Beyond integrative experiment design: Systematic experimentation guided by causal discovery AI.

Integrative experiment design is a needed improvement over ad hoc experiments, but the specific proposed method has limitations. We urge a further break with tradition through the use of an enormous untapped resource: Decades of causal discovery artificial intelligence (AI) literature on optimizing the design of systematic experimentation.

Causal discovery analysis: A promising tool in advancing precision medicine for eating disorders.

OBJECTIVE: Precision medicine (i.e., individually tailored treatments) represents an optimal goal for treating complex psychiatric disorders, including eating disorders. Within the eating disorders field, most treatment development efforts have been limited in their ability to identify individual-level models of eating disorder psychopathology and to develop and apply an individually tailored treatment for a given individual's personalized model of psychopathology. In addition, research is still needed to identify causal relationships within a given individual's model of eating disorder psychopathology. Addressing this limitation of the current state of precision medicine-related research in the field will allow us to progress toward advancing research and practice for eating disorders treatment. METHOD: We present a novel set of analytic tools, causal discovery analysis (CDA) methods, which can facilitate increasingly fine-grained, person-specific models of causal relations among cognitive, behavioral, and affective symptoms. RESULTS: CDA can advance the identification of an individual's causal model that maintains that individuals' eating disorder psychopathology. DISCUSSION: In the current article, we (1) introduce CDA methods as a set of promising analytic tools for developing precision medicine methods for eating disorders including the potential strengths and weaknesses of CDA, (2) provide recommendations for future studies utilizing this approach, and (3) outline the potential clinical implications of using CDA to generate personalized models of eating disorder psychopathology. PUBLIC SIGNIFICANCE STATEMENT: CDA provides a novel statistical approach for identifying causal relationships among variables of interest for a given individual. Person-specific causal models may offer a promising approach to individualized treatment planning and inform future personalized treatment development efforts for eating disorders.

Causal pathways to social and occupational functioning in the first episode of schizophrenia: uncovering unmet treatment needs.

BACKGROUND: We aimed to identify unmet treatment needs for improving social and occupational functioning in early schizophrenia using a data-driven causal discovery analysis. METHODS: Demographic, clinical, and psychosocial measures were obtained for 276 participants from the Recovery After an Initial Schizophrenia Episode Early Treatment Program (RAISE-ETP) trial at baseline and 6-months, along with measures of social and occupational functioning from the Quality of Life Scale. The Greedy Fast Causal Inference algorithm was used to learn a partial ancestral graph modeling causal relationships across baseline variables and 6-month functioning. Effect sizes were estimated using a structural equation model. Results were validated in an independent dataset (N = 187). RESULTS: In the data-generated model, greater baseline socio-affective capacity was a cause of greater baseline motivation [Effect size (ES) = 0.77], and motivation was a cause of greater baseline social and occupational functioning (ES = 1.5 and 0.96, respectively), which in turn were causes of their own 6-month outcomes. Six-month motivation was also identified as a cause of occupational functioning (ES = 0.92). Cognitive impairment and duration of untreated psychosis were not direct causes of functioning at either timepoint. The graph for the validation dataset was less determinate, but otherwise supported the findings. CONCLUSIONS: In our data-generated model, baseline socio-affective capacity and motivation are the most direct causes of occupational and social functioning 6 months after entering treatment in early schizophrenia. These findings indicate that socio-affective abilities and motivation are specific high-impact treatment needs that must be addressed in order to promote optimal social and occupational recovery.

Data-Driven Analytics to Discover APRN's Impact on Nursing Home Hospitalization: Causal Discovery Analysis.

OBJECTIVES: Research shows advanced practice registered nurses (APRNs) embedded in nursing homes (NHs) reduce resident hospitalizations. However, the specific APRN activities that reduce hospitalizations have not been adequately investigated. This study aims to identify the causal links between APRN activities and NHs resident hospitalization. The study also examined relationships among other variables, including advanced directives, clinical diagnosis, and length of hospitalization. DESIGN: Secondary data analysis. SETTING AND PARTICIPANTS: Residents of NHs participating in the Missouri Quality Initiative for Nursing Homes, 2016-2019. METHODS: We performed a secondary analysis of data from the Missouri Quality Initiative for Nursing Homes Intervention using causal discovery analysis, a machine learning, data-driven technique to determine causal relationships across data. The resident roster and INTERACT resident hospitalization datasets were combined to create the final dataset. Variables in the analysis model were divided into before and after hospitalization. Expert consensus was used to validate and interpret the outcomes. RESULTS: The research team analyzed 1161 hospitalization events and their associated NH activities. APRNs evaluated NH residents before a transfer, expedited follow-up nursing assessments, and authorized hospitalization when necessary. No significant causal relationships were found between APRN activities and the clinical diagnosis of a resident. The analysis also showed multifaceted relationships related to having advanced directives and duration of hospitalization. CONCLUSIONS AND IMPLICATIONS: This study demonstrated the importance of APRNs embedded in NHs to improve resident outcomes. APRNs in NHs can facilitate communication and collaboration among the nursing team, leading to early identification and treatment for resident status changes. APRNs can also initiate more timely transfers by reducing the need for physician authorization. These findings emphasize the crucial role of APRNs in NHs and suggest that budgeting for APRN services may be an effective strategy to reduce hospitalizations. Additional findings regarding advance directives are discussed.

Frontal tDCS reduces alcohol relapse rates by increasing connections from left dorsolateral prefrontal cortex to addiction networks.

BACKGROUND: Brain-based interventions are needed to address persistent relapse in alcohol use disorder (AUD). Neuroimaging evidence suggests higher frontal connectivity as well as higher within-network connectivity of theoretically defined addiction networks are associated with reduced relapse rates and extended abstinence during follow-up periods. OBJECTIVE: /Hypothesis: A longitudinal randomized double-blind sham-controlled clinical trial investigated whether a non-invasive neuromodulation intervention delivered during early abstinence can (i) modulate connectivity of addiction networks supporting abstinence and (ii) improve relapse rates. HYPOTHESES: Active transcranial direct current stimulation (tDCS) will (i) increase connectivity of addiction networks known to support abstinence and (ii) reduce relapse rates. METHODS: Short-term abstinent AUD participants (n = 60) were assigned to 5 days of either active tDCS or sham during cognitive training. Causal discovery analysis (CDA) examined the directional influence from left dorsolateral prefrontal cortex (LDLPFC, stimulation site) to addiction networks that support abstinence. RESULTS: Active tDCS had an effect on the average strength of CDA-determined connectivity from LDLPFC to the incentive salience and negative emotionality addiction networks - increasing in the active tDCS group only. Active tDCS had an effect on relapse rates following the intervention, with lower probability of relapse in the active tDCS vs. sham. Active tDCS showed an unexpected sex-dependent effect on relapse rates. CONCLUSION: Our results suggest that LDLPFC stimulation delivered during early abstinence has an effect on addiction networks supporting abstinence and on relapse rates. The unexpected sex-dependent neuromodulation effects need to be further examined in larger clinical trials.

One data set, many analysts: Implications for practicing scientists.

Researchers routinely face choices throughout the data analysis process. It is often opaque to readers how these choices are made, how they affect the findings, and whether or not data analysis results are unduly influenced by subjective decisions. This concern is spurring numerous investigations into the variability of data analysis results. The findings demonstrate that different teams analyzing the same data may reach different conclusions. This is the "many-analysts" problem. Previous research on the many-analysts problem focused on demonstrating its existence, without identifying specific practices for solving it. We address this gap by identifying three pitfalls that have contributed to the variability observed in many-analysts publications and providing suggestions on how to avoid them.

Causal Discovery for fMRI data: Challenges, Solutions, and a Case Study.

Designing studies that apply causal discovery requires navigating many researcher degrees of freedom. This complexity is exacerbated when the study involves fMRI data. In this paper we (i) describe nine challenges that occur when applying causal discovery to fMRI data, (ii) discuss the space of decisions that need to be made, (iii) review how a recent case study made those decisions, (iv) and identify existing gaps that could potentially be solved by the development of new methods. Overall, causal discovery is a promising approach for analyzing fMRI data, and multiple successful applications have indicated that it is superior to traditional fMRI functional connectivity methods, but current causal discovery methods for fMRI leave room for improvement.

Integrating Causal Discovery and Clinically-Relevant Insights to Explore Directional Relationships between Autistic Features, Sex at Birth, and Cognitive Abilities.

Prevalence in autism spectrum disorder (ASD) diagnosis has long been strongly male-biased. Yet, consensus has not been reached on mechanisms and clinical features that underlie sex-based discrepancies. Whereas females may be under-diagnosed because of inconsistencies in diagnostic/ascertainment procedures (sex-biased criteria, social camouflaging), diagnosed males may have exhibited more overt behaviors (e.g., hyperactivity, aggression) that prompted clinical evaluation. Applying a novel network-theory-based approach, we extracted data-driven, clinically-relevant insights from a large, well-characterized sample (Simons Simplex Collection) of 2175 autistic males (Ages = 8.9±3.5 years) and 334 autistic females (Ages = 9.2±3.7 years). Exploratory factor analysis (EFA) and expert clinical review reduced data dimensionality to 15 factors of interest. To offset inherent confounds of an imbalanced sample, we identified a subset of males (N=331) matched to females on key variables (Age, IQ) and applied data-driven CDA using Greedy Fast Causal Inference (GFCI) for three groups (All Females, All Males, and Matched Males). Structural equation modeling (SEM) extracted measures of model fit and effect sizes for causal relationships between sex, age, and, IQ on EFA-selected factors capturing phenotypic representations of autism across sensory, social, and restricted and repetitive behavior domains. Our methodology unveiled sex-specific directional relationships to inform developmental outcomes and targeted interventions.

Quantifying heterogeneity in mood-alcohol relationships with idiographic causal models.

BACKGROUND: Ecological momentary assessment (EMA) studies have provided conflicting evidence for the mood regulation tenet that people drink in response to positive and negative moods. The current study examined mood-to-alcohol relationships idiographically to quantify the prevalence and intensity of relationships between positive and negative moods and drinking across individuals. METHOD: We used two EMA samples: 96 heavy drinking college students (sample 1) and 19 young adults completing an ecological momentary intervention (EMI) for drinking to cope (sample 2). Mood and alcohol use were measured multiple times per day for 4-6 weeks. Mood-alcohol relationships were examined using three different analytic approaches: standard multilevel modeling, group causal modeling, and idiographic causal modeling. RESULTS: Both multilevel modeling and group causal modeling showed that participants in both samples drank in response to positive moods only. However, idiographic causal analyses revealed that only 63% and 21% of subjects (in samples 1 and 2, respectively) drank following any positive mood. Many subjects (24% and 58%) did not drink in response to either positive or negative mood in their daily lives, and very few (5% and 16%) drank in response to negative moods throughout the EMA protocol, despite sample 2 being selected specifically because they endorse drinking to cope with negative mood. CONCLUSION: Traditional group-level analyses and corresponding population-wide theories assume relative homogeneity within populations in mood-alcohol relationships, but this nomothetic approach failed to characterize accurately the relationship between mood and alcohol use in approximately half of the subjects in two samples that were demographically and clinically homogeneous. Given inconsistent findings in the mood-alcohol relationships to date, we conclude that idiographic causal analyses can provide a foundation for more accurate theories of mood and alcohol use. In addition, idiographic causal models may also help improve psychosocial treatments through direct use in clinical settings.

Explainable machine learning analysis reveals sex and gender differences in the phenotypic and neurobiological markers of Cannabis Use Disorder.

Cannabis Use Disorder (CUD) has been linked to a complex set of neuro-behavioral risk factors. While many studies have revealed sex and gender differences, the relative importance of these risk factors by sex and gender has not been described. We used an "explainable" machine learning approach that combined decision trees [gradient tree boosting, XGBoost] with factor ranking tools [SHapley's Additive exPlanations (SHAP)] to investigate sex and gender differences in CUD. We confirmed that previously identified environmental, personality, mental health, neurocognitive, and brain factors highly contributed to the classification of cannabis use levels and diagnostic status. Risk factors with larger effect sizes in men included personality (high openness), mental health (high externalizing, high childhood conduct disorder, high fear somaticism), neurocognitive (impulsive delay discounting, slow working memory performance) and brain (low hippocampal volume) factors. Conversely, risk factors with larger effect sizes in women included environmental (low education level, low instrumental support) factors. In summary, environmental factors contributed more strongly to CUD in women, whereas individual factors had a larger importance in men.

Performance of a Chest Radiograph AI Diagnostic Tool for COVID-19: A Prospective Observational Study.

Purpose: To conduct a prospective observational study across 12 U.S. hospitals to evaluate real-time performance of an interpretable artificial intelligence (AI) model to detect COVID-19 on chest radiographs. Materials and Methods: A total of 95 363 chest radiographs were included in model training, external validation, and real-time validation. The model was deployed as a clinical decision support system, and performance was prospectively evaluated. There were 5335 total real-time predictions and a COVID-19 prevalence of 4.8% (258 of 5335). Model performance was assessed with use of receiver operating characteristic analysis, precision-recall curves, and F1 score. Logistic regression was used to evaluate the association of race and sex with AI model diagnostic accuracy. To compare model accuracy with the performance of board-certified radiologists, a third dataset of 1638 images was read independently by two radiologists. Results: Participants positive for COVID-19 had higher COVID-19 diagnostic scores than participants negative for COVID-19 (median, 0.1 [IQR, 0.0-0.8] vs 0.0 [IQR, 0.0-0.1], respectively; P < .001). Real-time model performance was unchanged over 19 weeks of implementation (area under the receiver operating characteristic curve, 0.70; 95% CI: 0.66, 0.73). Model sensitivity was higher in men than women (P = .01), whereas model specificity was higher in women (P = .001). Sensitivity was higher for Asian (P = .002) and Black (P = .046) participants compared with White participants. The COVID-19 AI diagnostic system had worse accuracy (63.5% correct) compared with radiologist predictions (radiologist 1 = 67.8% correct, radiologist 2 = 68.6% correct; McNemar P < .001 for both). Conclusion: AI-based tools have not yet reached full diagnostic potential for COVID-19 and underperform compared with radiologist prediction.Keywords: Diagnosis, Classification, Application Domain, Infection, Lung Supplemental material is available for this article.. © RSNA, 2022.

The resting-state causal human connectome is characterized by hub connectivity of executive and attentional networks.

We demonstrate a data-driven approach for calculating a "causal connectome" of directed connectivity from resting-state fMRI data using a greedy adjacency search and pairwise non-Gaussian edge orientations. We used this approach to construct n = 442 causal connectomes. These connectomes were very sparse in comparison to typical Pearson correlation-based graphs (roughly 2.25% edge density) yet were fully connected in nearly all cases. Prominent highly connected hubs of the causal connectome were situated in attentional (dorsal attention) and executive (frontoparietal and cingulo-opercular) networks. These hub networks had distinctly different connectivity profiles: attentional networks shared incoming connections with sensory regions and outgoing connections with higher cognitive networks, while executive networks primarily connected to other higher cognitive networks and had a high degree of bidirected connectivity. Virtual lesion analyses accentuated these findings, demonstrating that attentional and executive hub networks are points of critical vulnerability in the human causal connectome. These data highlight the central role of attention and executive control networks in the human cortical connectome and set the stage for future applications of data-driven causal connectivity analysis in psychiatry.

Biased and inflexible interpretations of ambiguous social situations: Associations with eating disorder symptoms and socioemotional functioning.

BACKGROUND: Research indicates that difficulties across multiple socioemotional functioning domains (e.g., social emotion expression/regulation, response to social elicitors of emotion) and negatively biased interpretations of ambiguous social situations may affect eating disorder symptoms. The impact of inflexible interpretations of social situations on eating disorder symptoms is less clear. The present study therefore examined relations between inflexible and biased social interpretations, socioemotional functioning, and eating disorder symptoms. METHOD: A total of 310 participants from the general population, recruited from an online crowdsourcing platform, completed measures of socioemotional functioning (e.g., rejection sensitivity, negative social exchange), eating disorder symptoms, and positive and negative interpretation bias and inflexibility on a single measurement occasion. RESULTS: Socioemotional functioning impairments (Pillai's trace = 0.11, p < .001), but not negative (ß = .07, p = .162) or positive (ß = -.01, p = .804) interpretation bias or inflexible interpretations (ß = .04, p = .446), were associated with eating disorder symptoms in multiple regression models. In network analyses controlling statistically for multiple markers of socioemotional functioning, eating disorder symptoms were directly associated with negative (but not positive) interpretation bias. Inflexible interpretations were indirectly linked to symptoms via co-dampening of positive emotions. Exploratory causal discovery analyses suggested that several socioemotional functioning variables (social anxiety, depression, negative social exchange) may cause eating disorder symptoms. CONCLUSIONS: Consistent with cognitive-interpersonal models of disordered eating, our results suggest that less accurate (biased, inflexible) interpretations of social information contribute to patterns of cognition (anxious anticipation of rejection) and emotion regulation (down-regulation of positive social emotion) thought to encourage disordered eating. PUBLIC SIGNIFICANCE: This study suggests that less accurate interpretations of ambiguous social information encourage anxious anticipation of rejection and downregulation of positive social emotions, both of which are thought to promote eating disorder symptoms. Knowledge provided by this study about the likely relations between interpretive processes, social/emotional functioning, and eating disorder symptoms may help inform treatments for eating disorders, particularly those that attempt to modify patterns of interpretation.

Causal discovery replicates symptomatic and functional interrelations of posttraumatic stress across five patient populations.

Introduction: Approximately half of individuals with posttraumatic stress disorder (PTSD) may meet criteria for other psychiatric disorders, and PTSD symptoms are associated with diminished health and psychosocial functioning. However, few studies examine the longitudinal progression of PTSD symptoms concurrent with related symptom domains and functional outcomes, such that may neglect important longitudinal patterns of symptom progression beyond PTSD specifically. Methods: Therefore, we used longitudinal causal discovery analysis to examine the longitudinal interrelations among PTSD symptoms, depressive symptoms, substance abuse, and various other domains of functioning in five longitudinal cohorts representing veterans (n = 241), civilians seeking treatment for anxiety disorders (n = 79), civilian women seeking treatment for post-traumatic stress and substance abuse (n = 116), active duty military members assessed 0-90 days following TBI (n = 243), and civilians with a history of TBI (n = 43). Results: The analyses revealed consistent, directed associations from PTSD symptoms to depressive symptoms, independent longitudinal trajectories of substance use problems, and cascading indirect relations from PTSD symptoms to social functioning through depression as well as direct relations from PTSD symptoms to TBI outcomes. Discussion: Our findings suggest PTSD symptoms primarily drive depressive symptoms over time, tend to show independence from substance use symptoms, and may cascade into impairment in other domains. The results have implications for refining conceptualization of PTSD co-morbidity and can inform prognostic and treatment hypotheses about individuals experiencing PTSD symptoms along with co-occurring distress or impairment.

Willingness to vaccinate against SARS-CoV-2: The role of reasoning biases and conspiracist ideation.

BACKGR1OUND: Widespread vaccine hesitancy and refusal complicate containment of the SARS-CoV-2 pandemic. Extant research indicates that biased reasoning and conspiracist ideation discourage vaccination. However, causal pathways from these constructs to vaccine hesitancy and refusal remain underspecified, impeding efforts to intervene and increase vaccine uptake. METHOD: 554 participants who denied prior SARS-CoV-2 vaccination completed self-report measures of SARS-CoV-2 vaccine intentions, conspiracist ideation, and constructs from the Health Belief Model of medical decision-making (such as perceived vaccine dangerousness) along with tasks measuring reasoning biases (such as those concerning data gathering behavior). Cutting-edge machine learning algorithms (Greedy Fast Causal Inference) and psychometric network analysis were used to elucidate causal pathways to (and from) vaccine intentions. RESULTS: Results indicated that a bias toward reduced data gathering during reasoning may cause paranoia, increasing the perceived dangerousness of vaccines and thereby reducing willingness to vaccinate. Existing interventions that target data gathering and paranoia therefore hold promise for encouraging vaccination. Additionally, reduced willingness to vaccinate was identified as a likely cause of belief in conspiracy theories, subverting the common assumption that the opposite causal relation exists. Finally, perceived severity of SARS-CoV-2 infection and perceived vaccine dangerousness (but not effectiveness) were potential direct causes of willingness to vaccinate, providing partial support for the Health Belief Model's applicability to SARS-CoV-2 vaccine decisions. CONCLUSIONS: These insights significantly advance our understanding of the underpinnings of vaccine intentions and should scaffold efforts to prepare more effective interventions on hesitancy for deployment during future pandemics.

Heterogeneity in COVID-19 Patients at Multiple Levels of Granularity: From Biclusters to Clinical Interventions.

Several studies have shown that COVID-19 patients with prior comorbidities have a higher risk for adverse outcomes, resulting in a disproportionate impact on older adults and minorities that fit that profile. However, although there is considerable heterogeneity in the comorbidity profiles of these populations, not much is known about how prior comorbidities co-occur to form COVID-19 patient subgroups, and their implications for targeted care. Here we used bipartite networks to quantitatively and visually analyze heterogeneity in the comorbidity profiles of COVID-19 inpatients, based on electronic health records from 12 hospitals and 60 clinics in the greater Minneapolis region. This approach enabled the analysis and interpretation of heterogeneity at three levels of granularity (cohort, subgroup, and patient), each of which enabled clinicians to rapidly translate the results into the design of clinical interventions. We discuss future extensions of the multigranular heterogeneity framework, and conclude by exploring how the framework could be used to analyze other biomedical phenomena including symptom clusters and molecular phenotypes, with the goal of accelerating translation to targeted clinical care.

A Prospective Observational Study to Investigate Performance of a Chest X-ray Artificial Intelligence Diagnostic Support Tool Across 12 U.S. Hospitals.

Importance: An artificial intelligence (AI)-based model to predict COVID-19 likelihood from chest x-ray (CXR) findings can serve as an important adjunct to accelerate immediate clinical decision making and improve clinical decision making. Despite significant efforts, many limitations and biases exist in previously developed AI diagnostic models for COVID-19. Utilizing a large set of local and international CXR images, we developed an AI model with high performance on temporal and external validation. Objective: Investigate real-time performance of an AI-enabled COVID-19 diagnostic support system across a 12-hospital system. Design: Prospective observational study. Setting: Labeled frontal CXR images (samples of COVID-19 and non-COVID-19) from the M Health Fairview (Minnesota, USA), Valencian Region Medical ImageBank (Spain), MIMIC-CXR, Open-I 2013 Chest X-ray Collection, GitHub COVID-19 Image Data Collection (International), Indiana University (Indiana, USA), and Emory University (Georgia, USA). Participants: Internal (training, temporal, and real-time validation): 51,592 CXRs; Public: 27,424 CXRs; External (Indiana University): 10,002 CXRs; External (Emory University): 2002 CXRs. Main Outcome and Measure: Model performance assessed via receiver operating characteristic (ROC), Precision-Recall curves, and F1 score. Results: Patients that were COVID-19 positive had significantly higher COVID-19 Diagnostic Scores (median .1 [IQR: 0.0-0.8] vs median 0.0 [IQR: 0.0-0.1], p < 0.001) than patients that were COVID-19 negative. Pre-implementation the AI-model performed well on temporal validation (AUROC 0.8) and external validation (AUROC 0.76 at Indiana U, AUROC 0.72 at Emory U). The model was noted to have unrealistic performance (AUROC > 0.95) using publicly available databases. Real-time model performance was unchanged over 19 weeks of implementation (AUROC 0.70). On subgroup analysis, the model had improved discrimination for patients with "severe" as compared to "mild or moderate" disease, p < 0.001. Model performance was highest in Asians and lowest in whites and similar between males and females. Conclusions and Relevance: AI-based diagnostic tools may serve as an adjunct, but not replacement, for clinical decision support of COVID-19 diagnosis, which largely hinges on exposure history, signs, and symptoms. While AI-based tools have not yet reached full diagnostic potential in COVID-19, they may still offer valuable information to clinicians taken into consideration along with clinical signs and symptoms.

Neighborhood characteristics and psychiatric disorders in the aftermath of mass trauma: A representative study of New York City public school 4th-12th graders after 9/11.

Studies of the relationship between neighborhood characteristics and childhood/adolescent psychopathology in large samples examined one outcome only, and/or general (e.g., 'psychological distress') or aggregate (e.g., 'any anxiety disorder') measures of psychopathology. Thus, in the only representative sample of New York City public school 4th-12th graders (N = 8202) surveyed after the attacks of 9/11/2001, this study examined whether (1) indices of neighborhood Socioeconomic Status, Quality, and Safety and (2) neighborhood disadvantage (defined as multidimensional combinations of SES, Quality and Safety indicators) are associated with eight psychiatric disorders: posttraumatic stress disorder, separation anxiety disorder (SAD), agoraphobia, generalized anxiety disorder (GAD), panic disorder, major depression, conduct disorder, and alcohol use disorder (AUD). (1) The odds ratios (OR) of psychiatric disorders were between 0.55 (AUD) and 1.55 (agoraphobia), in low and intermediate-low SES neighborhoods, respectively, between 0.50 (AUD) and 2.54 (agoraphobia) in low Quality neighborhoods, and between 0.52 (agoraphobia) and 0.65 (SAD) in low Safety neighborhoods. (2) In neighborhoods characterized by high disadvantage, the OR were between 0.42 (AUD) and 1.36 (SAD). This study suggests that neighborhood factors are important social determinants of childhood/adolescent psychopathology, even in the aftermath of mass trauma. At the community level, interventions on modifiable neighborhood characteristics and targeted resources allocation to high-risk contexts could have a cost-effective broad impact on children's mental health. At the individual-level, increased knowledge of the living environment during psychiatric assessment and treatment could improve mental health outcomes; for example, specific questions about neighborhood factors could be incorporated in DSM-5's Cultural Formulation Interview.

A simple interpretation of undirected edges in essential graphs is wrong.

Artificial intelligence for causal discovery frequently uses Markov equivalence classes of directed acyclic graphs, graphically represented as essential graphs, as a way of representing uncertainty in causal directionality. There has been confusion regarding how to interpret undirected edges in essential graphs, however. In particular, experts and non-experts both have difficulty quantifying the likelihood of uncertain causal arrows being pointed in one direction or another. A simple interpretation of undirected edges treats them as having equal odds of being oriented in either direction, but I show in this paper that any agent interpreting undirected edges in this simple way can be Dutch booked. In other words, I can construct a set of bets that appears rational for the users of the simple interpretation to accept, but for which in all possible outcomes they lose money. I put forward another interpretation, prove this interpretation leads to a bet-taking strategy that is sufficient to avoid all Dutch books of this kind, and conjecture that this strategy is also necessary for avoiding such Dutch books. Finally, I demonstrate that undirected edges that are more likely to be oriented in one direction than the other are common in graphs with 4 nodes and 3 edges.

An integrated multimodal model of alcohol use disorder generated by data-driven causal discovery analysis.

Alcohol use disorder (AUD) has high prevalence and adverse societal impacts, but our understanding of the factors driving AUD is hampered by a lack of studies that describe the complex neurobehavioral mechanisms driving AUD. We analyzed causal pathways to AUD severity using Causal Discovery Analysis (CDA) with data from the Human Connectome Project (HCP; n = 926 [54% female], 22% AUD [37% female]). We applied exploratory factor analysis to parse the wide HCP phenotypic space (100 measures) into 18 underlying domains, and we assessed functional connectivity within 12 resting-state brain networks. We then employed data-driven CDA to generate a causal model relating phenotypic factors, fMRI network connectivity, and AUD symptom severity, which highlighted a limited set of causes of AUD. The model proposed a hierarchy with causal influence propagating from brain connectivity to cognition (fluid/crystalized cognition, language/math ability, & working memory) to social (agreeableness/social support) to affective/psychiatric function (negative affect, low conscientiousness/attention, externalizing symptoms) and ultimately AUD severity. Our data-driven model confirmed hypothesized influences of cognitive and affective factors on AUD, while underscoring that addiction models need to be expanded to highlight the importance of social factors, amongst others.