Mapping the interoceptive-internalising landscape: a network analysis and person-centred clustering of multidimensional interoceptive awareness in psychopathology

Background: Interoceptive awareness is increasingly conceptualised as a multidimensional brain-body process relevant to emotional regulation and internalising psychopathology. However, it remains unclear how interoceptive dimensions are embedded within the multivariate structure of depressive, anxiety, and dissociative symptoms, and whether distinct interoceptive-symptom profiles can be identified at the individual level.

Aim: This study aimed to characterise the network structure linking interoceptive awareness and internalising symptoms and to identify person-centred profiles reflecting heterogeneous constellations of interoceptive functioning and symptom severity.

Methods: Archival intake data from 475 adults who had applied to Koç University Psychological Counseling and Psychotherapy Center (KUPTEM), which is a comprehensive mental health clinic that provides various clinical services, including individual psychotherapy, psychiatric consultation, and treatment. Interoceptive awareness was assessed using the Turkish Multidimensional Assessment of Interoceptive Awareness, Version 2, and internalising symptoms using the Patient Health Questionnaire-9, Generalised Anxiety Disorder-7, in addition to the Dissociative Experiences Scale-Brief. Bivariate associations between interoceptive subscales and symptom measures were first examined using Spearman correlations with false discovery rate correction. Network analysis was conducted within a Gaussian graphical model framework using extended Bayesian information criterion graphical least absolute shrinkage and selection operator regularisation (γ = 0.25) based on polychoric and polyserial correlations. Nodes included interoceptive subscales and individual symptom items. Node importance was evaluated using strength and expected influence, and bridge centrality was calculated to identify interoceptive domains linking symptom communities. Network accuracy and stability were assessed using bootstrap procedures. Person-centred heterogeneity was examined using k-means clustering on standardised interoceptive and symptom variables. Candidate solutions (k = 3-7) were evaluated using silhouette coefficients, Davies–Bouldin indices, inertia, and stability across repeated initialisations, with model selection based on combined statistical and interpretability criteria.

Results: The network showed a sparse and well-regularised structure in which depressive, anxiety, and dissociative symptoms formed densely connected communities, while interoceptive domains occupied peripheral but systematically integrated positions. Regulatory interoceptive facets, including attention regulation, self-regulation, and trusting, demonstrated higher centrality, whereas not-worrying showed a prominent negative expected influence, reflecting inverse associations with anxiety-related symptoms. Bridge centrality analyses identified not-worrying and trusting as key connectors linking interoceptive domains to internalising symptom communities. Person-centred clustering identified six distinct subgroups characterised by differential interoceptive-symptom configurations (n = 63-116). The k = 5 solution provided a more granular representation of heterogeneity compared to lower-dimensional partitions. Stability analyses indicated moderate reproducibility (bootstrap ARI ≈ 0.508), with acceptable consistency under initialization sensitivity (mean Jaccard ≈ 0.41) and high agreement when extensive optimisation was applied. The identified clusters differentiated profiles characterised by high symptom burden and reduced regulatory interoceptive capacity, low symptom burden with preserved regulatory functioning, and intermediate patterns reflecting selective impairments in specific interoceptive domains.

Conclusions: Interoceptive awareness is differentially embedded within internalising symptom networks, with regulatory facets acting as key bridging and potentially protective components. The identification of distinct interoceptive symptom profiles supports a person-centred perspective on interoception in psychopathology. These findings highlight interoceptive regulation as a potential target for applied neuroscience models of emotional disorders. Longitudinal and multimodal studies are needed to clarify causal pathways.

References

No references