Identification of Neural Biomarkers of Major Depressive Disorder Symptom Severity Using Computerized Linguistic Analysis

Abstract

Although numerous treatments are available for major depressive disorder (MDD), patients can be refractory to sequential treatment regimens. Experimental studies have demonstrated promising results implementing deep brain stimulation (DBS) as a therapy for treatment resistant MDD. However, optimization of this technique requires repeated assessments of the clinical effects of treatment in each patient and the ability to reliably capture the complexity and dynamics of depression symptoms. In our initial studies evaluating the feasibility and preliminary efficacy of a novel closed-loop DBS (CL-DBS) approach, we have observed that repeated self-rated MDD metrics can be burdensome to complete and may not provide accurate measures of symptom severity fluctuations over time, making the identification of neural biomarkers of MDD a challenge. To address this, we evaluated if text analysis could identify linguistic indicators of depression, including providing insights into symptom severity. Using the Linguistic Inquiry and Word Count software, we analyzed written symptom reports from one patient in clinical trial for CL-DBS. We found significant linguistic predictors of depression symptoms that were associated with the same frequency- and region- specific spectral power correlates found when assessing symptoms captured by self-rated depression metrics. These preliminary findings suggest that the close association between language use and symptom strength could be utilized to detect neural biomarkers of depression and potentially to assess treatment outcome.