Clinical Report: The Potential of AI-Driven Speech Biomarkers in Mental Health Assessment
Overview
This study explores the use of AI-driven speech biomarkers to assess psychotic symptom severity in individuals with schizophrenia spectrum disorders. Findings indicate that speech features can provide estimations of symptom severity.
Background
Speech analysis has long been recognized as a valuable tool for understanding mental states in psychosis. Recent advancements in computational speech analysis have opened new avenues for tracking symptom fluctuations. This study investigates the relationship between speech features and psychotic symptoms in diverse cohorts.
Data Highlights
Cohort
Symptom Scale
Mean Absolute Error (MAE)
Dutch
Positive Symptoms
2.85
Dutch
Negative Symptoms
3.22
US
Thought Disturbance
3.00
US
Withdrawal
2.44
Key Findings
The study analyzed 773 brief speech recordings from 356 participants with schizophrenia spectrum disorders.
Automated speech analysis extracted over 100 features, including acoustic, semantic, syntactic, and sentiment markers.
Longer utterances and altered discourse organization were associated with positive symptoms.
Negative symptoms correlated with reduced speech output and flatter acoustic profiles.
Mean absolute errors for symptom severity estimations were clinically meaningful, ranging from 2.44 to 3.22.
Clinical Implications
The findings indicate that AI-driven speech analysis could facilitate monitoring of psychotic symptoms.
Conclusion
The study demonstrates the feasibility of using speech as a biosocial marker for psychosis, emphasizing the need for further research to ensure the generalizability and interpretability of these findings in clinical settings.
