Psychometric analysis of speech-based digital biomarkers of alogia symptoms in treatment resistant schizophrenia: Comparison with clinical ratings of expressive negative symptoms

Jean-Pierre Lindenmayer; Anzalee Khan; Beverly J Insel; Hardik Kothare; Michael Neumann; Danyah Nadim; Rachael Mishkind; Vikram Ramanarayanan

doi:10.1016/j.schres.2026.05.018

Psychometric analysis of speech-based digital biomarkers of alogia symptoms in treatment resistant schizophrenia: Comparison with clinical ratings of expressive negative symptoms

Schizophr Res. 2026 May 28:295:110-119. doi: 10.1016/j.schres.2026.05.018. Online ahead of print.

Authors

Jean-Pierre Lindenmayer¹, Anzalee Khan², Beverly J Insel³, Hardik Kothare⁴, Michael Neumann⁴, Danyah Nadim³, Rachael Mishkind⁵, Vikram Ramanarayanan⁶

Affiliations

¹ Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY, United States of America. Electronic address: jean-pierre.lindenmayer@nki.rfmh.org.
² Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY, United States of America; Manhattan Psychiatric Center, New York, NY, United States of America.
³ Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY, United States of America.
⁴ Modality.AI, Inc., San Francisco, CA, United States of America.
⁵ New York University, New York, NY, United States of America.
⁶ Modality.AI, Inc., San Francisco, CA, United States of America; University of California - San Francisco, San Francisco, CA, United States of America.

PMID: 42208341
DOI: 10.1016/j.schres.2026.05.018

Abstract

Background: The measurement of expressive negative symptoms (ENS) is challenging due to reliance on subjective measurements with variable reliability. Digital speech assessments using a cloud-based multimodal dialog platform offer a promising objective alternative. This study evaluates psychometric properties of acoustic speech markers as complementary measures of ENS in treatment-resistant schizophrenia (TRS). We also examine whether speech markers differentiate TRS from healthy controls (HC) and their association with clinician-rated ENS severity.

Method: 67 participants with DSM-5 TRS, and 63 HC completed two standardized, 10-min sessions using the Modality platform. Eight speech measures were retained. Trained raters assessed clinical measures. The analysis examined known-groups validity, test-retest reliability, and associations between speech measures and clinician-rated negative symptoms.

Results: Individuals with TRS demonstrated significant differences vs. HC in four speech measures. They required longer time to complete the reading passage (mean 44.09 vs. 31.90 s; p < 0.0001) and spoke less during spontaneous speech (mean 20.19 vs. 29.06 s; p < 0.0001). Higher Cepstral Peak Prominence was significantly associated with greater severity of ENS across 5 clinical measures. Speaking duration was significantly negatively associated with the PANSS ENS subfactor. Monotonous speech was significantly negatively associated with ENS.

Conclusion: Specific acoustic speech measures differentiated TRS from HC and were associated with clinician-rated ENS, particularly alogia and blunted affect. These findings support speech-based digital markers as objective correlates of ENS, while also indicating that they reflect only part of broader clinical construct. Longitudinal studies are needed to determine sensitivity, and whether these markers reflect trait versus state features.

Keywords: Digital assessment of speech; Expressive negative symptoms; Speech based biomarkers; Treatment-resistant schizophrenia.