BECAUSE OF AI, home

Measured AI benefits · as of

5.85 pointsfall in HADS psychological distress after 18 weeks with a voice AI coach, against 2.76 on a waitlist
receipt
"value": "-5.85",Results, secondary outcome HADS at 18 weeks, Psychological distress, Group 1 LumenClinicalTrials.gov NCT05603923 record with posted results (SPEAC, University of Illinois Chicago) · clinicaltrials.gov

Because of AI, adults with untreated depression or anxiety who did eight problem-solving sessions with Lumen, a voice-based AI coach, had their HADS distress score fall 5.85 points over 18 weeks, against 2.76 points on a waitlist.

receipt
"value": "-5.85",Results, secondary outcome HADS at 18 weeks, Psychological distress, Group 1 LumenClinicalTrials.gov NCT05603923 record with posted results (SPEAC, University of Illinois Chicago) · clinicaltrials.gov
"value": "-2.76",Results, secondary outcome HADS at 18 weeks, Psychological distress, Group 3 waitlistClinicalTrials.gov NCT05603923 record with posted results (SPEAC, University of Illinois Chicago) · clinicaltrials.gov

The SPEAC randomized trial at the University of Illinois Chicago assigned adults with mild-to-moderate untreated depression or anxiety to Lumen-coached problem-solving treatment, human-coached treatment, or a waitlist, and the posted results give the change in Hospital Anxiety and Depression Scale scores at 18 weeks.

Sources

Editor's notes

verify: the distress scale (HADS) is a secondary outcome; the primary outcome is change in fMRI activation, which this entry does not use. verify: the posted record gives group means and standard deviations but no between-group statistical test was found in the text read. verify: the human-coached arm (−7.33) improved more than the Lumen arm (−5.85); the AI coach was not better than a human coach on this measure. verify: the registry labels the waitlist arm PLACEBO_COMPARATOR but describes it as assessments only, with Lumen offered after 18 weeks.

Related

History

  • Added by the backfill run.

Verification

Drafted by
Anthropic claude-sonnet-5-5
Verified by
OpenAI gpt-6.1-sol · pass
How sure
Reasonable
  • The two checkers disagreed on evidence: the drafter said B and the verifier said A. The entry shows the more cautious of the two, B.
Record
data/verifications/speac-lumen-voice-ai-coach-depression-anxiety-rct/2026-11-17.json · The complete registry supports every receipted number and date in a randomized trial with a preregistered primary imaging endpoint; HADS was secondary. It reports greater mean distress reduction with the Lumen package than with waitlist, but provides no between-group statistical test or isolation of AI's contribution.

Two models from different vendors read the primary source independently; this entry is published because they agreed on every receipt and every grade. It is marked reasonable: Only the abstract or a summary of the source could be read, or the dates come from the abstract, or the two checking models disagreed on one grade and the entry shows the more cautious one. Every number is still quoted from the source, and both checking models support it. The models grading an AI registry have an interest in the answer. Cross-vendor checking and the receipts on this page are how the registry handles that, not a claim to have removed it. How the gate works · Dispute this entry

Cite this entry

Permalink https://becauseofai.org/entry/speac-lumen-voice-ai-coach-depression-anxiety-rct“Because of AI, adults with untreated depression or anxiety who did eight problem-solving sessions with Lumen, a voice-based AI coach, had their HADS distress score fall 5.85 points over 18 weeks, against 2.76 points on a waitlist.” Because of AI, updated Nov 17, 2026. https://becauseofai.org/entry/speac-lumen-voice-ai-coach-depression-anxiety-rct