Measurement for instrument, site and model change
When the instrument changes, know what changed for the patient.
We measure what an analyser, reagent or platform change does to results at the decision limits that matter, and hand you evidence you can re-run yourself. The same method applies to sites and devices in clinical trials and to the inputs of medical AI.
CD4 results that changed category at the decision lines (100, 200, 350, 500 cells/µL)
Crossed after the change: 535
Expected from repeat testing alone: about 238
Managed with the comparison line and a re-test band, on data the band was not fitted to:
- Excess removed
- 66%
- of the excess removed
- Results re-tested
- 15%
- of results re-tested
Fig. 1. Sample built from public data: 1,885 paired CD4 counts from an open-access deposit, one platform replaced by another. Reported without adjustment, 535 results change category against about 238 expected from repeat testing alone. Correcting with the comparison line and re-testing inside the band removes 66% of that excess on data the band was not fitted to, while re-testing 15% of results.
Source Open-access deposit PMC5669480, CC BY 4.0. Repeat-test imprecision is assumed. No laboratory commissioned or reviewed it.
Show the numbers
| Measure | Value |
|---|---|
| Paired CD4 counts | 1,885 |
| Decision lines (cells/µL) | 100, 200, 350, 500 |
| Results that changed category, reported without adjustment | 535 |
| Expected to change from repeat testing alone (about) | 238 |
| Share of the excess removed on data the band was not fitted to (median) | 66% |
| Share of results re-tested inside the band (median) | 15% |
01The report
What the report gives you
From the paired samples a laboratory already runs for a method comparison, we produce seven things.
The comparison line, with its interval.
How the new analyser's results relate to the old one's across the whole range you measured.
Bias, with limits of agreement.
How far apart the two analysers read on the same sample, and how much that varies.
Bias at each of your own decision limits.
Not an average: the difference at the concentrations where your clinicians act.
Patients across each limit, in each direction.
How many results changed category, set against how many repeat testing alone would have moved.
A re-test band for the transition.
A narrow range around each limit inside which a new result is repeated, sized to a budget you choose.
A one-page note for requesting clinicians.
What changed and what it means for their patients, in their words rather than ours.
For networks, whether sites report interchangeably.
The same analysis across sites, at your limits, showing which site moves which patients.
Your validation software tells you the bias. It does not tell you which patients now sit on the other side of a decision limit, how many of those moves repeat testing would have caused anyway, or what to re-test while both analysers are in use. Those three answers are what the report adds, and it can run on the same paired results you already collected.
It runs on your own computer. No patient row leaves your laboratory; only aggregate files do, and you sign off what is released.
The report is something we do for you, described here rather than shown as a track record. How the method works.
02Evidence
What you can open today
Each item carries its label and the run it came from. You do not need to ask for any of it.
- Public data
Analyser change explorer
34 method comparisons at 14 stand-in sites, from 12 open-access studies. Of the 34 method comparisons, 21 get a full re-test band and 5 an indicative one, because they have fewer pairs than the minimum we set; the other 8 are shown as a comparison only or are too small for a band. In 6 of the 34 comparisons, the change moved fewer results across a limit than repeat testing alone would. All five checks we fixed before the run passed. The checks are ours, not an outside review.
10,579paired results
- Sample
Sample analyser-change report, CD4
A sample report built from one public platform change, as if a laboratory had exported its pairs. It shows three of the seven report parts in full and three in part; the other one needs more than one site. The sample page lists each part and what is not shown, with the reason. Reported without adjustment, 535 of 1,885 results change CD4 category, against about 238 expected from repeat testing alone. Nobody commissioned it; it is there so you can see what you would receive.
535results changed category
- Live demo
MedEval-1
A working version of a versioned evaluation standard for medical AI, built by NakedSignal. Its held-out track seals, scores and rotates its test set. 48 of 48 published scores were re-derived by our own audit from the published manifests.
48 of 48scores re-derived
03Standards and labels
Built for the work you already do
The report is built in the vocabulary of ISO 15189:2022 comparability of results, CLSI EP09 method comparison, and the clinical decision limits your laboratory already publishes. It supports your verification work. It does not replace it, and it is not a certification or a regulatory clearance.
Every result on this site carries one of five labels: public data, sample, offer, live demo or designed. Nothing here comes from a customer’s data.
Public dataSampleOfferLive demoDesigned
Changing an analyser soon?
Send us the change you are planning. We will tell you what the report would show and what it needs from you.