botminds.aiRequest a demo
WorkflowRegulatory & Clinical Evidence

IVDR Performance Evaluation Report

An IVDR performance evaluation report built claim by claim: each performance claim in the instructions for use is read against the study that should carry it, and people sign the gaps and the wording.

See the demoBook a demo
Follow claim C-02 from one limit to two ↓
Watch with transcript →
The problem

Instructions for use that drifted away from the studies behind them

A legacy assay moving to IVDR carries dozens of performance claims in its instructions for use, written years ago, and the evidence sits in study reports, bridging studies and literature filed across several systems. Someone cross-reads them by hand, claim by claim.

What slips through is rarely a missing study. It is a limit rounded the wrong way, a sample type nobody tested, or an interference study run on the reagent version before a supplier change. The notified body finds it, and each finding costs a question round.

Built forClinical affairs teamsRegulatory affairs teamsAssay developersBiostatisticians

The Problem Today

Each claim in a diagnostic test's instructions for use is cross-read against dozens of study reports by hand.

300 h
to build or update one evaluation report by hand our estimate

Cross-read by hand today

  • Claims
  • Studies
  • Versions
  • Literature
  • Gaps found by the reviewerEach missing study costs a question round
  • Evidence filed elsewhereBridging studies sit under change records

The Solution

  1. Each claim listedValue, units, sample type and population taken from the instructions
  2. Three pillars per claimScientific validity, analytical and clinical studies under each claim
  3. Claim against its studyA limit that differs from its study's own table is flagged
  4. Owners decide, people signEach finding has one owner, and four named people sign the report

Built-in controls

  • An approver opens the study before signing
  • Agents read study reports and literature only
  • Agents never sign, decide or write conclusions

What You Get

Notified body questions start from a file whose gaps were already found, owned and closed.

$1.2M
a year of clinical scientist time at a hundred reports a year our estimate
  • A cited report, by sectionEach drafted sentence carries its source passage
  • Label changes drafted from tablesWording from the study table, with touched claims re-checked
  • An evidence trail for inspectorsWho signed, when, what it meant and in which session
One troponin assay, one transition

A performance evaluation report for a legacy class C troponin assay

A high-sensitivity troponin I assay, self-declared under the old directive, now class C. Its IFU carries 24 performance claims, and the technical documentation is due to the notified body in about six weeks.

demoThis troponin assay, its studies, its staff and its numbers are sample data, built for the working solution.

  1. 01Before the read

    24 claims on a skyline, and not one marked before the agents read

    Clinical affairs specialist

    Each IFU claim is a column, with three pillars waiting under it: scientific validity, analytical performance, clinical performance. The evidence on file is counted beside it: 34 analytical reports, 7 clinical, 16 sources. Until the read runs, every pillar says not read yet, so nothing can be mistaken for a check.

  2. 02The read

    20 claims stand, 1 contradicts its own study, and 3 have gaps

    Clinical affairs specialist

    She starts the read. Eight agents work through the IFU, the study reports and the sources, then raise each claim's pillars block by block. The skyline settles at 20 that stand, 1 that contradicts and 3 gaps. The claim that contradicts is the 99th percentile limit, the one the notified body will open first.

  3. 03The read

    A rule-out claim rests only on two papers about other makers' assays

    Clinical affairs specialist

    C-24 says a single result below 3 ng/L at presentation rules out infarction. Its scientific validity pillar holds two papers on other manufacturers' assays and a source saying such thresholds are assay-specific. The clinical study never looked below the 99th percentile. The gap goes to her, due in ten working days.

  4. 04Review

    One limit of 25 ng/L in the instructions; 14.3 and 31.6 in the study

    Clinical affairs specialist

    The claim page puts the IFU sentence beside the reference interval study's own Table 4. All adults: 24.7 ng/L. Women 14.3, men 31.6, and the difference meets the study's own test for separate limits. Value and units match; population does not. The study supports two limits and the label gives one.

  5. 05Review

    The new wording comes from the table, and two more claims move with it

    Clinical affairs specialist

    She asks for the wording. The change request drafter writes sex-specific limits from Table 4, rounded as the IFU rounds them: 14 for women, 32 for men, 25 for all adults. It also finds two other claims that quote the old limit, imprecision and clinical sensitivity, and re-checks each against its own study.

  6. 0608:10

    She signs the three changed claims as author, with her own ID

    Clinical affairs specialist

    The signature dialog shows the old sentence struck through and the new one beside it. She signs under her personal login, and the meaning names all three claims and says the wording follows the cited reports. Words she typed by hand would be flagged unchecked, and the approver must read them.

  7. 0708:12

    The approve button waits until he has read all three studies

    Regulatory affairs manager

    Viewing as the regulatory affairs manager, the change request lists the three touched claims. His approval needs the change packet open for six seconds: the reference interval study, the imprecision panels and the re-analysis at sex-specific limits. Until each record has been on his screen that long, approving on her summary is not possible.

  8. 0808:13

    A new cut-off is a significant change, so it waits for the IVDR file

    Regulatory affairs manager

    The re-analysis lowers specificity from 90.4% to 89.2%, and the drafted assessment proposes calling that a change to the intended purpose. On the legacy label, it would end the assay's transition. He decides significant and approves: the new wording goes into the IVDR technical documentation and the current instructions stay on the market.

  9. 0908:14

    21 of 24 claims are ready, and the changed ones show their new wording

    Clinical affairs specialist

    Back on the skyline, the contradiction is resolved and the count reads 21 of 24 ready. Opening the sensitivity claim shows the new sex-specific figures, matched again field by field against the re-analysis. The statistician still checks it before the report is signed. Three gaps remain open, owned and dated.

Four claims that looked settled until their studies were opened

Each one cites the passage it rests on, and each one goes to a named owner.

One reference limit for all adults

The instructions say 25 ng/L for men and women alike. The study's own table gives 14.3 ng/L for women and 31.6 ng/L for men, and meets its test for separate limits.

A sample type nobody tested

Detection at 2.0 ng/L is claimed for three sample types. The two studies used serum and lithium heparin; the K2-EDTA equivalence study starts at 10.4 ng/L.

Biotin tested on the old reagent

The interference claim rests on a study of reagent version 3. A version 4 study exists, filed under the supplier change record and missing from the plan's evidence list.

A rule-out cut-off from other makers

A single result below 3 ng/L is claimed to rule out infarction. The only support is two papers on other manufacturers' assays.

demoSample data, from the working solution.

The cost of two limits: sensitivity up, specificity down

The sensitivity claim, C-23, from the clinical performance study, recomputed at the single limit and at the sex-specific limits.

Clinical performanceOne limit, 25 ng/LWomen 14, men 32 ng/L
Patients, with infarction1,560, 2501,560, 250
Sensitivity224 of 250, 89.6%228 of 250, 91.2%
Specificity1,184 of 1,310, 90.4%1,169 of 1,310, 89.2%
Imprecision near the limits4.8% CV at 24.9 ng/L6.1% at 13.2, 4.3% at 32.8

demoSample data, from the working solution. All three CVs sit inside the 10% criterion. The drop in specificity is what makes the new wording a significant change for the legacy device.

Around the report

Evidence, findings and signers around one IVDR performance evaluation

The screens the assay scientist, the statistician and the regulatory manager open between the read and the signatures.

FindingsEach contradiction and gap with its owner, due date and stage. The report cannot be signed while any is open.
Evidence58 records read in place, each with its method, date and the reagent version it tested, and the claims behind it.
Statistician's checkA recorded random sample of 5 claims that stand, plus every decided finding and every claim a decision changed. Each record is opened before agreeing.
The reportDrafted sections with a citation per sentence and four signers in order. Signing stays blocked while gaps are open.
Legacy devices312 legacy devices by class, owner, application, written agreement and report date, for planning the wave behind this one.
Class D specificationsFor a class D assay, study results against each row of the common specifications. Cross-reactivity has 96 specimens of the 100 required.
Audit trailWho signed, when, what each signature meant, the session and the reason, exportable as CSV.

demoSample data, from the working solution.

Four signatures in order, and the conclusions written by hand

Nobody checks their own item, and reopening the report clears every signature.

Clinical affairs specialist
Sees
Each claim's three pillars and the drafted wording
Signs
Signs her own claim decisions as author
Assay development scientist
Sees
Gaps on detection and interference, with the studies found
Signs
Narrows a claim, plans a study or confirms evidence
Biostatistician
Sees
A recorded random sample of standing claims and every decided finding
Signs
Signs the statistical review for exactly the rows she opened
Head of IVD clinical affairs
Sees
The drafted sections and the open findings
Signs
Writes the conclusions and benefit-risk herself
Regulatory affairs manager
Sees
The change packet behind each touched claim
Signs
Approves label changes, records the significant-change decision, signs last
Who reads the evidence

Eight agents read the claims and studies, and none writes the conclusions

They list, read, match and draft. People decide, approve and sign.

Claim extractor

Lists each performance claim with value, units, sample type, population and section.

  • Never rewrites the instructions for use
Study report reader

Reads design, method, lots, sites, results and the reagent version of each study.

  • Never opens complaint, vigilance or field records
Scientific validity reader

Appraises each source for what it actually measured.

  • Never counts another maker's results as this assay's clinical performance
Claim matcher

Gives each claim a verdict and re-checks the claims a change touches.

  • Never passes a claim without a report
  • Never treats typed wording as compared
Common specifications checker

Compares class D study figures with the common specification tables.

  • Never applies them to a class C device
Report drafter

Drafts the evidence sections with a citation per sentence.

  • Never writes the conclusions or the benefit-risk
Change request drafter

Drafts new wording from the study table and lists each claim it touches.

  • Never sends a change request unsigned
Gap keeper

Gives each finding one owner and a due date, and draws the statistician's sample.

  • Never closes a gap

What IVD clinical affairs leads ask about a performance evaluation report

What is an IVDR performance evaluation report?

A performance evaluation report sets out the scientific validity, analytical performance and clinical performance behind each claim an in vitro diagnostic makes. Here agents read each IFU claim against those three pillars, and people decide and sign.

Does it write the conclusions or the benefit-risk section?

No. The agents draft the evidence sections with a citation per sentence. The head of IVD clinical affairs writes the conclusions by hand, and the regulatory affairs manager signs last after reading them.

What happens when a label change touches other claims?

The agents find each claim that names the same limit and re-check it against its own study. The approver must open the studies behind every touched claim before the approve button works.

Can papers on another manufacturer's assay support a claim?

They can support scientific validity. They never count as this assay's clinical performance, so a claim resting only on them is shown as a gap with an owner.

Does it read post-market or complaint data?

No. The agents read the IFU, the analytical and clinical study reports and the literature. Post-market records stay inside one signed summary from the team that owns them, which the agents never open and the report cites as a whole.

Who can see the working file before submission?

The manufacturer's own team. A notified body reviewer sees the filed package only, once all four people have signed.

Regulatory & Clinical Evidence

Regulatory evidence work beside the performance evaluation

All 5 solutions →
Global Device Registration Dossier Assembly

Every country's dossier from one core file, and every change traced to its filings.

1-min demo510(k) Submission Consistency Review

FDA's questions found and fixed before FDA asks them.

1-min demoNotified Body Deficiency Response

Every repeated question shows why the last answer failed, and every answer cites its page.

1-min demo

Bring one assay's instructions for use
and the studies filed behind it.

We will read every claim against its studies and point to the ones a reviewer would question first.

Book a demoBrowse solutions