Search Results

The Human Influenza Virus Real-time RT-PCR Detection and Characterization Panel (rRT-PCR Flu Panel) is intended for use in Real-time RT-PCR assays on an AB1 7500 Fast Dx Real-time PCR instrument in conjunction with clinical and epidemiological information:

• for qualitative detection of influenza virus type A or B in symptomatic patients from viral RNA in nasopharyngeal and/or nasal swab specimens,
• for determination of the subtype of seasonal human influenza A virus, as seasonal A/H1 or A/H3, if present, from viral RNA in nasopharyngeal and/or nasal swab specimens,
• for presumptive identification of virus in patients who may be infected with influenza A subtype A/H5 (Asian lineage) from viral RNA in human respiratory specimens and viral culture in conjunction with clinical and epidemiological risk factors.
• to provide epidemiologic information for surveillance for influenza viruses.

The CDC Human Influenza Virus Real-time RT-PCR Detection and Characterization Panel (rRT-PCR Flu Panel) is a panel of oligonucleotide primers and dual-labeled hydrolysis (TagMan®) probes which may be used in real-time RT-PCR assays using the ABI 7500 Fast Dx Real-Time PCR instrument for the in vitro qualitative detection and characterization of human influenza viruses (RNA) in respiratory specimens from patients presenting with influenza-like illness (ILI). Detection of viral RNA not only aids in the diagnosis of illness caused by seasonal and novel influenza viruses in patients with ILI, but also provides epidemiologic information on influenza and aids in the presumptive laboratory identification of specific novel influenza A viruses.

Acceptance Criteria and Device Performance for CDC Human Influenza Virus Real-time RT-PCR Detection and Characterization Panel

1. Table of Acceptance Criteria and Reported Device Performance

Study Details:

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

• Clinical Performance (Seasonal Influenza - InfA, InfB, H1, H3):
  • Sample Size: 415 total prospective specimens.
  • Data Provenance: Prospective collection from US state public health laboratories during the 2006-2007 respiratory virus season (February-April).
  • Specimen Type: Nasal and nasopharyngeal swabs.
• Clinical Performance (Influenza A/H5N1):
  • Clinical Specimens: 24 retrospective A/H5 influenza samples.
  • Data Provenance: Retrospective from suspect positive cases received at CDC, "diverse geographic locations."
  • Cultured Specimens: Not explicitly stated sample size, but also used for H5 detection with 100% agreement.
  • Prospective Clinical Specimens (for negative agreement): Implied to be part of the 415 prospective seasonal specimens, where no H5 was detected.

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

• No specific number or qualifications of experts are explicitly mentioned for establishing ground truth.
• The reference method for prospective seasonal influenza samples was "rapid culture (shell vial) followed by direct fluorescent antibody screening and identification." This implies laboratory personnel with expertise in virology and cell culture.
• Influenza A/H5 and discrepant prospective seasonal influenza samples were "further analyzed by bidirectional sequencing," suggesting molecular biology experts.

4. Adjudication Method for the Test Set:

• No formal adjudication method (e.g., 2+1, 3+1) is explicitly described.
• The ground truth for prospective seasonal influenza was based on rapid culture (shell vial) followed by direct fluorescent antibody screening and identification.
• Discrepant prospective seasonal influenza samples and influenza A/H5 samples were further analyzed by bidirectional sequencing. This serves as an additional, higher-resolution method for resolving discrepancies or confirming difficult cases.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done:

• No, an MRMC comparative effectiveness study was not done. This device is a diagnostic assay (RT-PCR panel) for detecting and characterizing influenza viruses, not an imaging device or a device that requires human interpretation in the same way an AI-powered image analysis tool would. Therefore, the concept of "human readers improve with AI vs. without AI assistance" does not apply here.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop) Performance Was Done:

• Yes, the performance presented for the rRT-PCR Flu Panel is standalone (algorithm only, without human-in-the-loop performance). This device is a molecular diagnostic kit, and its performance metrics (sensitivity, specificity, LoD, etc.) reflect the intrinsic analytical and clinical capabilities of the assay itself when executed according to protocol. Human involvement is in sample preparation, running the assay, and interpreting the raw output (e.g., Ct values), but the "diagnosis" of the presence and type of virus is determined by the output of the PCR reaction.

7. The Type of Ground Truth Used:

• Expert Consensus and "Gold Standard" Laboratory Methods:
  • For prospective seasonal influenza, the ground truth was established by gold standard virus culture testing (rapid culture (shell vial) followed by direct fluorescent antibody screening and identification).
  • For influenza A/H5 and discrepant prospective seasonal influenza samples, the ground truth was further analyzed and confirmed by bidirectional sequencing. This suggests a form of expert-validated, highly accurate laboratory confirmation.

8. The Sample Size for the Training Set:

• Not Applicable / Not Explicitly Stated for a "training set" as understood in machine learning. This device is a molecular diagnostic panel based on designed primers and probes, not a machine learning model that undergoes explicit training. The "development" and "optimization" would have involved testing various primer/probe combinations against known virus strains, but this isn't analogous to a "training set" for an AI algorithm.
• However, for Analytical Specificity (Inclusivity), the document mentions initial testing with "ten (10) influenza virus strains of A/H/N1, A/H3N2, and influenza B" and "24 influenza A/H5N1 clinical samples tested retrospectively" to demonstrate the panel's ability to detect different strains. This could be considered akin to an internal validation set during development.

9. How the Ground Truth for the Training Set Was Established:

• As above, the concept of a "training set" for this type of medical device (RT-PCR panel) is not directly applicable in the same way it would be for an AI algorithm.
• The "ground truth" for the various influenza virus strains and non-influenza pathogens used in analytical studies would have been primarily established through known viral cultures, characterized bacterial/yeast cultures, and molecular sequencing (e.g., 16S ribosomal RNA bi-directional sequencing for bacteria, bi-directional sequencing for non-influenza respiratory viruses), all validated by standard laboratory practices.

Ask a specific question about this device