(61 days)
The G5 I HbA₁c assay is an affinity chromatography method and is intended for the in-vitro quantitative determination of HbA₁c in capillary blood taken from a finger prick or whole blood in EDTA. The test is indicated for use by diabetics for monitoring the time averaged blood glucose levels of known diabetics as an indicator of overall glycaemic control. The G5 I HbA₁c assay is intended for use in a physicians/doctors office. The assay is not intended for use as a home use or for self-testing.
The G5 II HbA1c assay is an affinity chromatography method and is intended for the in-vitro quantitative determination of HbA1c in capillary blood taken from a finger prick. The test is indicated for use by diabetics for monitoring the time averaged blood glucose levels of known diabetics as an indicator of overall glycaemic control. The G5 II HbA1c assay is intended for use as a prescription home use test.
Instrument read, single use in vitro test for the quantitative determination of glycated haemoglobin (GHb) in diabetics.
Here's an analysis of the G5 I HbA1c and G5 II HbA1c Test Systems based on the provided document, addressing the requested criteria:
1. Table of Acceptance Criteria and Reported Device Performance
| Performance Metric | Acceptance Criteria (Implied / Stated in document) | Reported Device Performance |
|---|---|---|
| Clinical Equivalence | Correlation coefficient to predicate device (Glycosal assay) | - Fresh finger prick: 0.96 |
- Stored EDTA blood: 0.97 |
| POL Study - Accuracy | Accuracy within ± 10% | Achieved within ± 10% (for G5 I) |
| POL Study - Precision | CV less than 4.6% | Achieved less than 4.6% (for G5 I) |
| POL Study - Correlation | Correlation coefficients of >0.95 | Achieved >0.95 (for G5 I) |
| Linearity | Linear over the assay range | Linear between 6% and 14% HbA1c |
| Reproducibility (Overall CV) | Overall CV precision of less than 5% | Achieved less than 5% |
| Reproducibility (Inter-batch CV) | Assay %CVs of less than 5% (intra and inter-batch) | Achieved less than 5% |
| Haemoglobin Range | Performs acceptably within a specified range | Acceptable within 11-18 g/dl |
| Haematocrit Range | Performs acceptably within a specified range | Acceptable within 35% to 55% |
| Cartridge Stability | Stable for a specified duration | Stable for at least 12 months at 2-8°C |
| Equilibration Time | Minimum time required to equilibrate to room temperature | 2 hours from 2-8°C |
| Sample Insertion Time | Acceptable time after sample addition prior to insertion into reader | Must be inserted immediately after sample addition |
2. Sample Size Used for the Test Set and Data Provenance
- Clinical Trial (for G5 HbA1c - likely refers to both G5 I and G5 II):
- Sample Size: 74 patients.
- Data Provenance: Southport DGH, UK (retrospective due to using "fresh finger prick and stored EDTA blood" which implies samples were collected then analyzed/compared).
- POL Studies (G5 I):
- Sample Size: At least 20 blood samples per site (across 3 sites). This means a minimum of 60 unique blood samples were used. Additionally, 5 standards were run in triplicate at each site.
- Data Provenance: Not explicitly stated, but "Physician Office Laboratory (POL) studies" suggest prospective collection in a clinical setting.
- Home Use - Consumer Study (G5 II):
- Sample Size: Not explicitly stated, but involved "untrained subjects" and "trained subjects" at 3 separate sites. No specific number of subjects or samples is given.
- Data Provenance: 3 separate sites in the USA. Likely prospective, as it involved untrained subjects using the packaging.
- Non-Clinical Laboratory Studies (Linearity, Haemoglobinopathies, Abnormal blood chemistries, Interfering Drugs, Labile HbA1c, Reproducibility, Inter-batch variation, Stored blood, Total Haemoglobin/Haematocrit, Cartridge Stability, Equilibration Time, Sample Insertion Time):
- Sample Size: Not explicitly stated for each individual study beyond "a study was conducted" or "a study was performed." The reproducibility studies mention "a normal and an abnormal control" and "3 %HbA1c standards."
- Data Provenance: In-house laboratory studies by Provalis Diagnostics. The references for Haemoglobinopathies and Labile HbA1c point to an external scientific publication (Ann Clin Biochem 1997; 34: 17-31 by WG John), indicating reliance on established scientific understanding rather than a direct study conducted by the company for these specific aspects.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
The document does not explicitly state the number of experts used to establish ground truth or their qualifications for any of the studies.
- For the clinical trial, the "Glycosal assay" is used as the comparative method, implying it serves as the reference standard. The document doesn't detail the ground truth method for the Glycosal assay itself, but it is implied to be a legally marketed predicate device.
- The POL studies involve "trained operators," but their expert qualifications or role in defining ground truth (beyond running the predicate device) are not specified.
4. Adjudication Method for the Test Set
The document does not mention any formal adjudication method (e.g., 2+1, 3+1) for establishing ground truth in the test sets. The primary method of comparison is against a predicate device (Glycosal assay), implying that the result from the predicate device serves as the reference for determining accuracy and correlation.
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, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. This device is an in-vitro diagnostic (IVD) assay for quantitative determination of HbA1c, not an AI-assisted imaging device or a diagnostic that relies on human reader interpretation of complex data that could be augmented by AI. The studies evaluate the device's performance against a predicate device and its ability to be used by trained/untrained operators.
6. If a Standalone (i.e. algorithm only, without human-in-the-loop performance) was done
This question is not directly applicable in the context of this device. The G5 HbA1c Test System is a hardware/reagent-based IVD device with an instrument reader. Its "standalone performance" is essentially what is evaluated in the reproducibility, linearity, and interference studies – how the instrument/assay performs on its own when given a sample. There isn't a separate "algorithm only" component that would be evaluated distinctly from the physical device's performance. The "human-in-the-loop" aspect relates to the operators performing the test (trained vs. untrained), which was assessed in the POL and Home Use studies.
7. The Type of Ground Truth Used
The primary type of "ground truth" used for performance evaluation is comparison against a legally marketed predicate device (Glycosal HbA1c Test / Glycosal II HbA1c Test).
- For the clinical study, the G5 HbA1c assay was compared to the Glycosal assay.
- For the POL studies, performance was evaluated by comparing results from the G5 I HbA1c test with results obtained by a trained operator, likely using a reference method or the predicate.
- For the Home Use study, untrained subjects' results were compared to trained operators' results, implying the trained operator's result (presumably using the G5 II and/or a reference method) served as a benchmark.
8. The Sample Size for the Training Set
The document does not provide details on a specific "training set" in the context of machine learning or AI. As an IVD assay, it's developed and validated using traditional analytical and clinical performance studies, not typically "trained" in the way an AI algorithm is.
9. How the Ground Truth for the Training Set Was Established
Since there is no "training set" in the AI/ML context for this device, this question is not applicable. The development of the assay likely involved internal R&D validation using various samples with known HbA1c levels established by reference methods, but this is not described as a formal "training set" with ground truth establishment in the provided information.
§ 864.7470 Glycosylated hemoglobin assay.
(a)
Identification. A glycosylated hemoglobin assay is a device used to measure the glycosylated hemoglobins (A1a , A1b , and A1c ) in a patient's blood by a column chromatographic procedure. Measurement of glycosylated hemoglobin is used to assess the level of control of a patient's diabetes and to determine the proper insulin dosage for a patient. Elevated levels of glycosylated hemoglobin indicate uncontrolled diabetes in a patient.(b)
Classification. Class II (performance standards).