Gating CNS Enrollment Capital Against Unit Economics

The CNS Enrollment Kill Switch: Gating Lundbeck's Brain-Health AI Bets Against Real Cost

You are looking at a brain-health pipeline carrying a $1.76 million per-patient reconciliation tax, while the physical floor of visits, labs, and imaging runs about $21,275 — an 83x waste gap that no AI vendor slide deck closes. That gap is the job. Not "find patients faster." Not "hours saved." Not the OpenAI/Iambic/Gefion partnership press line. The job is to neutralize a quantified future revenue liability — every month a Phase III CNS protocol misses its randomization window burns modelable NPV — by transferring it into an audited, fully-loaded cost per eligible randomized patient the CFO can sign. The math is brutal: MASCOT's empirical baseline reads roughly $84,000 per randomized patient against a physical floor of $6,000 to $9,000, a 4x to 5x multiple sitting in fat inversion territory before any matcher touches a protocol. Penn CNS coordinator K. Vasquez logged 38 inbound names across two weeks, nine reached Visit 1, zero randomized — and a single screen-failed Parkinson's patient named Mr. D. consumed 6.5 coordinator hours on a DaTscan insurance appeal that blew the 14-day window. The old scoreboard treats "ahead of schedule" as proof; the new scoreboard treats it as a time metric, not a P&L line, and it cannot fund capital.

Traditional executives fall into the illusion of optimization trap by asking vendors and PIs what features they want — and customers always say speed, cost, and accuracy, but they cannot tell you why the industry fails to deliver. That is analogical thinking: copying Deep 6 AI, Mendel, or Trial Library pitch decks because the capability is visible at the top of the funnel. The fix is First Principles engineering — mathematically calculating N (fully-loaded sites, coordinators, screen-fail labor, startup amortization against the 8% fixed protocol-average hurdle rate, elapsed time) divided by D (the IRB-approved Schedule of Activities floor), proving the strategic hypothesis with the customer's own cost before the bet touches a patient. The MASCOT single-protocol audit container enforces this: tenant-level CTMS fencing prevents hedge-spread across THRIVE and DEEp OCEAN, so the gap measurement cannot be laundered into a partnership win. Jevons Paradox fires hard here — at E = 1.30, every 10% lift in inbound candidates produces a 13% lift in coordinator work against an unchanged D, and tribal knowledge decay is why every coordinator turnover resets the organization's enrollment intelligence.

Three structural cures collapse the 83x cost gap. Computational Removal replaces CTMS spiral-notebook backfills and DaTscan insurance appeals with continuous, automated per-patient cost taggers that pull site activation, coordinator labor, and CRO rates from the ERP/CTMS financial ledger in real time. Cognitive Elimination codifies the Vasquez "chart-pull, MRI safety, hemolyzed redraw" tribal heuristics into native EHR data-abstraction rules that route eligibility queries through the same function already reviewing prior authorization charts — no headcount expansion, just sunk-capacity arbitrage. Architecture Simplification externalizes site startup to CNS-specialty SMOs that already amortize IRB and contracting across portfolios, and amortizes MASCOT site activation over the DEEp OCEAN trial's already-qualified footprint at near-marginal cost, killing the re-amortization multiplier that inflates N super-linearly.

This architecture builds an un-rippable moat because ripping it out is an operational lobotomy — replacing the MASCOT audit container means deleting the locked Finance allocation conventions, the Big 4-reviewable baseline ledger, the pre-signed CFO side letter, and the co-signed kill sentence that fires mechanically when the 11-to-14 week DSMB window closes. Every randomized patient tagged with fully-loaded cost feeds the data flywheel: local transactions train the global rule-based eligibility engine, calibration tightens with every Visit 1, and the 30% matcher-spend cap on measured savings prevents vendor capture from owning the registry inversion relationships. The kill sentence — "no audited cost-per-randomized-eligible-patient baseline exists, so this bet cannot pass the CFO unit economics gate" — is what protects the $559.1 million annual strategic value stack: $52.0M direct OpEx reclaimed, $460.9M in transaction pipeline preserved against the 35% friction abandonment risk, and $46.1M in 10% value-pricing capture toll.