Can linguistic analysis be used to identify whether adolescents with a chronic illness are depressed?
Jones, Lauren Stephanie, Anderson, Emma, Loades, Maria et al. · Clinical psychology & psychotherapy · 2020 · DOI
Quick Summary
This study looked at whether doctors could use the words young people write in online therapy messages to figure out if they also have depression alongside ME/CFS. Researchers developed a system to analyze language patterns in 16 teenagers' messages, but found it didn't work very well at identifying depression compared to standard depression screening tests. A few language patterns showed promise—for example, teenagers with depression used more words about the past and fewer words showing mixed feelings—but the method overall wasn't reliable enough to be useful.
Why It Matters
Depression commonly co-occurs with ME/CFS in adolescents and affects treatment outcomes, yet it's often underdiagnosed in online therapy settings. Developing objective linguistic markers could help clinicians more readily identify youth needing mental health intervention. This early-stage research contributes to understanding how language patterns might signal depression in chronic illness populations.
Observed Findings
Only 3 of 16 participants (19%) showed agreement between the linguistic coding scheme and self-reported depression assessment.
'Past focus' words appeared more frequently in e-consultations from adolescents with probable depression (mean rank 10.70) compared to no depression (mean rank 1.50; p<.05).
'Discrepancy' words (indicating mixed feelings) appeared less frequently in probable depression (mean rank 6.40) versus no depression (mean rank 16.00; p<.05).
The linguistic coding scheme identified potential depression in only a minority of participants whose self-report assessments indicated depression.
Inferred Conclusions
The developed linguistic coding scheme had insufficient discriminative validity to support clinical identification of comorbid depression in adolescents with CFS/ME within this sample.
Two specific linguistic features—increased past-oriented language and decreased discrepancy language—showed promise as potential markers of depression and warrant further investigation.
Larger, well-powered studies are needed before linguistic analysis tools can be implemented in clinical practice for depression screening.
Remaining Questions
Would the linguistic features identified (past focus and discrepancy words) replicate in a larger, more diverse sample of adolescents with ME/CFS?
Can the coding scheme be refined to improve its sensitivity and specificity for detecting depression in online therapy contexts?
What This Study Does Not Prove
This study does not prove that linguistic analysis cannot identify depression in ME/CFS—the small sample size (n=16) and preliminary nature of the coding scheme limit generalizability. The findings do not establish causation between language use and depression, only associations. The poor overall accuracy means this method cannot yet be recommended for clinical use without substantial further development.
Tags
Symptom:Fatigue
Phenotype:Pediatric
Method Flag:Weak Case DefinitionNo ControlsSmall SampleExploratory Only