[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Reading Between the Lines: Combining Pause Dynamics and Semantic Coherence for Automated Assessment of Thought Disorder

Created by
  • Haebom

Author

Feng Chen, Weizhe Xu, Changye Li, Serguei Pakhomov, Alex Cohen, Simran Bhola, Sandy Yin, Sunny X Tang, Michael Mackinley, Lena Palaniyappan, Dror Ben-Zeev, Trevor Cohen

Outline

This study utilized automatic speech recognition (ASR) technology for objective and scalable assessment of formal thought disorder (FTD), a core symptom of schizophrenia spectrum disorder. In order to overcome the limitations of existing clinical assessment scales, linguistic and temporal features of speech obtained through ASR, especially pause movements, were analyzed and utilized to predict FTD severity. Using three datasets (natural self-recorded diaries, structured picture descriptions, and dream stories), support vector regression (SVR) analysis was performed by combining pause-related features and existing semantic consistency measures. As a result, it was confirmed that pause features alone could strongly predict FTD severity, and the model that integrated pause features and semantic consistency measures showed better prediction performance than the model that only considered semantics (maximum correlation coefficient ρ = 0.649, AUC = 83.71%). These results suggest that a framework that combines temporal and semantic analysis can improve the assessment of disorganized language and contribute to the development of automatic speech analysis in psychosis.

Takeaways, Limitations

•
Takeaways:
◦
We present an objective and scalable FTD evaluation method based on automatic speech recognition (ASR).
◦
We demonstrate that pause features play an important role in predicting FTD severity.
◦
Improved FTD prediction performance by integrating stopping features and semantic consistency measures.
◦
Consistently improved performance across a variety of situations (natural diary writing, picture descriptions, dream stories).
•
Limitations:
◦
The size of the dataset used may be relatively small.
◦
Further research is needed to determine generalizability across different linguistic and cultural backgrounds.
◦
There is a dataset dependency of the stopping pattern.
◦
Additional validation and standardization processes are needed for clinical use.
👍
Made with Slashpage