Prospective by default
Every visit is captured as defined variables as it happens, so the data for a study is collected before anyone asks the question.
For research
A research database that builds itself at every visit.
Every visit is captured as defined variables, the same way at every center, with outcomes attached over time. Studies start from structured data, not chart review.
The idea
What a defined data layer makes possible.
When every visit is stored as variables defined by the proprietary database structure, the dataset builds itself as patients are seen.
Pools across centers
Every center records the same variables the same way, so combining their data doesn’t start with cleaning it.
Outcomes on the same record
Patient-reported outcomes, the scores patients give their own pain and function, will sit on the same record as the visit, tied to the exact procedure and the patient’s baseline.
Built for causal methods
Causal methods need to know what was known at the decision, the decision itself and what happened next. A defined data layer records each one.
In practice
Spend the time on the study, not the chart review.
Your own research database
Every visit lands in one analysis-ready dataset. The research team queries it directly.
Skip the re-abstraction
Re-abstraction means reading old charts by hand to pull the data back out. Here a study starts from structured data instead of 18–36 months of that.
Lines up with registries
The variables line up with registries and with standard patient-reported outcome questionnaires, the short forms patients fill in about their own recovery.
Causal AI
Most reports tell you what happened.
Causal AI estimates what would have happened if you’d chosen differently. It only works if it knows what you knew at the visit.
Coming later
The trap
Patients who got surgery sooner did better. Was it the timing, or were those patients healthier to start with? A normal report can’t tell you. Causal AI can, but only if it knows what you knew at the visit.
What causal AI needsWhat the data layer records
What was known before the decision, and whenThe diagnosis, exam and imaging, recorded at the visit
The decision itselfThe treatment chosen, saved as its own defined variable
How it turned outOutcomes at 6 weeks, 3 months and 1 year, once Outcomes is live
One standard across every siteThe same defined variables at every center
Facts a physician confirmedVariables confirmed by the physician before the note is closed
Causal AI needs to know why you made the call. The EMR records what you billed. LastPatient records what you saw.
The data layer
Each visit type knows what it should capture.
Every visit is saved as defined variables, grouped the way surgeons think about the visit. The note is then written from that record.
Context
- VisitFollow-up
- AfterInjection
- Timing6 weeks
- HistoryPrior MRI
One patient’s record
Defined variables
Diagnosis
- primary
- glenohumeral osteoarthritis
- Source: Heard
- laterality
- right
- Source: From history
Exam
- forward elevation
- 140°
- Source: Macro (dot phrase)
- external rotation
- 30°
- Source: Macro (dot phrase)
Imaging
- modality
- X-ray
- Source: Typed
- classification
- Walch B2
- Source: Dictated
Plan
- treatment
- physical therapy
- Source: Heard
- follow-up
- 6 weeks
- Source: Typed
Multi-center by default
Every center defines the same variables the same way.
So pooling data across centers comes down to a query.
Center A
- Dxcuff tear
- Oprepair
Center B
- Dxcuff tear
- Oprepair
Center C
- Dxcuff tear
- Oprepair
One pooled dataset
Every center uses the same defined variables, so nobody re-abstracts charts before the analysis.
Planning a multi-center study? We’re setting up studies with research teams now, and every center defines the same variables the same way from day one.
Running a multi-center study? Start with the visit.
Bring the next protocol. Our research team will walk through it.