Change check
Check an analyser change in your browser.
Drop the paired results from your method comparison. For one analyte, free, you see the bias at each of your decision lines and how many results crossed each line. Try the public example to see the full change report free. Your file is read on your own computer and never uploaded.
Run it
Your paired results
Drop your paired-results file here
A CSV with one row per sample and analyte: pair_id, analyte, unit, value_old, value_new, old_system, new_system. Comma, semicolon, tab or bar separated; decimal commas are read.
Your file stays on this computer. This page reads it in your browser and uploads nothing. The only things this page can send are a payment check (a payment reference, nothing about your file), a cookieless count of what was done here (for example “a file was opened”, never the file, its name or a result) and, if you choose to, the summary shown under “Contribute”.
The example is 1,428 paired full blood count results from an open-access deposit (PMC12956494 (opens in a new tab), CC BY 4.0), with the systems shown under generic names. On the example the full change report is shown free. Download the CSV template.
US$490 per change report
Every analyte in your file, the crossing counts set beside repeat testing, the category matrix, the crossed-sample list and a printable report. Paid by card through Stripe. Terms of sale.
Stripe takes the payment for CareerLoop Inc., the company behind NakedSignal; that name appears on your card statement.
Full refund within 14 days of payment, for any reason.
Payment opens in a new tab. Keep this tab open: when payment completes, the report unlocks here too, with your file still loaded.
Need an invoice or a purchase order instead of a card? Ask for an invoice.
Running method comparisons for client laboratories? Buy one change report per study, at the same price, and ask for the invoice in your client’s name if that suits them.
01Intended use
What this check is, and what it is not
The NakedSignal change check is laboratory-management software. It is intended for use by laboratory scientists to describe, for a population of paired results, how a change of analyser, reagent or method moved results across the decision lines the laboratory chose, and to plan the laboratory’s own verification, re-testing capacity and communication with clinicians.
It is not intended to diagnose, monitor or treat any person, to change, correct or re-report any patient result, or for any time-critical decision. The list of individual samples is for review by the laboratory under its own procedures and is not a recommendation about any patient.
It is for a laboratory professional responsible for method verification and method change: a clinical or biomedical scientist, laboratory director or quality manager who knows method comparison and the laboratory’s own decision limits. It is not intended for clinicians or patients.
Warnings
- For laboratory professionals. Not for use by clinicians or patients.
- Not for diagnosis, monitoring or treatment of any person. It does not change, correct or re-report any result.
- Not for any time-critical decision, including critical-value handling or results awaiting release.
- The crossed-sample list is for the laboratory’s review under its own procedures. Do not write it into the laboratory information system or the patient record as a flag or alert.
- The decision lines are the laboratory’s. The check does not say which method is right or that any result is wrong.
- Figures marked ASSUMED rest on a default precision, not the laboratory’s own.
- It describes the pairs supplied. It is not an ISO 15189 verification, not a device approval, and not a CE- or UKCA-marked product.
- Resolve or acknowledge every data-check warning before the figures are used.
02Method
How the figures are computed
Per analyte and pair of systems: the comparison line is a Passing-Bablok regression of new on old; the average difference and limits of agreement are Bland-Altman; the bias at a decision line is the comparison line’s value there minus the line. A result crosses a line when its old and new values fall on different sides of it. “Expected from repeat testing” models a repeat on one system as normal with the repeat-test CV, so the crossings a change causes can be set beside those that no change at all would cause. Groups with fewer than twenty usable pairs are not computed.
The browser engine gives the same figures as the reference program behind our full site report: on 14 public comparison data sets (34 analyte groups), 2,165 of 2,165 figures agreed. Not computed in the browser: bootstrap intervals and the re-test band, which are part of the full site report (see the sample report).
Default decision lines, assumed precisions and acceptance limits are common published values, each labelled with its source; replace them with your laboratory’s own. More on the method: how it works.
03Your data
Where your file goes: nowhere
The file is opened by this page in your browser and computed there. It is not uploaded to us or to anyone else, and the page keeps no copy once you close it. Use your own pair IDs, never a patient identifier or name.
The page makes two kinds of request about your check, both optional: a payment check, which sends only the payment reference Stripe returns, and, if you choose to contribute and press send, the derived summary you see before sending, to the Analyser-Change Registry. Like every page on this site it also sends cookieless page counts, and here a count of what was done (the example tried, a file opened, the payment or invoice link pressed), never the file, its name or a result: see the privacy notice. Terms of sale and the refund rule: change check terms.