Automated Speech-Based Modeling of Item-Level Symptom Severity in Schizophrenia - Takeaways - MDSpire

Automated Speech-Based Modeling of Item-Level Symptom Severity in Schizophrenia

  • By

  • Silvia Ciampelli

  • Janna N. de Boer

  • Sanne Koops

  • Evan Troelstra

  • Almut Jebens

  • Jan-Bernard C. Marsman

  • Arnout C. Smit

  • Amir Hossein Nikzad

  • Ryan Partlan

  • Philipp Homan

  • Wolfram Hinzen

  • Sunny X. Tang

  • Iris E. C. Sommer

  • June 25, 2026

  • 0 min

Share

  • 1

    Speech serves as a primary medium for patients to express mental states and for clinicians to assess symptoms in psychiatry.

  • 2

    Automated speech recognition and natural language processing enable scalable monitoring of speech changes to detect early signs of relapse in psychosis.

  • 3

    This study evaluated the association between speech samples and psychotic symptom severity using the PANSS and BPRS scales in Dutch and US cohorts.

  • 4

    A total of 160 features were extracted from Dutch speech samples, while 138 features were extracted from US samples for analysis.

  • 5

    Principal component analyses were conducted to reduce dimensionality of speech features, which were then tested against symptom scores using mixed-effects models.

Original Source(s)

Related Content