One-Minute Speech Test Explored for Anxiety and Depression Screening

A study published in JASA Express Letters explored whether acoustic features of a one-minute spoken-word task could help screen for co-occurring anxiety disorder and major depressive disorder. The work involved researchers from the University of Illinois Urbana-Champaign, the University of Illinois College of Medicine in Peoria and Southern Illinois University School of Medicine. It is a research-stage screening approach, not an established diagnosis from a voice recording.

As background, the article cites US figures for 2021: 8.3% of adults with major depressive disorder and 19.1% with anxiety disorder. It gives diagnosis and treatment figures of 61.0% for major depression and 36.9% for anxiety, linking gaps in care with social, cognitive and structural barriers. The exact definitions and denominators behind those background figures are not specified in the recovered article.

A one-minute word task

Author Mary Pietrowicz said delays in diagnosis and treatment motivated the research. The team considered speech because acoustic signals may reflect several mental, neurological and physical conditions. She noted that anxiety, depression and their co-occurrence may have different acoustic features; markers for one condition can partly offset markers for the other.

Women with and without co-occurring anxiety and major depression were recorded through a secure telehealth platform. In a timed semantic-fluency task, they named as many animals as they could. The researchers extracted acoustic and phonemic features from the recordings and trained a machine-learning model to distinguish the groups. They reported that the one-minute test showed promise for screening in this study, but the article does not provide enough sample or external-validation details to establish clinical reliability.

Pietrowicz said the co-occurring group tended to use simpler words and showed less variation in phoneme length and lower levels and variation of phonemic similarity. She plans to investigate possible biological mechanisms, expand the size and diversity of the data, include more modalities and refine the model. More data across populations and conditions would be needed before a diagnostic tool could be developed.

Original report: AIP Publishing.