Language-based detection of depression with machine learning: systematic review and meta-analysis - Takeaways - MDSpire

Language-based detection of depression with machine learning: systematic review and meta-analysis

  • By

  • Hadar Fisher

  • Nigel M. Jaffe

  • Kristina Pidvirny

  • Anna O. Tierney

  • Mia S. Vaidean

  • Poorvesh Dongre

  • Christian A. Webb

  • February 24, 2026

  • 0 min

Share

  • 1

    Automated detection of depression using NLP and ML shows a pooled accuracy of 0.80 based on 43 studies with 40,983 text samples.

  • 2

    Subgroup analyses revealed accuracy variations based on language, text source, feature type, and classifier used in the studies.

  • 3

    Studies utilizing structured clinical interviews and non-English languages demonstrated the highest accuracy in depression detection.

  • 4

    Text source was identified as a significant predictor of performance, explaining 13.6% of the variance in detection accuracy.

  • 5

    The findings highlight the potential of text-based depression detection while emphasizing the need for methodological standardization.

Original Source(s)

Language-based detection of depression with machine learning: systematic review and meta-analysis

  • by Hadar Fisher, Nigel M. Jaffe, Kristina Pidvirny, Anna O. Tierney, Mia S. Vaidean, Poorvesh Dongre, Christian A. Webb

  • February 24, 2026

Related Content

FaReWell Depression – a randomized controlled trial of a physiotherapeutic program for the facial rehabilitation of wellbeing in depression

Identifying key predictors of post-stroke depression and cognitive impairment in acute stroke survivors

Perceived stress and depression among Chinese nurses: a cross-sectional mediation analysis of psychological flexibility and its components