The evidence
What this does not do
The honest boundaries: what public data cannot tell you, which numbers are upper bounds, and where a human has to make the call.
Every analytical product has a boundary. Most of them are discovered by a customer, in a meeting, at the worst possible moment. Ours is written down here.
None of the limits below is a defect. They are the honest shape of what public evidence can support — and knowing them is what lets you use the parts that are strong.
It is not a prediction#
Probability of success is a base rate drawn from comparable programmes, presented as directional evidence with its assumptions visible. It is not calibrated to your programme and is not a forecast or a guarantee of your trial’s outcome.
The same applies to confidence scores: while uncalibrated they order results rather than measure them.
Some numbers are upper bounds#
Cohort figures are computed from epidemiology and eligibility modelling. They describe how many patients could in principle qualify — not how many you will enrol.
- The eligible pool passes the modelled criteria and a screen-failure prior. No reach, catchment or consent stage is modelled, so it is an upper bound.
- Prevalence snapshots are point estimates for sizing, not a live epidemiology feed.
- Intervals around a modelled figure come from the modelling. Where the underlying epidemiology publishes no interval, the spread is an assumption and is labelled as one.
Budgets are bands, and wide ones#
Cost feasibility is built from public per-patient priors, presented as a low–median–high band rather than a quote. Treat public-data budgets as roughly ±50%. They are useful for comparing designs and countries; they are not a substitute for a CRO bid.
It only knows what is public#
- Unregistered trials, unpublished results and confidential competitor programmes are invisible to it.
- Registry data is only as accurate and as current as the sponsor who filed it.
- Sparse indications produce sparse assessments — which the evidence bands and the evidence ledger will show you plainly rather than papering over.
A human makes the call#
The platform produces decision-support material. It is not a medical device, it does not practise medicine, and it does not approve anything. Deliverables are drafts until a named person on your team reviews and signs them.
Decision-support only — not a medical device.
Related
Next
Start here
Reading an assessment
The six pillars, the questions each one answers, and how to tell a number backed by several independent sources from one resting on a single record.
OpenThe evidence
Where the numbers come from
Around fifty live public-data connectors, seven of them load-bearing, each pinned to the snapshot it was read from — and an honest ledger of the ones that returned nothing.
OpenYour data
Your data
What the platform stores, where it is processed, what it is never used for, and how long it is kept — in plain terms, with the legal detail linked.
OpenSee a verdict you can actually check.
Send us a protocol — or just a molecule and an indication. We'll return a fully cited feasibility assessment you can trace, line by line, back to public data — yours to defend in a bid, take to your board or investment committee, or hand to a regulator.