NdDB Readiness Map · Free framework
Neuroimaging-derived digital biomarkers (NdDB) are brain measurements produced by imaging software. The Readiness Map is a free framework to check whether such a measurement can support its intended role in a trial — and what is proportionate to get it there.
6
Failure modes
3
Steps in the map
18
Self-check questions
Jump to Self-check · White paper · How the map works ↓
01 — Where imaging endpoints break
Most imaging biomarkers don’t fail because the signal is wrong. They fail in the gaps between teams — found after commitments are locked.
Hover or tap a failure mode to see where the map answers it.
Tap a failure mode to see where the map answers it.
Endpoint role not prespecified
The claim isn’t defined yet
The trial role — and the minimum evidence it needs — isn’t fixed before the protocol is locked.
Regulatory posture misaligned
Clearance is assumed to cover the trial claim
A cleared product covers its own intended use — not your trial role.
Acquisition chain underestimated
Sites don’t scan the same way
Scanner differences, deviations and drift aren’t controlled to the level the role requires.
Processing pipeline not endpoint-grade
The pipeline works in research, not yet in a trial
Versioning, QC traceability and reprocessing rules are missing.
Change control not defined upfront
No one has decided what counts as a change
Updates to software, settings or QC rules can shift endpoint values unnoticed.
Vendor governance gaps
The vendor’s processing is a black box
Contracts don’t secure audit access, specifications or change control.
Not sure which applies to you? The self-check shows where your programme stands in about five minutes.
02 — How the map works
Two rooms for every candidate, and an advanced door when AI is in the pipeline. The aim is proportion: avoid overbuilding governance too early and underbuilding it until it is too late.
Room 1 · Triage
"Does this output unambiguously match the intended clinical variable for the prespecified trial role?"
Tick what is true for your candidate. The door opens only while all three hold.
You stay in this room. Redefine the claim and the output before choosing how to deliver it — candidates that fail here should not proceed.
Once through, the output is classified (classes A–H) — what kind of output it is.
Room 2 · Pipeline grade
"Which trial roles are credible for this output, given the delivery system we can realistically govern?"
Behind this door — delivery
Credible trial roles
Decision flag
Open a door to see what lies behind it.
Grades are not progressive by default: upgrades are triggered by the intended trial role, not by technical sophistication.
Does the pipeline include AI?
That’s the map for your candidate: a defined output, and a grade matched to its role. Check your programme or read the full method.
Advanced door · only if AI is in the pipeline
"Is version locking, update policy, and change control proportionate to the trial criticality of this output?"
AI does not automatically change the risk class — but it raises lifecycle risk. In a trial, stability matters as much as accuracy.
Open a door to see what lies behind it.
If the output can vary during the study because the model or pipeline is not locked, avoid decision-driving use.
You’ve walked the whole map. Check your programme or read the full method.
The principle underneath: regulation applies to the software that generates the measurement, not to the signal — and the intended use decides how much rigour is needed.
03 — Why an independent view
Independent of execution
No scans, reads or software to sell — so the answer can be “narrow the claim” or “stop”.
Claim-first and proportionate
The evidence your decision actually needs — not more.
Both sides of the endpoint
Device-science rigour and the disease perspective, specific to brain imaging.
“The steps are structured. The judgment required to apply them is not.”
04 — Go further
Self-check
Where does your programme stand?
Six dimensions, 18 questions, an immediate profile.
White paper
The full methodology
Output classes, maturity-grade criteria and the AI governance framework, in detail.
Free course
Clinical Evidence in SaMD Development
Analyse a claim with clinical and scientific reasoning — and recognise when it’s an overclaim.
Need an independent view on your programme? See the three offers or get in touch.