Search Results

Immunoassay for the in vitro quantitative determination of N-terminal pro-Brain natriuretic peptide in human serum and plasma. This assay is used as an aid in the diagnosis of individuals suspected of having heart failure. It can be used as an aid in the diagnosis of acute decompensated heart failure (ADHF) in patients presenting with signs and symptoms of ADHF to the emergency department (ED). The test is further indicated for the risk stratification of patients with acute coronary syndrome and heart failure. The test may also serve as an aid in the assessment of increased risk of cardiovascular events and mortality in patients at risk for heart failure who have stable coronary artery disease.

Elecsys proBNP II and proBNP II STAT are second-generation assays by Roche Diagnostics for the in vitro quantitative determination of N-terminal pro-Brain natriuretic peptide (NT-proBNP) in human serum and plasma. The STAT and the 18 Minute assays are intended for use on the cobas e 601. The cobas e family of analyzers employs the electrochemiluminescence immunoassay "ECLIA" technology. The assays are sandwich principle methods using two monoclonal antibodies which are specifically directed against NT-proBNP. For the neutralization of free biotin in serum and plasma, Roche developed an antibody, which binds to free biotin. The antibodies are specific for free biotin and do not bind to or interact with the biotin-linker conjugates.

The provided document describes the Elecsys proBNP II and Elecsys proBNP II STAT assays, which are in vitro quantitative determination tests for N-terminal pro-Brain natriuretic peptide (NT-proBNP) used as an aid in the diagnosis of heart failure, specifically acute decompensated heart failure (ADHF). The document details the analytical and clinical performance evaluations of these devices.

Since the device is an in vitro diagnostic (IVD) test and not an AI-powered medical device for image analysis or similar, the acceptance criteria and study that proves the device meets them are focused on analytical performance (precision, sensitivity, linearity, interference) and clinical performance (diagnostic accuracy based on cut-points), rather than typical AI/ML metrics like AUC, sensitivity, specificity for image interpretation, or MRMC studies.

Therefore, many of the requested AI/ML specific information points (e.g., number of experts to establish ground truth for test set, adjudication method, MRMC study, sample size for training set, how ground truth for training set was established) are not directly applicable in the context of this IVD device's evaluation as described.

Here's an analysis based on the available information:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly present a single table of "acceptance criteria" alongside "reported device performance" for the clinical diagnostic accuracy in a pass/fail format. Instead, it provides detailed analytical performance results and clinical likelihood ratios for various patient subgroups. However, for analytical performance, conclusions are stated:

Analytical Performance Acceptance Criteria and Results:

Clinical Performance:
For clinical performance, the criterion is generally that the device provides "adequate performance when aiding in the diagnosis of acutely decompensated heart failure" and supports a "substantial equivalent decision" to the predicate. This is demonstrated through likelihood ratios (LR+ and LR-) for the various age and gender groups. While no explicit "acceptance criteria" for these LR values are given (e.g., LR+ > X and LR- < Y), the presentation of these values, along with post-test probabilities, is intended to show clinical utility and substantial equivalence.

2. Sample Size Used for the Test Set and Data Provenance

• Clinical Performance Test Set (ADHF Diagnosis Study):

  • Sample Size: 1485 subjects.
  • Data Provenance: Multi-center trial performed in the United States (US subjects).
  • Retrospective/Prospective: The description "multi-center trial was performed" suggests a prospective study design. Patients were enrolled with suspected ADHF and their ADHF status was subsequently adjudicated.

• Analytical Performance Test Sets:

  • Precision: 8 serum samples and 2 controls (n=84 determinations for each sample/control).
  • LoB: 60 determinations of an analyte-free sample.
  • LoD: 5 low-level human serum samples, 60 determinations per reagent lot (total of three reagent lots tested).
  • LoQ: 10 native, unaltered serum samples, 5 replicates per sample, over 5 days (n=25 measured values per sample).
  • Linearity: 10 levels of diluted high analyte human native serum.
  • Endogenous Interferences: Three different analyte concentration levels for serum samples.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

• Clinical Performance Test Set (ADHF Diagnosis Study):
  • Ground truth for ADHF diagnosis was established "based on adjudication by a clinical events committee."
  • The number of experts on this committee and their specific qualifications (e.g., radiologist with X years of experience) are not explicitly specified in the provided text.

4. Adjudication Method for the Test Set

• Clinical Performance Test Set:
  • Ground truth was based on "adjudication by a clinical events committee."
  • The specific adjudication method (e.g., 2+1, 3+1, majority vote, etc.) for this committee is not explicitly detailed in the document.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

• No, this is not applicable. The device is an in vitro diagnostic (IVD) test (immunoassay), not an AI-powered diagnostic imaging tool. Therefore, an MRMC study and the concept of "human readers improving with AI assistance" are not relevant to this type of device and its evaluation as described. The clinical performance evaluation focuses on the assay's ability to aid in diagnosis using predefined cut-points.

6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done

• Yes, in essence. The "standalone performance" of the Elecsys proBNP II and Elecsys proBNP II STAT assays is demonstrated by their analytical performance metrics (precision, LoB, LoD, LoQ, linearity, interference) and their diagnostic performance (likelihood ratios) when applied to patient samples. There isn't an "algorithm" in the typical AI sense, but rather a direct measurement of NT-proBNP concentration against established cut-points. The interpretation relies on these quantitative measurements.

7. The Type of Ground Truth Used (expert consensus, pathology, outcomes data, etc.)

• Clinical Performance Test Set (ADHF Diagnosis Study): The ground truth for Acute Decompensated Heart Failure (ADHF) status was established by adjudication of a clinical events committee. This implies a panel of medical experts reviewed patient data (e.g., symptoms, signs, other diagnostic tests, clinical course) to determine the true ADHF status. This falls under a form of expert consensus on clinical diagnosis.

8. The Sample Size for the Training Set

• Not Applicable in the AI/ML sense. This is an in vitro diagnostic immunoassay, not an AI/ML device that requires a separate "training set" to develop an algorithm. The assay's analytical characteristics and clinical utility are established through the studies mentioned (analytical and clinical performance evaluation). The "training" of the device is inherent in its chemical and biological design and manufacturing, not a machine learning process.

9. How the Ground Truth for the Training Set was Established

• Not Applicable. As explained in point 8, there isn't a "training set" for an AI/ML algorithm in the context of this immunoassay. The analytical performance is validated against known standards and controls, and clinical performance is validated against clinical diagnoses adjudicated by a committee.

Ask a specific question about this device