Sign in
Easiest: open the study bot in Telegram and tap the 📊 Dashboard button. You get a link that signs you in for 30 days on this device.
Or sign in with the team password:
Review queue — submitted assessments
Scores are provisional until confirmed. Owners never see scores. Adjust if needed (scale 0, 1, 2, 3, 5 — never 4), add your interpretation, then confirm.
Cohort overview
| Pseudonym | Breed | Condition | State | Doses | Visits | Flags | Locale |
|---|
4A — cohort averages by visit (confirmed scores)
| Visit | Aggressivity | Anxiety | Attachment | Self-control | Global | n |
|---|
Per-item heat-map
Cohort themes (Claude)
Regenerated every 6h if the API key is configured.
Per-dog drilldown
Inbox
Database structure
SQLite. *_encrypted fields hold AES-256-GCM ciphertext.
Analysis bundle
De-identified (pseudonym only, no free-text, no identifiers). Start with the codebook.
Tidy dataset (long)
One row per dog × visit × item, submitted and confirmed side by side. Ready for R / pandas.
Scores (wide)
Per-dog × visit, 20 items + subtotals. Confirmed by default.
Dosing / adherence
One row per scheduled dose: times, administered, mg, batch.
Codebook
Data dictionary + the {0,1,2,3,5} scale + de-identification note.
Case-study Markdown
Prose per completed dog, with clinical interpretations.
Double-blind CSV
Per-dog global vectors for an external blinded reviewer.
Bot script — every owner-facing prompt
The complete script the bot says to owners, pulled live from the system so it always matches the real bot. Review it here instead of walking Telegram.
Team access
Who can sign in to this dashboard (via the bot's Dashboard button or the team password). Add/remove members by editing STUDY_TEAM_TELEGRAM_IDS on the server.