Understanding the Core Structure and Heterogeneity of Eating Disorder Symptoms: A Group Iterative Multiple Model Estimation (GIMME) Approach

Eating disorders (EDs) are a critical public health concern, with the second-highest mortality rate out of any mental health disorder. In the US, 10,200 deaths each year are the direct result of an ED, equating to a death every 52 minutes (ANAD, 2023). This represents an essential area of focus to determine ED maintenance and inform novel treatments to improve ED recovery rates. ED network research demonstrates that there is significant heterogeneity in ED symptoms at an individual level. An idiographic approach is critical to understanding individual differences and informing treatment improvements for individuals who may not be represented by group-level analyses. While network analysis has been increasingly applied to characterize the structure of eating disorder psychopathology and to identify central symptoms, networks are usually analyzed at the group level or at the individual level, rather than looking at both levels simultaneously, whereas GIMME gives us this capability. GIMME is a type of modeling technique that can integrate complex data to model both group and within person variance. We can use these analyses to understand how ED symptoms maintain each other over time and for whom relationships among symptoms exist. These are questions that are critical to improving remission rates and treatment outcomes. GIMME is a hierarchical discovery algorithm developed for person-specific time-series models. It leverages group-level information to develop individual models. This type of model allows us to see connections at the group and individual levels. Both contemporaneous and lagged effects will be identified in the model. 1. Aim 1: Identify group-level and individual-level symptom maintenance pathways within eating disorder symptom networks using GIMME. a. This aim focuses on distinguishing symptom dynamics that are shared across individuals from those that reflect meaningful heterogeneity in maintenance processes. With this aim, we will determine where ED symptom structures vary within and between individuals, highlighting which symptom structures are most common and how they change over time. 2. Aim 2 : Evaluate cross-domain maintenance processes linking cognitive, affective, and behavioral symptoms. a. This aim tests whether ED maintenance is driven primarily by cognitive–behavioral pathways, affect-driven pathways, or behavioral feedback loops, and whether these processes differ across individuals. 3. Aim 3: Characterize temporal directionality and centrality of eating disorder symptoms within individuals. a. This aim examines which symptoms temporally precede changes in other symptoms and which symptoms function as dynamic drivers of symptom change over time. To our knowledge, no peer-reviewed studies have yet applied GIMME to core ED symptom time-series data to simultaneously estimate group- and individual-level temporal networks, allowing us to determine directionality, temporal precedence, within-person processes, and heterogeneity of symptom maintenance. Including both direction and time in our models allows us to determine possible maintenance mechanisms. This project thus represents a novel extension of network psychopathology research in EDs with the potential to reveal dynamic mechanisms of symptom maintenance and heterogeneity in individual symptom interactions. This has the potential to generate clinically actionable insights regarding which symptoms drive ED maintenance for whom and when. Eating Disorder Statistics | ANAD - National Association of Anorexia Nervosa and Associated Disorders. (2023, November 29). https://anad.org/eating-disorder-statistic/