For hospital systems
Your EHR ends at the parking lot. Your readmissions don't.
Under HRRP, Medicare withholds up to 3% of inpatient payments for excess readmissions — and the drivers live outside your walls, where clinical systems can't see. Variate adds the field layer: the neighborhoods your discharges go home to, and which ones can't support recovery.
~$17,700
direct cost per readmission (AHRQ HCUP)
75–82%
of hospitals projected in penalty territory, FY2027
30 days
to live deployment
Three problems, one blind spot.
Pain 1 — the penalty math
Every excess readmission compounds
- ~$17,700 direct cost per readmission — before HRRP multiplies the damage across your Medicare book.
- FY2027: Medicare Advantage patients enter readmission cohorts for the first time.
Pain 2 — the blind spot
The drivers aren't in the chart
- Transportation, housing, distance to pharmacy, who's home to help.
- The forces that bounce a patient back live in the field, where clinical systems can't see.
Pain 3 — the stack gap
Nothing you own builds the list
- Care-transition software works the list you already have.
- Your stack has nothing that builds the list from the field — that layer doesn't exist in it.
We don't promise a percentage. We show you the field — with receipts.
The engine composites 200+ federal health variables into precision maps of the neighborhoods your discharges go home to. Every answer ships with a decision receipt: the variables, the weights, the data vintages, the human who approved. Inspect the proof layer →
One page, every seat at your table.
The forwardable page: send this one URL to whoever needs to evaluate us — each block answers a different room.
For your CFO
The penalty math, in your market
- Name your market — get your exposure number, free, usually same day.
- Leading indicators (field visits, follow-up completion) move in weeks; HRRP cohorts move on multi-year windows. We'll say which is which, unprompted.
For population health
Ask in plain English, challenge the weights
- “Where are dual-eligible seniors with high heart-failure risk and no transit access?” — answered in ~10 seconds, census-tract level.
- Freshness table on every layer: source, vintage, refresh cadence. See the data dictionary →
For your data science team
Weighted overlay you can audit
- Deterministic MCDA, not a model mood: same question + same weights + same vintage = same map, forever.
- Full data dictionary with 200+ variables. Per-tract decomposition: which variables drove this score is arithmetic, and we show it.
- Population math done properly — never county averages smeared across empty space.
For your security review
We don't want your PHI
- Analysis starts from public federal datasets, not patient records.
- Google Cloud under a signed BAA; Vertex AI, Cloud Run, Cloud SQL.
- Governed agents with receipts: every decision logs actor type (HUMAN / AGENT / SYSTEM), state, and confidence. Security & procurement docs →