Network modeling of anxiety and psychological characteristics on suicidal behavior: Cross-sectional study

Explainable multimodal prediction of treatment-resistance in patients with depression leveraging brain morphometry and natural language processing

Although 20 % of patients with depression receiving treatment do not achieve remission, predicting treatment-resistant depression (TRD) remains challenging. In this study, we aimed to develop an explainable multimodal prediction model for TRD using structured electronic medical record data, brain morphometry, and natural language processing. In total, 247 patients with a new depressive episode were included. TRD-predictive models were developed based on the combination of following parameters: selected tabular dataset features, independent components-map weightings from brain T1-weighted magnetic resonance imaging (MRI), and topic probabilities from clinical notes. All models applied the extreme gradient boosting (XGBoost) algorithm via five-fold cross-validation. The model using all data sources showed the highest area under the receiver operating characteristic of 0.794, followed by models that used combined brain MRI and structured data, brain MRI and clinical notes, clinical notes and structured data, brain MRI only, structured data only, and clinical notes only (0.770, 0.762, 0.728, 0.703, 0.684, and 0.569, respectively). Classifications of TRD were driven by several predictors, such as previous exposure to antidepressants and antihypertensive medications, sensorimotor network, default mode network, and somatic symptoms. Our findings suggest that a combination of clinical data with neuroimaging and natural language processing variables improves the prediction of TRD.

Insomnia moderates the association between psychotic-like experiences and suicidal ideation in a non-clinical population: a network analysis

Psychotic-like experiences (PLEs) have been associated with poor sleep quality and increased suicide risk. However, the association between PLEs, insomnia and suicide risk has not been thoroughly investigated in prior studies. In this study, we aimed to explore as to whether insomnia moderates the association between PLEs and suicidal ideation. The study was performed in 4203 young adults (aged 18-35 years, 63.8% females). Data were collected using self-reports. Moderation analysis demonstrated that PLEs are associated with higher levels of the current suicidal ideation only in participants with greater severity of insomnia (B = 0.003, p < 0.001). This analysis included age, gender, education, occupation and depressive symptoms as covariates. Moreover, the network analysis demonstrated that nodes representing PLEs are connected to the node of current suicidal ideation only in participants with greater severity of insomnia. The nodes of PLEs connected to the current suicidal ideation node captured PLEs representing deja vu experiences, auditory hallucination-like experiences and paranoia (edge weights between 0.011 and 0.083). Furthermore, nodes representing PLEs were the three most central nodes in the network analysis of individuals with higher levels of insomnia (strength centrality between 0.96 and 1.10). In turn, the three most central nodes were represented by depressive symptoms in the network analysis of individuals with lower levels of insomnia (strength centrality between 0.67 and 0.79). Findings from this study indicate that insomnia might be an important risk factor of suicide in people with PLEs, especially those reporting deja vu experiences, auditory hallucination-like experiences and paranoia.

A Network Analysis Approach on the Psychological Flexibility/Inflexibility Model

Acceptance and Commitment Therapy (ACT) is purported to work via targeting six interrelated processes summarized as the Psychological Inflexibility/Psychological Flexibility (PI/PF) model. However, the theoretical structure and interconnections of this model have not been sufficiently explored. Lacking are examinations of the interrelations among its components. Network Analysis (NA) can model PI/PF as a system of interconnected variables. We aimed at exploring the role and associations of the PI/PF model's components using NA in two different samples and sets of scales, and compare its structure across sub-samples. Sample 1 consisted of 501 individuals, who completed an online battery of questionnaires including the Multidimensional Psychological Flexibility Inventory, and Sample 2 consisted of 428 people, who completed an online set of six ACT measures, each assessing a component of the PI/PF model. NA could not verify the six ACT dimensions as distinct components. Values and Committed Action components were found to be strongly associated and combined in a group in both sets of measures and samples. Interestingly, Acceptance and Defusion were not the most central components as purported in some ACT conceptualizations, whereas Self-as-Context had a key role on both sets of measures and its items were often merged with Present Moment Awareness items. No significant differences were found in the comparison of networks across different subsamples and sets of scales. After combining different sets of scales, the six ACT components could not be completely verified as distinct entities, which might reflect problems with the theoretical model, or with the scales used. All components had critical roles in the model, particularly Self-as-Context, which reflects the need to redirect research towards this understudied construct. Findings point towards considerations of a triflex instead of a hexaflex ACT model.

Predictors of suicidal ideation severity among treatment-seeking young people with major depressive disorder: The role of state and trait anxiety

OBJECTIVE

Depression and suicidal ideation are closely intertwined. Yet, among young people with depression, the specific factors that contribute to changes in suicidal ideation over time are uncertain. Factors other than depressive symptom severity, such as comorbid psychopathology and personality traits, might be important contributors. Our aim was to identify contributors to fluctuations in suicidal ideation severity over a 12-week period in young people with major depressive disorder receiving cognitive behavioural therapy.

METHODS

Data were drawn from two 12-week randomised, placebo-controlled treatment trials. Participants (N = 283) were 15-25 years old, with moderate to severe major depressive disorder. The primary outcome measure was the Suicidal Ideation Questionnaire, administered at baseline and weeks 4, 8 and 12. A series of linear mixed models was conducted to examine the relationship between Suicidal Ideation Questionnaire score and demographic characteristics, comorbid psychopathology, personality traits and alcohol use.

RESULTS

Depression and anxiety symptom severity, and trait anxiety, independently predicted higher suicidal ideation, after adjusting for the effects of time, demographics, affective instability, non-suicidal self-injury and alcohol use.

CONCLUSIONS

Both state and trait anxiety are important longitudinal correlates of suicidal ideation in depressed young people receiving cognitive behavioural therapy, independent of depression severity. Reducing acute psychological distress, through reducing depression and anxiety symptom severity, is important, but interventions aimed at treating trait anxiety could also potentially be an effective intervention approach for suicidal ideation in young people with depression.

Gender differences in anxiety and depressive symptomatology determined by network analysis in panic disorder

BACKGROUND

It has been suggested that gender differences in anxiety and depressive symptoms characterize panic disorder (PD) in terms of vulnerability to stressful life events, anxiety, depressive symptom patterns, and brain structure. However, few studies have investigated the gender differences in PD using a network approach.

METHODS

This study included 619 participants with PD (313 men). The Panic Disorder Severity Scale, Albany Panic and Phobia Questionnaire, and Beck Depression Inventory-II were used to evaluate symptomatology. To investigate the PD-related white matter (WM) neural correlates, tract-based spatial statistics were used. The PD-related clinical scales and WM neural correlates were included in the network analysis to identify associations between variables. To evaluate network differences between genders, network comparison tests were conducted.

RESULTS

Our findings revealed that agoraphobia in men was the strongest central symptom. In addition, loss of pleasure, and not anxiety or panic symptoms, was the strongest central symptom in women with PD. The network comparison test revealed that the bridge strength score was higher in agoraphobia and tiredness in men and in self-criticalness in women. Furthermore, in the network that includes neural correlates of WM, the bridge strength score was higher in the cingulate gyrus WM in men and the cingulum hippocampus in women.

LIMITATIONS

Since this is a cross-sectional network study of PD patients, the causal relationship between interactions in this network analysis for both genders may not be accurately determined.

CONCLUSION

Network structures of anxiety and depressive symptomatology and related WM neural correlates can differ according to gender in PD patients.

Psychosis Relapse Prediction Leveraging Electronic Health Records Data and Natural Language Processing Enrichment Methods

BACKGROUND

Identifying patients at a high risk of psychosis relapse is crucial for early interventions. A relevant psychiatric clinical context is often recorded in clinical notes; however, the utilization of unstructured data remains limited. This study aimed to develop psychosis-relapse prediction models using various types of clinical notes and structured data.

METHODS

Clinical data were extracted from the electronic health records of the Ajou University Medical Center in South Korea. The study population included patients with psychotic disorders, and outcome was psychosis relapse within 1 year. Using only structured data, we developed an initial prediction model, then three natural language processing (NLP)-enriched models using three types of clinical notes (psychological tests, admission notes, and initial nursing assessment) and one complete model. Latent Dirichlet Allocation was used to cluster the clinical context into similar topics. All models applied the least absolute shrinkage and selection operator logistic regression algorithm. We also performed an external validation using another hospital database.

RESULTS

A total of 330 patients were included, and 62 (18.8%) experienced psychosis relapse. Six predictors were used in the initial model and 10 additional topics from Latent Dirichlet Allocation processing were added in the enriched models. The model derived from all notes showed the highest value of the area under the receiver operating characteristic (AUROC = 0.946) in the internal validation, followed by models based on the psychological test notes, admission notes, initial nursing assessments, and structured data only (0.902, 0.855, 0.798, and 0.784, respectively). The external validation was performed using only the initial nursing assessment note, and the AUROC was 0.616.

CONCLUSIONS

We developed prediction models for psychosis relapse using the NLP-enrichment method. Models using clinical notes were more effective than models using only structured data, suggesting the importance of unstructured data in psychosis prediction.