Contribution of physiological dynamics in predicting major depressive disorder severity.

Abstract

This study aimed to explore the physiological dynamics of cognitive stress in patients with Major Depressive Disorder (MDD) and design a multiparametric model for objectively measuring severity of depression. Physiological signal recordings from 40 MDD patients and 40 healthy controls were collected in a baseline stage, in a stress-inducing stage using two cognitive tests, and in the recovery period. Several features were extracted from electrocardiography, photoplethysmography, electrodermal activity, respiration, and temperature. Differences between values of these features under different conditions were used as indexes of autonomic reactivity and recovery. Finally, a linear model was designed to assess MDD severity, using the Beck Depression Inventory scores as the outcome variable. The performance of this model was assessed using the MDD condition as the response variable. General physiological hyporeactivity and poor recovery from stress predict depression severity across all physiological signals except for respiration. The model to predict depression severity included gender, body mass index, cognitive scores, and mean heart rate recovery, and achieved an accuracy of 78%, a sensitivity of 97% and a specificity of 59%. There is an observed correlation between the behavior of the autonomic nervous system, assessed through physiological signals analysis, and depression severity. Our findings demonstrated that decreased autonomic reactivity and recovery are linked with an increased level of depression. Quantifying the stress response together with a cognitive evaluation and personalization variables may facilitate a more precise diagnosis and monitoring of depression, enabling the tailoring of therapeutic interventions to individual patient needs.