STRATUM COLLECTIVE
Internal
Corpus & Intelligence HQ
The operating surface for Stratum's corpus-first strategy: precedent assets, evidence health, market signals, the regulatory clock, and H1 discovery — in one place.
loading live data…
The Mandate — Corpus First
Decided 2026-07-15: Stratum's product is the denial pattern corpus and the intelligence layer on top of it — not a platform app waiting for tenants. The platform is cold-stored (holding page live, ~1h re-activation runbook). The clock that matters is competitive, not regulatory: build corpus depth in BH clinical necessity before the volume-RCM players' implicit corpora get good enough. Sequence: Spravato corpus now → H1 discovery validation → psilocybin corpus intake at approval (2027–28).
Decision records: stratum-platform/docs/cold-storage-runbook.md · stratum-cli/docs/plans/pr6-data-plane-retirement.md · Ops Tracker
Corpus Precedents
89
6 payers · 61.8% wtd overturn
Evidence Items
—
registry
Approved Phrases
—
phrase registry
New Signals
—
UIS inbox
COMP360 NDA Window
—
days to Q4 2026 filing window
Corpus Assets
BH Denial Corpus
LIVE · N=89 · 6 PAYERS
The foundation asset. Category overturn rates, payer heatmap, weighted corpus analysis, 4 structured appeal playbooks, 6 KB denial patterns including NQTL parity violations (70–86% appeal success).
Open Corpus Overview →
Spravato Wedge Corpus
NEXT BUILD · NOT STARTED
The revenue wedge and the psilocybin proof-of-concept. Map clinical necessity criteria across UHC, Cigna, Anthem, Horizon, Florida Blue (all published); systematize J&J's PA Toolkit + Exceptions & Appeals logic. Target market: 300–400 specialized TRD/psychiatric clinics. Time-sensitive — the volume-RCM implicit corpus clock is running.
Strategy brief →
SUD + Psychedelic LOC Intersection
TRACKING · NEW CATEGORY
PHP/IOP authorization denials for patients who received ketamine/psilocybin in SUD treatment. Genuinely new denial category — too small for statistical corpus approaches, longest lead time against implicit-corpus competition. Zero cost to track; corpus unit = drug PA denial + LOC denial in the same episode.
Payer Registries
LIVE · PUBLIC DATA
MA payer registry (6 insurers, KFF PA metrics — 80.7% MA appeal overturn), Cigna/regional precedents (204 records, 72.3% individual case success), 5-state Medicaid registry structure. Canonical data in stratum-corpus-data.
Pattern analysis →
Signal Feed · UIS · LAST 7 DAYS
loading…
Evidence Registry · LATEST ENTRIES
loading…
H1 Discovery Tracker · MANUAL — UPDATE IN THIS FILE
H1 (knowledge asymmetry) is validated through discovery conversations, not product. Target: 20–30 conversations with BH RCM directors. The three questions that test the thesis:
QuestionWhat validates H1Signal count
Do they know their denial patterns? Can't name top-3 denial codes by revenue impact from memory; don't track Spravato appeal success at code level 0 / 0 asked
Do they reuse what works? Winning appeal documentation is not systematized — each appeal built from scratch 0 / 0 asked
Will they pay for cross-org precedents? "Would 200 clinics' Cigna appeal outcomes change how you build your next appeal?" → yes + willingness to pay 0 / 0 asked
Honest state: 0 of 20–30 target conversations completed as of 2026-07-15. This table is hand-updated — no automation writes here. When conversations start, log per-question tallies in this file (grep for id="disc1").
Regulatory & Competitive Clock
2026-04-18 · DONE
Trump EO — psychedelic fast-track + $50M ARPA-H
Q4 2026
COMPASS COMP360 rolling NDA filing
Q4 2027 – 2028
Plausible FDA approval → REMS + launch
POST-APPROVAL
CPT 0820T–0823T Category I vote → payer adjudication begins
corpus value inflection
The real clock is competitive: Waystar/Cohere/R1 implicit corpora improve with every Spravato + Auvelity claim they process. The window is "before their BH signal gets good enough," not "before FDA approval." Monitor: BH-specific PA intelligence features in their product announcements.