Answers are grounded in this paper and cite their sources.
Automated speech analysis offers a promising avenue for objective assessment in schizophrenia by capturing its core feature of disorganized thinking, but the reliability of speech markers across different conversational contexts remains a critical unknown. We hypothesized that more complex speech elicitation tasks would lead to larger patient-control differences. We collected speech samples from patients (diagnosed with schizophrenia-spectrum disorders) and controls using tasks of varying complexity: reading, storyboard, and personal narrative. We extracted four speech markers known to be linked to schizophrenia and applied linear mixed-effects models to examine within-task stability and task effects. Results indicated that while speech markers are stable within tasks, significant group differences emerge specifically in the more cognitively demanding personal narrative and storyboard tasks, but not in the simpler reading task. Thus, it appears that clinical utility of routinely assessed automated speech variables in psychosis is task dependent. Our results provide evidence to guide the selection of sufficiently complex elicitation tasks for both research and applied clinical tool development.
No takes yet. Share an insight, caveat, or question.
Mouslih et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: