Claire E. Cusack

Set-theoretic configural psychometrics identifies minimally sufficient conditions (concurrent state combinations) for clinical targets. We tested whether intraindividual rules derived from early ecological momentary assessment (EMA; first 60%) consistently account for future suicidal ideation and self-injury (held out 40%), and whether engine performance profiles can be preregistered. Across discovery (N = 46) and replication (N = 54) samples of recently suicidal trans and nonbinary adults, we evaluated constructive and exhaustive engines across four targets and tested nine preregistered behavioral criteria. Person-specific rules replicated forward in time (median replication .65–1.00). The constructive engine achieved higher specificity, whereas the exhaustive engine yielded higher sensitivity, coverage, and discrimination. Up to eight of nine preregistered criteria were confirmed, with replication strongly graded by target base rates. Configural psychometrics recovers valid intraindividual rules for future observations. Its performance is predictable and highest for chronically, rather than episodically, symptomatic individuals.

Scientific research is a high-impact experience for undergraduate students. While there are many benefits of involving students in research programs, mentored research models cannot accommodate all interested students due to limited time and bandwidth of faculty mentors. Additionally, these experiences often cannot be accessed by students from marginalized backgrounds, amplifying issues of inequity in the sciences. In response to these limitations, other models of student research engagement such as course-based undergraduate research experiences (CUREs) have been widely implemented. Although CUREs allow for greater scalability than one-on-one mentored research approaches, they fail to engage students in research activities in a sustained way that leads to higher persistence in STEM. Below, we first outline the benefits and problems associated with both mentored research and CURE models. We then propose Structured Team-based Inquiry in Labs (STIL)—an alternative strategy for researchers interested in providing undergrads with research opportunities. This hybrid approach bridges traditional mentored research and CURE-based opportunities to maximize the benefits of both approaches. To illustrate this model, we present two case studies in a microbial ecology and a clinical psychology research laboratory, showcasing the interdisciplinary implementation potential of the STIL model.

Network analysis is a popular method researchers use to characterize the structure of psychopathology and inform personalized treatments. Typically, applied researchers, based on network theory, interpret symptoms with the highest strength centrality as most important to network structure and represent amenable treatment targets. This study examines the stability of strength centrality in idiographic networks in a sample of participants with eating disorders (N = 26, 90-day assessment, M = 356.00 observations per person) and a second sample of participants with social anxiety disorder (N = 42, 30-day assessment, M = 201.90 observations per person). We estimated idiographic networks using three different item-inclusion approaches and accounted for time using a "sliding window" method (e.g., Window 1 = data from Days 1-15, Window 2 = data from Days 2-16). Items included in networks were selected in three ways: default networks (six items with the highest means at Window 1), changing means networks (six items with the highest means at each respective Window), and random ensembles (random combinations of any six items across all sliding windows). In both samples, we found that the most central symptom in the default network was central in less than half of idiographic changing means networks (maximum = 29.41% of networks). Our results show that node strength centrality estimates are sensitive to item ensemble and temporal effects. We discuss implications concerning inferences assigned to strength centrality given the frequency at which strength centrality changes and future efforts developing network-informed personalized treatment. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Objective: Psychological treatment effects and response rates have largely plateaued over the past few decades. A potential answer to this problem is personalized treatment approaches, which match treatment to a client’s specific presenting concerns, which should increase its precision and efficacy. Examining predictors and moderators of treatment outcome (who is likely to benefit from treatment, and which treatment) is one way to guide such personalized decision making. The current study aimed to investigate the relationship between network-derived subgroups, treatment condition, and treatment response using data from two clinical trials on eating disorders (N = 80). This study is also a proof-of-concept for the clinical utility of network-based subgroups. Method: Subgroups were identified using Subgrouping Group Iterative Multiple Model Estimation (S-GIMME) and ANOVAs were used to compare changes in symptom severity and clinical impairment among subgroups. Results: We found three subgroups (n = 71; mean age = 34.4 [SD = 11.8], 87.3% cisgender women, 85.9% white, non-Hispanic), which were differentiated by how shame and guilt were related in the network. The subgroup with a contemporaneous pathway from guilt to shame showed the least improvement in clinical impairment from pre- to posttreatment (F(2, 64) = 5.92, p = .004). Conclusions: Overall, our findings suggest that network-based subgroups may have utility as prognostic indicators, at least in the context of eating disorders, though replication of present findings is warranted. Limitations included potentially unstable subgroups and use of mostly cisgender women and white samples.

Individuals with eating disorders present with heterogenous cognitive-affective symptom profiles, whereas behavioral symptoms tend to be more homogenous. For example, dietary restriction is a common behavioral symptom across the eating disorders; however, the cognitive-affective factors that predict restriction may vary among persons. The aim of this study was to examine which variables are associated with dietary restriction in group-level models of eating disorder pathology and describe how these symptoms compare to the associations evidenced in person-specific models. We collected time-intensive longitudinal data from 26 participants with eating disorders. Participants provided a mean of 355.92 observations across a 90-day period. We estimated group-level models using group iterative multiple model estimation including nodes with the six highest means for the sample. We estimated idiographic (N=1) models by using regularized graphical vector autoregressive models including nodes with the six highest means for each participant. Nodes included in the group-level model were anxiety, feeling fat, fear of weight gain, overevaluation of weight and shape, drive for thinness, and the urge to restrict. In the group-level model, aside from the autoregressive relationship, fear of weight gain was the only variable associated with the urge to restrict. In idiographic models, the urge to body check was the most common variable associated with the urge to restrict, followed by anxiety and feeling fat. Findings from the group-level model were rarely maintained at the individual level. Idiographic models may provide more tailored conceptualizations of eating disorder symptom dynamics related to dietary restriction relative to group-based models.Keywords: eating disorders, dietary restriction, idiographic, network analysis, anorexia nervosa, bulimia nervosa, OSFED

Objective: Dysregulated eating is common among youth and is associated with trait‐level negative affect and emotion regulation difficulties. Despite the transient nature of affect, momentary associations among affect and eating behavior are unclear, which limits development of more impactful treatment tools, such as “just‐in‐time” intervention approaches (JITAI). The current study (N = 62) drew from two ecological momentary assessment (EMA) studies involving children and adolescents who endorsed loss of control (LOC) eating symptoms during a two‐week assessment period.Method: Intensive time series network analysis tested concurrent and prospective relationships across six specific affective states (i.e., upset, guilty, scared, tired, excited, attentive) and four eating‐related experiences (e.g., LOC, overeating, hunger, craving) in real time. Additionally, we repeated these models within demographic subgroups of the sample based on age, race, and sex.Results: In the full‐sample models, contemporaneously assessed guilt was associated with craving and LOC eating, and tiredness was associated with LOC eating. In the prospective analysis, tiredness was negatively predicted by LOC eating and positively predicted by overeating at the previous timepoint, and attentiveness positively predicted craving. Differences in affect‐eating relationships were identified across teens and preteens as well as male and female participants.Discussion: These results suggest that specific affective states are associated with dysregulated eating‐related experiences in real time among youth, and associations may differ depending on demographic characteristics. Findings may be used to inform the development and tailoring of momentary interventions.

Objective: Body mass index (BMI) is the primary criterion differentiating anorexia nervosa (AN) and atypical anorexia nervosa despite prior literature indicating few differences between disorders. Machine learning (ML) classification provides us an efficient means of accurately distinguishing between two meaningful classes given any number of features. The aim of the present study was to determine if ML algorithms can accurately distinguish AN and atypical AN given an ensemble of features excluding BMI, and if not, if the inclusion of BMI enables ML to accurately classify between the two.Methods: Using an aggregate sample from seven studies consisting of individuals with AN and atypical AN who completed baseline questionnaires (N = 448), we used logistic regression, decision tree, and random forest ML classification models each trained on two datasets, one containing demographic, eating disorder, and comorbid features without BMI, and one retaining all features and BMI.Results: Model performance for all algorithms trained with BMI as a feature was deemed acceptable (mean accuracy = 74.98%, mean area under the receiving operating characteristics curve [AUC] = 74.75%), whereas model performance diminished without BMI (mean accuracy = 59.37%, mean AUC = 59.98%).Discussion: Model performance was acceptable, but not strong, if BMI was included as a feature; no other features meaningfully improved classification. When BMI was excluded, ML algorithms performed poorly at classifying cases of AN and atypical AN when considering other demographic and clinical characteristics. Results suggest a reconceptualization of atypical AN should be considered.Public Significance: There is a growing debate about the differences between anorexia nervosa and atypical anorexia nervosa as their diagnostic differentiation relies on BMI despite being similar otherwise. We aimed to see if machine learning could distinguish between the two disorders and found accurate classification only if BMI was used as a feature. This finding calls into question the need to differentiate between the two disorders.

Item selection is a critical decision in modeling psychological networks. The current preregistered two-study research used random selections of 1,000 symptom networks to examine which eating disorder (ED) and co-occurring symptoms are most central in longitudinal networks among individuals with EDs (N = 71, total observations = 6,060) and tested whether centrality changed based on which items were included in the network. Participants completed 2 weeks of ecological momentary assessment (five surveys/day). In Study 1, we obtained initial strength centrality values by estimating an a priori network using eight items with the highest means. We then estimated 1,000 networks and their centrality from a random selection of unique eight-item symptom combinations. We compared the strength centrality from the a priori network to the distribution of strength centrality estimates from the random-item networks. In Study 2, we repeated this procedure in an independent longitudinal dataset (N = 41, total observations = 4,575) to determine if our results generalized across samples. Shame, guilt, worry, and fear of losing control were consistently central across networks, regardless of items included in the network or sample. Results suggest that these symptoms may be important to the structure of ED psychopathology and have implications for how we understand the structure of ED psychopathology. Existing methods for item inclusion in psychological networks may distort the structure of ED symptom networks by either under- or overestimating strength centrality, or by omitting consistently central symptoms that are nontraditional ED symptoms. Future research should consider including these symptoms in models of ED psychopathology. (PsycInfo Database Record (c) 2023 APA, all rights reserved).