Automated, Objective Speech and Language Markers of Longitudinal Changes in Psychosis Symptoms
Tang, S. X.; Spilka, M. J.; John, M.; Birnbaum, M. L.; Saito, E.; Berretta, S. A.; Behbehani, L. M.; Liberman, M. Y.; Malhotra, A. K.; Simpson, W.; Kane, J. M.
Show abstract
Background and HypothesesWe sought to evaluate the ability of automated speech and language features to track fluctuations in the major psychosis symptoms domains: Thought Disorder, Negative Symptoms, and Positive Symptoms. Study DesignSixty-six participants with psychotic disorders were longitudinally assessed soon after inpatient admission, at discharge, and at 3- and 6-months. Psychosis symptoms were measured with semi-structured interviews and standardized scales. Recordings were collected from paragraph reading, fluency, picture description, and open-ended tasks. Longitudinal relationships between psychosis symptoms and 357 automated speech and language features were analyzed using a single component score and as individual features, using linear mixed models. Study ResultsAll three psychosis symptom domains demonstrated significant longitudinal relationships with the single component score. Thought Disorder was particularly related to features describing more subordinated constructions, less efficient identification of picture elements, and decreased semantic distance between sentences. Negative Symptoms was related to features describing decreased speech complexity. Positive Symptoms appeared heterogeneous, with Suspiciousness relating to greater use of nouns, and Hallucinations related to decreased semantic distances. These relationships were largely robust to interactions with gender and race. However, interactions with timepoint revealed variable relationships during different phases of illness (acute vs. stable). ConclusionsAutomated speech and language features show promise as scalable, objective markers of psychosis severity. The three symptom domains appear to be distinguishable with different features. Detailed attention to clinical setting and patient population is needed to optimize clinical translation; there are substantial implications for facilitating differential diagnosis, improving psychosis outcomes and enhancing therapeutic discovery.