Longitudinal Dynamics and Pluripotentiality of Polysymptomatic Clustering in Adolescent Mental Health.

Abstract

Background: Adolescence represents a sensitive developmental period characterised by an increased incidence of emerging mental health symptoms and formal diagnostic onset. These conditions can remain a significant burden throughout life. The Longitudinal Adolescent Brain Study (LABS) commenced in 2018 to track the onset and trajectory of mental health symptoms among general population participants. This research aims to identify polysymptomatic clusters of emerging mental health symptoms in adolescents and examine how these clusters vary by age and change over time, providing insights into the pluripotentiality of disorder development. Methods: LABS participants aged 12-17 years (n=166) completed the Mini International Neuropsychiatric Interview (MINI Kid) approximately every 4 months, with up to 15 timepoints. Due to this high dimensional dataset, the data was first processed using a dimensionality reduction step (uniform manifold approximation and projection; UMAP). Following this, the data was clustered using Bayesian model averaging of k-means, gaussian mixture model and hierarchical clustering to identify distinct symptom clusters. Symptom clusters were described in terms of the original neuropsychiatric interview responses using separate XGBoost classifier models. Symptom cluster dynamics were analysed using Markov chain transition probability matrices and longitudinal analysis. To explore the relationship between symptom clusters and psychological distress and wellbeing, correlational analyses were conducted using scores from the Kessler Psychological Distress Scale (K10) and the COMPAS-W Wellbeing Scale. Outcomes: Six symptom-based clusters (states) were identified: attention, anxiety, depression, manic episode - heritability, anhedonia, and well. Depression and anxiety clusters had the greatest pluripotentiality. Analysis of psychological distress and wellbeing demonstrated an inverse relationship between the states: those with greater psychological distress had more symptoms, conversely those with greater wellbeing had fewer symptoms. Interpretations: Mapping polysymptomatic clusters of mental health symptoms and their pluripotential and transitory trajectories in adolescents enables more effective targeting of preventive interventions. This approach moves beyond categorical classifications to mitigate the progression of early symptoms into enduring psychiatric disorders.