(165 days)
The Shield Activated Factor XI I (FXIIa) assay is a semi-quantitative direct enzyme immunoassay for the detection, in human citrated plasma, of activated factor X11 (x-X1 la and B- X11a). The test may be used as an index of activation of the contact (intrinsic) system of coagulation
ACTIVATED FACTOR X11 (FX11a) ELISA TEST
The document provided is a 510(k) clearance letter from the FDA for an in vitro diagnostic device (ELISA assay), not an AI/ML medical device. Therefore, much of the information requested in your prompt regarding acceptance criteria and studies relevant to AI/ML devices (e.g., sample sizes for test sets, data provenance, number of experts for ground truth, adjudication methods, MRMC studies, standalone performance, training set details) is not applicable to this type of document.
The document primarily states that the device is "substantially equivalent" to legally marketed predicate devices.
However, I can extract the following information that is somewhat related to your request:
1. A table of acceptance criteria and the reported device performance
This document does not contain a table of acceptance criteria or reported device performance in the manner you'd expect for an AI/ML device (e.g., sensitivity, specificity, AUC). For an in vitro diagnostic device like an ELISA, the "performance" would typically be evaluated against predicate devices, ensuring it performs comparably in terms of analytical sensitivity, specificity, precision, accuracy, etc. These specific metrics are not detailed in this FDA clearance letter. The letter simply states the device is "substantially equivalent," implying these tests were performed and deemed acceptable by the FDA to reach that conclusion.
2. Sample sized used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)
- Not explicitly stated in this document. For an in vitro diagnostic device, performance studies typically involve patient samples, but the number of samples, their origin, or whether the study was retrospective or prospective is not provided in this regulatory letter.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)
- Not applicable / Not explicitly stated for this type of device. The "ground truth" for an ELISA assay would usually be established by comparing its results to a reference method or clinical diagnosis based on established laboratory and clinical procedures, not by expert consensus on images.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
- Not applicable / Not explicitly stated. Adjudication methods are typically associated with expert review of images or clinical cases for AI/ML ground truth, which is not relevant here.
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. Not applicable. MRMC studies are specific to AI-assisted interpretation, which this device is not.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
- Yes, implicitly. An ELISA assay is by its nature a "standalone" test. It generates a result based on the sample reacting with reagents. There is no "human-in-the-loop" once the assay is run and interpreted (though a human performs the test and reads the result). However, this concept is different from standalone performance of an AI algorithm.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)
- Not explicitly stated. For an ELISA, ground truth regarding the presence of Activated Factor XII (FXIIa) would likely be established either through:
- Comparison to a gold standard laboratory method (if one exists).
- Correlation with clinical outcomes or diagnosis in patient populations, where the clinical diagnosis serves as the "truth."
- Comparison to an established predicate device.
8. The sample size for the training set
- Not applicable / Not explicitly stated. ELISA assays are not "trained" in the way AI/ML algorithms are. They operate based on biochemical reactions.
9. How the ground truth for the training set was established
- Not applicable. As stated above, ELISA assays are not "trained."
Summary regarding this specific document:
This FDA letter is a 510(k) clearance for an In Vitro Diagnostic (IVD) device, specifically an ELISA assay. The criteria relevant to this type of device are primarily about its equivalence to a predicate device in terms of analytical and clinical performance. The letter confirms substantial equivalence, allowing the device to be marketed. It does not provide the detailed performance metrics or study designs typically requested for AI/ML device submissions.
§ 864.7290 Factor deficiency test.
(a)
Identification. A factor deficiency test is a device used to diagnose specific coagulation defects, to monitor certain types of therapy, to detect coagulation inhibitors, and to detect a carrier state (a person carrying both a recessive gene for a coagulation factor deficiency such as hemophilia and the corresponding normal gene).(b)
Classification. Class II (performance standards).