Predicting individual clinical trajectories of depression with generative embedding

Abstract

ABSTRACTPatients with major depressive disorder (MDD) show heterogeneous treatment response and highly variable clinical trajectories: while some patients experience swift and enduring recovery, others show relapsing-remitting or chronic disease course. Predicting individual clinical trajectories at an early disease stage is a key challenge for psychiatry and might facilitate individually tailored interventions. So far, however, reliable predictors at the single-patient level are absent.Here, we evaluated the utility of a machine learning strategy – generative embedding – which combines an interpretable generative model with a discriminative classifier. Specifically, we used functional magnetic resonance imaging (fMRI) data of emotional face perception in 85 MDD patients from the multi-site longitudinal NEtherlands Study of Depression and Anxiety (NESDA) who had been followed up over two years and classified into three subgroups with distinct clinical trajectories. Combining a generative model of effective (directed) connectivity with support vector machines (SVMs), it was possible to predict whether a given patient will experience chronic depression vs. fast remission with a balanced accuracy of 79%. Gradual improvement vs. fast remission could still be predicted above-chance, but less convincingly, with a balanced accuracy of 61%. Importantly, generative embedding outperformed conventional (descriptive) measures such as functional connectivity or local BOLD activity, which did not predict clinical trajectories with above-chance accuracy. Furthermore, the predictive performance of generative embedding could be assigned to a specific network property: the dynamic modulation of connections by the emotional content of the trial-by-trial stimuli. Our findings suggest that a mechanistically informed generative model of a neuronal circuit underlying emotional face perception may have predictive utility for distinguishing disease courses in MDD patients.