Speech and Language Markers of Bipolar Disorder: Challenges and Opportunities.
Level 3 - non-randomized controlled study
Systematic review of predominantly cross-sectional observational studies
PubMed 42360412 · doi:10.1111/bdi.70119
What was done
Systematic review of Medline and Google Scholar databases up to February 2026 for English-language studies evaluating speech markers in bipolar disorder. The authors performed topic modeling thematic analysis on abstracts and qualitative gap analysis across 43 included studies (screened from 867 records).
What was found
Of 867 screened records, 43 studies were included. Most studies focused on cross-sectional detection of mood states or diagnosis rather than prospective prediction. Speech features distinguished bipolar disorder from schizophrenia, depression, and healthy controls. Manic states were characterized by measures of pressured speech, derailment, grammatical errors, and word repetition; depressive states by increased personal pronoun use, reduced verbal fluency, and reduced speech quantity. Replications were limited, and no quantitative effect sizes or diagnostic accuracy metrics were reported in the abstract.
Why it matters
Speech analysis offers non-invasive markers of current mood states in bipolar disorder, but current evidence does not support its clinical application for relapse prediction, treatment response monitoring, or diagnosing mixed episodes.
Limits
The abstract provides no numerical effect sizes, diagnostic accuracy statistics, or total participant sample sizes. The underlying evidence base is largely cross-sectional, lacks longitudinal evaluation, and exhibits minimal independent replication.