Metastability from recurrence analysis in depression

Abstract

Abstract This study investigates recurrent patterns in Ecological Momentary Assessment (EMA) data from a single-case longitudinal study (239 days) of a patient diagnosed with major depressive disorder undergoing antidepressant dosage adjustments. We construct a dynamic network based on the synchrony in the rate of change of questionnaire items. Furthermore, we quantify quasi-stationary regions—defined as metastates —for individual items using recurrence quantification analysis and Markovian optimization. Our results show that an abrupt increase in the average degree of the network, together with a decrease in the correlation between node degree and the number of metastable states, precedes the depressive transition identified by the mean SCL-90 score. These results highlight the potential of Recurrence Plots to map the complex landscape of psychological states, providing a robust framework for quantifying stability and change in high-frequency longitudinal assessments.