has a long tradition in the field, but it has been challenged recently by exploratory graph analysis (EGA), an approach based on network psychometrics. EGA first estimates a network and then applies the Walktrap community detection algorithm. Simulation studies have demonstrated that EGA has comparable or better accuracy for recovering the same number of communities as there are factors in the simulated data than factor analytic methods. Despite EGA's effectiveness, there has yet to be an investigation into whether other sparsity induction methods or community detection algorithms could achieve equivalent or better performance. Furthermore, unidimensional structures are fundamental to psychological measurement yet they have been sparsely studied in simulations using community detection algorithms. In the present study, we performed a Monte Carlo simulation using the zero-order correlation matrix, GLASSO, and two variants of a non-regularized partial correlation sparsity induction methods with several community detection algorithms. We examined the performance of these method-algorithm combinations in both continuous and polytomous data across a variety of conditions. The results indicate that the Fast-greedy, Louvain, and Walktrap algorithms paired with the GLASSO method were consistently among the most accurate and least-biased overall.

Aesthetic emotions are affected by context: a psychometric network analysis

Aesthetic emotions are defined as emotions arising when a person evaluates a stimulus for its aesthetic appeal. Whether these emotions are unique to aesthetic activities is debated. We address this debate by examining if recollections of different types of engaging activities entail different emotional profiles. A large sample of participants were asked to recall engaging aesthetic (N = 167), non-aesthetic (N = 160), or consumer (N = 172) activities. They rated the extent to which 75 candidate aesthetic emotions were evoked by these activities. We applied a computational psychometric network approach to represent and compare the space of these emotions across the three conditions. At the behavioral level, recalled aesthetic activities were rated as the least vivid but most intense compared to the two other conditions. At the network level, we found several quantitative differences across the three conditions, related to the typology, community (clusters) and core nodes (emotions) of these networks. Our results suggest that aesthetic and non-aesthetic activities evoke emotional spaces differently. Thus, we propose that aesthetic emotions are distributed differently in a multidimensional aesthetic space than for other engaging activities. Our results highlight the context-specificity of aesthetic emotions.

Comorbidity among depression, anxiety and stress symptoms in naturalistic clinical samples: A cross-cultural network analysis

Comorbidity between depression and anxiety is well-established across various settings and cultures. We approached comorbidity from the network psychopathology perspective and examined the depression, anxiety/autonomic arousal and stress/tension symptoms in naturalistic clinical samples from Serbia, Italy and Croatia. This was a multisite study in which regularized partial correlation networks of the symptoms, obtained via self-reports on the Depression Anxiety and Stress Scales-21 (DASS-21) in three cross-cultural, clinical samples (total N = 874), were compared with respect to centrality, edge weights, community structure and bridge centrality. A moderate degree of similarity in a number of network indices across the three networks was observed. While negative mood emerged to be the most central node, stress/tension nodes were the most likely bridge symptoms between depressive and anxiety/autonomic arousal symptoms. We demonstrated that the network structure and features in mixed clinical samples were similar across three different languages and cultures. The symptoms such as agitation, restlessness and inability to relax functioned as bridges across the three symptom communities explored in this study. Important theoretical and clinical implications were derived.

Network structure of the links between extracurricular time-use and delinquent behaviors: Moving forward and beyond linear relations

Using psychological network analysis, this study explored the heterogeneity of the network structure between extracurricular time-use and delinquency using a nationally representative longitudinal survey of at-school students in China (N = 10,279, 47.3% female, average age 13.6, 91.2% Han ethnicity). The results are threefold: First, time stimulation of activities occurs on weekdays, while time displacement and stimulation occur on weekends. Second, delinquent behaviors are positively correlated, forming a problem behavior syndrome. Smoking or drinking is the central delinquent behavior. Third, negative consequences of specific time-use behaviors are more likely to occur on weekends than on weekdays, and time-use behavior may function differently on weekdays versus weekends. Among them, going to coffeenets or game-centers serves as the bridge with the highest potential of triggering delinquency.

Unique Variable Analysis: A Network Psychometrics Method to Detect Local Dependence

The local independence assumption states that variables are unrelated after conditioning on a latent variable. Common problems that arise from violations of this assumption include model misspecification, biased model parameters, and inaccurate estimates of internal structure. These problems are not limited to latent variable models but also apply to network psychometrics. This paper proposes a novel network psychometric approach to detect locally dependent pairs of variables using network modeling and a graph theory measure called weighted topological overlap (wTO). Using simulation, this approach is compared to contemporary local dependence detection methods such as exploratory structural equation modeling with standardized expected parameter change and a recently developed approach using partial correlations and a resampling procedure. Different approaches to determine local dependence using statistical significance and cutoff values are also compared. Continuous, polytomous (5-point Likert scale), and dichotomous (binary) data were generated with skew across a variety of conditions. Our results indicate that cutoff values work better than significance approaches. Overall, the network psychometrics approaches using wTO with graphical least absolute shrinkage and selector operator with extended Bayesian information criterion and wTO with Bayesian Gaussian graphical model were the best performing local dependence detection methods overall.

Networks of Neurodevelopmental Traits, Socioenvironmental Factors, Emotional Dysregulation in Childhood, and Depressive Symptoms Across Development in Two U.K. Cohorts

OBJECTIVE

Previous population-based studies have identified associations between childhood neurodevelopmental traits and depression in childhood, adolescence, and young adulthood. However, neurodevelopmental traits are highly correlated with each other, which could confound associations when traits are examined in isolation. The authors sought to identify unique associations between multiple neurodevelopmental traits in childhood and depressive symptoms across development, while taking into account co-occurring difficulties, in multivariate analyses.

METHODS

Data from two U.K. population-based cohorts, the Twins Early Development Study (TEDS) (N=4,407 independent twins) and the Avon Longitudinal Study of Parents and Children (ALSPAC) (N=10,351), were independently analyzed. Bayesian Gaussian graphical models were estimated to investigate pairwise conditional associations between neurodevelopmental traits (autism and ADHD symptoms and general cognitive, learning, and communication abilities), socioenvironmental stressors (academic performance and peer relations), and emotional dysregulation in childhood (ages 7-11) and depressive symptoms across development (ages 12, 16, and 21).

RESULTS

In both cohorts, bivariate correlations indicated several associations between neurodevelopmental traits and depressive symptoms across development. However, based on replicated findings across cohorts, these pairs of variables were mostly conditionally independent, and none were conditionally associated, after accounting for socioenvironmental stressors and emotional dysregulation. In turn, socioenvironmental stressors and emotional dysregulation were conditionally associated with both neurodevelopmental traits and depressive symptoms. Based on replicated findings across cohorts, neurodevelopmental traits in childhood could be associated only indirectly with depressive symptoms across development.

CONCLUSIONS

This study indicates that associations between childhood neurodevelopmental traits and depressive symptoms across development could be explained by socioenvironmental stressors and emotional dysregulation. The present findings could inform future research aimed at the prevention of depression in youths with neurodevelopmental disorders.

Psychometric Properties of the Chinese Version of the 10-Item Social Provisions Scale in Chinese Populations

This study performed a cross-cultural validation of the Chinese version of the 10-item Social Provisions Scale (C-SPS-10) in Chinese populations. Study 1 examined the factor structure, internal reliability, discrimination, criterion validity, and network structure of C-SPS-10 by utilizing a sample of disaster victims in the 2021 Henan floods. Study 2 substantiated the findings of Study 1 in a general population sample. Measurement invariances between populations and between sexes in terms of the C-SPS-10 were also tested using the network approach. Study 3 used three samples to examine the test-retest reliability of the C-SPS-10 over three different time periods. The general results showed that the C-SPS-10 has excellent factor structure, internal reliability, discrimination, and criterion validity. The C-SPS-10 was confirmed to have good psychometric properties. Although the full scale functions well, problems may exist at a domain level. Moreover, the full scale of the C-SPS-10 was varied as a useful tool to capture trait-like characteristics of individuals' perceptions of social support for the general population.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10862-023-10047-7.

Alcohol, cannabis, and nicotine use have distinct associations with COVID-19 pandemic-related experiences: An exploratory Bayesian network analysis across two timepoints

BACKGROUND

Substance use trends during the COVID-19 pandemic have been extensively documented. However, relatively less is known about the associations between pandemic-related experiences and substance use.

METHOD

In July 2020 and January 2021, a broad U.S. community sample (N = 1123) completed online assessments of past month alcohol, cannabis, and nicotine use and the 92-item Epidemic-Pandemic Impacts Inventory, a multidimensional measure of pandemic-related experiences. We examined links between substance use frequency, and pandemic impact on emotional, physical, economic, and other key domains, using Bayesian Gaussian graphical networks in which edges represent significant associations between variables (referred to as nodes). Bayesian network comparison approaches were used to assess the evidence of stability (or change) in associations between the two timepoints.

RESULTS

After controlling for all other nodes in the network, multiple significant edges connecting substance use nodes and pandemic-experience nodes were observed across both time points, including positive- (r range 0.07-0.23) and negative-associations (r range -0.25 to -0.11). Alcohol was positively associated with social and emotional pandemic impacts and negatively associated with economic impacts. Nicotine was positively associated with economic impact and negatively associated with social impact. Cannabis was positively associated with emotional impact. Network comparison suggested these associations were stable across the two timepoints.

CONCLUSION

Alcohol, nicotine, and cannabis use had unique associations to a few specific domains among a broad range of pandemic-related experiences. Given the cross-sectional nature of these analyses with observational data, further investigation is needed to identify potential causal links.

Disentangling heterogeneity of psychosis expression in the general population: sex-specific moderation effects of environmental risk factors on symptom networks

BACKGROUND

Psychosis expression in the general population may reflect a behavioral manifestation of the risk for psychotic disorder. It can be conceptualized as an interconnected system of psychotic and affective experiences; a so-called 'symptom network'. Differences in demographics, as well as exposure to adversities and risk factors, may produce substantial heterogeneity in symptom networks, highlighting potential etiological divergence in psychosis risk.

METHODS

To explore this idea in a data-driven way, we employed a novel recursive partitioning approach in the 2007 English National Survey of Psychiatric Morbidity (N = 7242). We sought to identify 'network phenotypes' by explaining heterogeneity in symptom networks through potential moderators, including age, sex, ethnicity, deprivation, childhood abuse, separation from parents, bullying, domestic violence, cannabis use, and alcohol.

RESULTS

Sex was the primary source of heterogeneity in symptom networks. Additional heterogeneity was explained by interpersonal trauma (childhood abuse and domestic violence) in women and domestic violence, cannabis use, ethnicity in men. Among women, especially those exposed to early interpersonal trauma, an affective loading within psychosis may have distinct relevance. Men, particularly those from minority ethnic groups, demonstrated a strong network connection between hallucinatory experiences and persecutory ideation.

CONCLUSION

Symptom networks of psychosis expression in the general population are highly heterogeneous. The structure of symptom networks seems to reflect distinct sex-related adversities, etiologies, and mechanisms of symptom-expression. Disentangling the complex interplay of sex, minority ethnic group status, and other risk factors may help optimize early intervention and prevention strategies in psychosis.

Gaussian graphical models with applications to omics analyses

Gaussian graphical models (GGMs) provide a framework for modeling conditional dependencies in multivariate data. In this tutorial, we provide an overview of GGM theory and a demonstration of various GGM tools in R. The mathematical foundations of GGMs are introduced with the goal of enabling the researcher to draw practical conclusions by interpreting model results. Background literature is presented, emphasizing methods recently developed for high-dimensional applications such as genomics, proteomics, or metabolomics. The application of these methods is illustrated using a publicly available dataset of gene expression profiles from 578 participants with ovarian cancer in The Cancer Genome Atlas. Stand-alone code for the demonstration is available as an RMarkdown file at https://github.com/katehoffshutta/ggmTutorial.

Probabilistic edge inference of gene networks with markov random field-based bayesian learning

Current algorithms for gene regulatory network construction based on Gaussian graphical models focuses on the deterministic decision of whether an edge exists. Both the probabilistic inference of edge existence and the relative strength of edges are often overlooked, either because the computational algorithms cannot account for this uncertainty or because it is not straightforward in implementation. In this study, we combine the Bayesian Markov random field and the conditional autoregressive (CAR) model to tackle simultaneously these two tasks. The uncertainty of edge existence and the relative strength of edges can be measured and quantified based on a Bayesian model such as the CAR model and the spike-and-slab lasso prior. In addition, the strength of the edges can be utilized to prioritize the importance of the edges in a network graph. Simulations and a glioblastoma cancer study were carried out to assess the proposed model’s performance and to compare it with existing methods when a binary decision is of interest. The proposed approach shows stable performance and may provide novel structures with biological insights.

Relating balance and conditional independence in graphical models

When data are available for all nodes of a Gaussian graphical model, then, it is possible to use sample correlations and partial correlations to test to what extent the conditional independencies that encode the structure of the model are indeed verified by the data. In this paper, we give a heuristic rule useful in such a validation process: When the correlation subgraph involved in a conditional independence is balanced (i.e., all its cycles have an even number of negative edges), then a partial correlation is usually a contraction of the corresponding correlation, which often leads to conditional independence. In particular, the contraction rule can be made rigorous if we look at concentration subgraphs rather than correlation subgraphs. The rule is applied to real data for elementary gene regulatory motifs.

Network Analysis of Time Use and Depressive Symptoms Among Emerging Adults: Findings From the Guizhou Population Health Cohort Study

Background

To date, the relationship between diverse time use behaviors and depression status among emerging adults have not been disentangled in the literature. Therefore, if and how the time displacement mechanism activates depressive symptoms among emerging adults remains unclear.

Methods

To fill this gap in the literature, we employed a network analysis to make estimations. The emerging adult sample (N = 1,811) was collected by the Guizhou Population Health Cohort Study. Time use behaviors were measured by an adaption of the self-administered International Physical Activity Questionnaire, and depressive symptoms were assessed using the 9-item Patient Health Questionnaire (PHQ-9).

Results

The results revealed that the time displacement mechanism of emerging adults differed from that of adolescents. Sleep duration was not crowded out by other activities, while the time spent on computer use was found to be negatively related to time spent on heavy work activities. Moreover, computer use behavior triggered three depressive symptoms ("Anhedonia," "Guilt," and "Motor"), but inhibited "Suicide." The results of the directed acyclic graph revealed that females and heavy drinkers were at risk of depression.

Limitations

The study sample was confined to only one province, which may limit its generalizability. The cross-sectional design impeded the ability to draw causal inferences.

Conclusion

Our results enhance the current understanding of the internal mechanism of how time use behaviors influence depressive symptoms among emerging adults.

Comparing PTSD symptom networks in type I vs. type II trauma survivors

Background: Network analysis has gained increasing attention as a new framework to study complex associations between symptoms of post-traumatic stress disorder (PTSD). A number of studies have been published to investigate symptom networks on different sets of symptoms in different populations, and the findings have been inconsistent. Objective: We aimed to extend previous research by testing whether differences in PTSD symptom networks can be found in survivors of type I (single event; sudden and unexpected, high levels of acute threat) vs. type II (repeated and/or protracted; anticipated) trauma (with regard to their index trauma). Method: Participants were trauma-exposed individuals with elevated levels of PTSD symptomatology, most of whom (94%) were undergoing assessment in preparation for PTSD treatment in several treatment centres in Germany and Switzerland (n = 286 with type I and n = 187 with type II trauma). We estimated Bayesian Gaussian graphical models for each trauma group and explored group differences in the symptom network. Results: First, for both trauma types, our analyses identified the edges that were repeatedly reported in previous network studies. Second, there was decisive evidence that the two networks were generated from different multivariate normal distributions, i.e. the networks differed on a global level. Third, explorative edge-wise comparisons showed moderate or strong evidence for specific 12 edges. Edges which emerged as especially important in distinguishing the networks were between intrusions and flashbacks, highlighting the stronger positive association in the group of type II trauma survivors compared to type I survivors. Flashbacks showed a similar pattern of results in the associations with detachment and sleep problems (type II > type I). Conclusion: Our findings suggest that trauma type contributes to the heterogeneity in the symptom network. Future research on PTSD symptom networks should include this variable in the analyses to reduce heterogeneity.

Psychopathological symptom network structure in transgender and gender queer youth reporting parental psychological abuse: a network analysis

BACKGROUND

This is the first study to investigate the effect of parental psychological abuse on potential psychopathological symptoms in gender minority youth subgroups, including transgender women, transgender men, and gender queer individuals.

METHODS

Data was analysed from the Chinese National Transgender Survey in 2017; the survey was distributed through community-based organizations to transgender adolescents and adults residing in China, with representation from all 32 provinces and autonomous regions. A total of 1293 youth that self-identified as transgender or gender queer completed the study. Measures covered psychopathological symptoms including depression, anxiety, risk of suicideand self-harm. Parental psychological abuse was assessed in terms of neglect and avoidance, force to change, and verbal insults. Both the edges and centralities were computed via network analysis, and the network properties were then compared among the three gender minority subgroups. In addition, linear regression was adopted to test the predictive ability of node centrality for low self-esteem.

RESULTS

Descriptive analysis revealed that among the three subgroups, transgender women had more severe psychopathological symptoms and reported the most psychological abuse. Network analysis revealed that the risk of suicide and self-harm was directly connected with one type of parental psychological abuse ("neglect and avoidance"). Node centrality was significantly associated with the predicting value of the nodes on low self-esteem (r² = 0.25, 0.17, 0.31) among all three gender minority subgroups.

CONCLUSIONS

The distinctive core psychopathological symptoms, within the networks of the gender minority subgroups, revealed specific symptoms across each group. The significant association between node centrality and low self-esteem indicated the extent of parental psychological abuse. Parental psychological abuse directed towards gender minority youth should be recognized as a form of family cold violence. It is recommended that schools and local communities should support early intervention to improve psychological well-being.

The exploratory value of cross-sectional partial correlation networks: Predicting relationships between change trajectories in borderline personality disorder

Objective

Within the network approach to psychopathology, cross-sectional partial correlation networks have frequently been used to estimate relationships between symptoms. The resulting relationships have been used to generate hypotheses about causal links between symptoms. In order to justify such exploratory use of partial correlation networks, one needs to assume that the between-subjects relationships in the network approximate systematic within-subjects relationships, which are in turn the results of some within-subjects causal mechanism. If this assumption holds, relationships in the network should be mirrored by relationships between symptom changes; if links in networks approximate systematic within-subject relationships, change in a symptom should relate to change in connected symptoms.

Method

To investigate this implication, we combined longitudinal data on the Borderline Personality Disorder Severity Index from four samples of borderline personality disorder patients (N = 683). We related parameters from baseline partial correlation networks of symptoms to relationships between change trajectories of these symptoms.

Results

Across multiple levels of analysis, our results showed that parameters from baseline partial correlation networks are strongly predictive of relationships between change trajectories.

Conclusions

By confirming its implication, our results support the idea that cross-sectional partial correlation networks hold a relevant amount of information about systematic within-subjects relationships and thereby have exploratory value to generate hypotheses about the causal dynamics between symptoms.

Assisted estimation of gene expression graphical models

In the study of gene expression data, network analysis has played a uniquely important role. To accommodate the high dimensionality and low sample size and generate interpretable results, regularized estimation is usually conducted in the construction of gene expression Gaussian Graphical Models (GGM). Here we use GeO-GGM to represent gene-expression-only GGM. Gene expressions are regulated by regulators. gene-expression-regulator GGMs (GeR-GGMs), which accommodate gene expressions as well as their regulators, have been constructed accordingly. In practical data analysis, with a "lack of information" caused by the large number of model parameters, limited sample size, and weak signals, the construction of both GeO-GGMs and GeR-GGMs is often unsatisfactory. In this article, we recognize that with the regulation between gene expressions and regulators, the sparsity structures of a GeO-GGM and its GeR-GGM counterpart can satisfy a hierarchy. Accordingly, we propose a joint estimation which reinforces the hierarchical structure and use the construction of a GeO-GGM to assist that of its GeR-GGM counterpart and vice versa. Consistency properties are rigorously established, and an effective computational algorithm is developed. In simulation, the assisted construction outperforms the separation construction of GeO-GGM and GeR-GGM. Two The Cancer Genome Atlas data sets are analyzed, leading to findings different from the direct competitors.