A framework for parsing psychopathological heterogeneity: initial application in a large-scale unselected community sample.

Abstract

Background: Traditional descriptive nosology arbitrarily distinguishes between mental illness and health, hindering the progress of scientific research and clinical practice. Building on recent advancements in psychiatric conceptualization, this study proposes an innovative phased framework for deconstructing psychopathological heterogeneity. The framework involves four key steps: extraction of symptom dimensions, identification of psychopathological subtypes, characterization of symptom interaction patterns using a network approach, and validation of their incremental validity through links to neurobehavioral functions. This framework is preliminarily applied to a large, non-selective community sample (N = 4102) to explore its utility and potential for deconstructing psychopathological heterogeneity.

Methods: Data on comprehensive psychopathology and RDoC negative valence constructs were collected from the sample. Factor analysis and exploratory graph analysis were used to extract symptom dimensions. Latent profile analysis based on these dimensions was applied to identify psychopathological profiles. Partial correlation networks were estimated for each profile, and symptom network characteristics were compared across profiles. Finally, hierarchical multiple regression was applied to assess incremental validity.

Results: The first step of the phased framework involves extracting homogeneous dimensions based on symptom co-occurrence patterns, yielding seven distinct dimensions: Obsessive-Compulsive, Emotional Distress, Eating-Related, Substance-Related, Aggressive, Psychotic, and Somatoform dimensions. The second step involves applying a person-centered approach to identify latent subgroups based on these symptom dimensions. Four profiles were identified, namely Substance Use Group, Moderate Symptomatology Group, Disengaged from Symptomatology Group, and Severe Symptomatology Group. The third step involves characterizing symptom interaction patterns across subgroups. Using a network approach, the Severe Symptomatology Group exhibited the densest interconnections and the highest global network strength, with Aggressive and Psychotic dimensions serving as core issues compared to other profiles. Finally, incremental validity was assessed through associations with self-reported neurobehavioral functions. Results showed that these profiles provided unique predictive value for RDoC negative valence constructs beyond both dichotomous diagnostic status and purely dimensional approach.

Conclusions: This study introduces a fine-grained framework for deconstructing psychopathological heterogeneity, providing a comprehensive approach to parsing psychopathology. While the framework is preliminarily applied to a large sample from the Chinese population, future studies should integrate multimodal neurobehavioral measures, employ intensive longitudinal designs to track symptom dynamics, and validate its consistency across diverse cultural and regional contexts.