Search Results

EliA Ro52 is intended for the in vitro semi-quantitative measurement of IgG antibodies directed to Ro52 in human serum as an aid in the diagnosis of Sjögren's syndrome (SS), systemic lupus erythematosus (SLE), systemic sclerosis (SSc) and idiopathic inflammatory myopathies (IIM) in conjunction with other laboratory and clinical findings. EliA Ro52 uses the EliA IgG method.

EliA Ro60 is intended for the in vitro semi-quantitative measurement of IgG antibodies directed to Ro60 in human serum as an aid in the diagnosis of Sjögren's syndrome (SS) and systemic lupus erythematosus (SLE) in conjunction with other laboratory and clinical findings. EliA Ro60 uses the EliA IgG method.

The EliA Ro52 and EliA Ro60 Immunoassays are semi-quantitative solid-phase fluoroenzyme immunoassays, for the determination of autoantibodies against SS-A/Ro 52 kDa and 60 kDa proteins. The EliA Ro52 and EliA Ro60 test System is a fully integrated and automated system composed of assay-specific reagents, EliA method-specific reagents, and general reagents.

The provided document describes the EliA Ro52 and EliA Ro60 Immunoassays, which are intended for the semi-quantitative measurement of IgG antibodies directed to Ro52 and Ro60 in human serum, respectively. These devices aid in the diagnosis of specific autoimmune diseases. The document includes detailed analytical and clinical performance data to support the substantial equivalence claim.

Here's an analysis of the acceptance criteria and the study that proves the device meets them:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly state formal "acceptance criteria" in a tabulated format. Instead, it presents various analytical and clinical performance metrics, implying that these metrics, when compared to the predicate device and established clinical utility, are deemed acceptable for the device's intended use. Based on the provided performance data, the implicit acceptance criteria would likely be related to aspects like precision, linearity, detection limits, specificity (interference), and clinical sensitivity/specificity.

Below is a table summarizing key performance indicators reported in the document:

Table 1: Key Performance Metrics for EliA Ro52 and EliA Ro60

The acceptance of these performance metrics is implicitly based on comparison with the predicate devices (QUANTA Flash Ro52 and QUANTA Flash Ro60) and the established clinical utility in diagnosing the mentioned autoimmune diseases. The FDA's substantial equivalence determination (K210902) confirms that the submitted data supports the claim that the new devices perform comparably to the predicate devices.

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

The document describes several test sets for different performance characteristics:

• Precision/Reproducibility:
  • Phadia 250: 5 samples, each with 252 replicate determinations (across 3 instruments, 7 days, 21 runs).
  • Within-lab Imprecision: Samples in 80 replicates on 1 instrument over 20 days (40 runs).
  • Phadia 2500 and Phadia 5000 series (E-module): 5 samples, each with 84 replicates (across 3 instruments, 7 days, 21 runs).
• Linearity/Assay Reportable Range: 3 patient serum samples serially diluted in at least 9 dilution steps, tested in quadruplicates.
• Detection Limit: 4 blank and 4 low-level samples, tested in 5-fold determination, across 2 different reagent sets (totaling 120 combined observations for blank and low-level per instrument type).
• Analytical Specificity (Interference): 3 serum samples (negative, equivocal, high positive), spiked with interfering substances, tested in triplicates.
• Method Comparison with Predicate Device:
  • EliA Ro52: 208 patient serum samples (only 181 used in agreement calculations due to samples outside measuring range).
  • EliA Ro60: 208 patient serum samples (only 104 used in agreement calculations due to samples outside measuring range).
• Clinical Sensitivity and Specificity:
  • EliA Ro52: 755 clinically and ethnically defined serum samples. These include:
    • Diagnostic group: SLE (n=120), Sjögren's syndrome (n=60), IIM (n=94), SSc (n=91) - total 365.
    • Disease control group: 390 samples (various autoimmune and infectious diseases).
  • EliA Ro60: 713 clinically and ethnically defined serum samples. These include:
    • Diagnostic group: SLE (n=120), Sjögren's syndrome (n=60) - total 180.
    • Disease control group: 533 samples (various autoimmune and infectious diseases, excluding SSc).

Data Provenance:
The document states that the clinical samples used for sensitivity and specificity determination were from "clinically and ethnically defined serum samples, including those of US origin." This indicates a mix of retrospective (clinically defined, likely banked samples) and potentially prospective (if US origin implies some form of fresh collection for the study) data. It is not explicitly stated whether it was purely retrospective or involved prospective collections, but the description "clinically and ethnically defined serum samples" strongly suggests retrospective use of samples with established diagnoses.

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

The document does not provide information on the number of experts used to establish the ground truth for the test sets (e.g., for diagnoses of SLE, SS, IIM, SSc). It only states that samples were "clinically and ethnically defined." This typically implies that patient diagnoses were established through standard clinical practice by medical professionals, but the specifics of these experts' qualifications or the consensus process are not detailed in this submission summary.

4. Adjudication Method for the Test Set

The document does not specify an adjudication method for establishing the ground truth for the test set. It refers to diagnoses as "clinically defined," which generally suggests diagnosis was made by treating physicians following clinical guidelines, but no adjudication process like 2+1 or 3+1 by independent experts is mentioned for the study.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. This document describes an in vitro diagnostic (IVD) device, specifically an immunoassay for measuring antibody levels. MRMC studies are typically performed for imaging devices or other diagnostic tools where human interpretation is involved and needs to be compared with and/or augmented by AI. This device is an automated laboratory test, and its performance is evaluated directly through analytical measurements and comparison to a predicate device, as well as clinical sensitivity and specificity against established clinical diagnoses.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done

Yes, this is a standalone device. The EliA Ro52 and EliA Ro60 Immunoassays are "automated semi-quantitative solid phase fluoroenzymeimmunoassays" run on "Phadia 250 instrument and the Phadia 5000 instrument series (E-modules)." The results are automatically converted to EliA U/mL. The performance characteristics described (precision, linearity, detection limit, analytical specificity, clinical sensitivity, specificity) are all based on the output of the automated instrument and reagents, independent of human interpretation during the actual test process itself. The interpretation of results (negative, equivocal, positive) is based on defined cut-off values.

7. The Type of Ground Truth Used (Expert Consensus, Pathology, Outcomes Data, etc.)

For the clinical sensitivity and specificity studies, the ground truth was clinical diagnosis. The samples were "clinically and ethnically defined serum samples" with established diagnoses of various autoimmune diseases (SLE, SS, IIM, SSc) and control conditions. This implies that the diagnosis was based on the standard clinical criteria and medical records established by healthcare professionals, rather than a separate expert consensus panel or specific pathology results for the study.

8. The Sample Size for the Training Set

The document does not describe a "training set" in the context of machine learning or AI models, as this is an immunoassay and not an AI/ML-based device.

However, if "training set" is generally interpreted as samples used for assay development, optimization, or establishing parameters like cut-offs, the relevant information is:

• Cut-off definition: A cohort of 69 healthy blood donors, 19 SLE patients, and 9 Sjögren's syndrome (SS) patients for EliA Ro52, and 70 healthy blood donors, 22 SLE patients, and 6 Sjögren's syndrome patients for EliA Ro60, were used to define the cut-off values. This can be considered the 'development' or 'calibration' set for determining the interpretive thresholds.

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

As noted above, this device does not utilize a machine learning "training set" in the conventional sense. For the samples used to establish the cut-off values (which could be considered analogous to determining a 'decision boundary' in an ML context), the ground truth was established based on the clinical diagnoses of the patient samples (healthy blood donors, SLE patients, Sjögren's syndrome patients). The document states that the initial cut-off for EliA Ro52, set using SLE and SS patients, was later verified with additional IIM and SSc patient sera. This indicates clinical diagnosis as the gold standard.

Ask a specific question about this device

EliA Rib-P is intended for the in vitro semi-quantitative measurement of IgG antibodies directed to Rib-P in human serum as an aid in the diagnosis of systemic lupus erythematosus (SLE) in conjunction with other laboratory and clinical findings. EliA Rib-P uses the EliA IgG method.

EliA Rib-P is a semi-quantitative solid-phase fluoroenzymeimmunoassay, for the determination of autoantibodies against Rib-P. The EliA Rib-P test System is fully integrated and automated system which comprises of assay-specific reagents, EliA method-specific reagents, and general reagents.

Assav-Specific Reagents include:

• EliA Rib-P Wells: coated with human recombinant ribosomal P-proteins P0, P1 . and P2 - 2 carriers (12 wells each), ready to use;
• . EliA ANA 3 Positive Control 250 or 2500/5000: Human monoclonal antibodies in Tris buffer containing IgG antibodies to Ro52, Rib-P and RNA Pol III – 6 single use vials, 0.3 mL each, ready to use;
• . EliA IgG/IgM/IgA Negative Control 250 or 2500/5000: Human blood preparation from healthy donors in PBS containing BSA, detergent and 0.095% sodium azide - 6 single-use vials, 0.3 mL each, ready to use;

EliA Method-Specific Reagents include:

• EliA Sample Diluent: PBS containing BSA, detergent and 0.095% sodium azide . - 6 bottles, 48 mL each, ready to use; or 6 bottles, 400 mL each, ready to use;
• I EliA IqG Conjuqate 50 or 200: ß-Galactosidase labeled anti-lgG (mouse monoclonal antibodies) in PBS containing BSA and 0.06% sodium azide - 6 wedge shaped bottles, 5 mL each, ready to use; or 6 wedge shaped bottles, 19 mL each, ready to use:
• . EliA IgG Calibrator Strips: Human IqG (0, 4, 10, 20, 100, 600 uq/L) in PBS containing BSA, detergent and 0.095% sodium azide - 5 strips, 6 single-use vials per strip, 0.3 mL each, ready to use;
• EliA IgG Curve Control Strips: Human IgG (20 ug/L) in PBS containing BSA, . detergent and 0.095% sodium azide - 5 strips, 6 single-use vials per strip, 0.3 mL each, ready to use:
• EliA IgG Calibrator Well: coated with mouse monoclonal antibodies 4 carriers . (12 wells each), ready to use;

General Reagents include:

• Development Solution: 0.01% 4-Methylumbelliferyl-β-D-galactoside, 110, 6x >170, or 6x >1165 determinations;
• I Stop Solution: 4% Sodium Carbonate - 6 bottles (65 mL, 119 mL, or 2800 mL each), sufficient for 6x >292, 6x >560, or 6x >13100 determinations;
• 트 Washing Solution Additive: detergent, preservative

The provided document is a 510(k) Summary for the EliA Rib-P device, an in vitro semi-quantitative immunoassay for detecting IgG antibodies directed to Rib-P as an aid in diagnosing Systemic Lupus Erythematosus (SLE). This document does not describe a study involving an AI-enabled device or human-in-the-loop performance. Instead, it focuses on the analytical and clinical performance of a standalone laboratory diagnostic device and its substantial equivalence to a predicate device.

However, I can extract and structure the information related to acceptance criteria and the performance study of this specific diagnostic device in the format requested, even without AI-specific parameters.

Acceptance Criteria and Device Performance Study for EliA Rib-P

The EliA Rib-P device is an automated semi-quantitative solid phase fluoroenzymeimmunoassay for the measurement of IgG antibodies directed to Rib-P in human serum. This summary outlines the performance characteristics tested to demonstrate its substantial equivalence to a legally marketed predicate device, the Quanta Lite Ribosome P ELISA.

1. Table of Acceptance Criteria and the Reported Device Performance

• Range of Total Imprecision %CV across 5 samples: 4.0% - 13.3%
• Within-lab Imprecision %CV across 4 samples: 3.5% - 13.8%
  Phadia 2500/5000:
• Range of Total Imprecision %CV across 5 samples: 5.7% - 18.4% (one sample had high CV, 18.4%) |
  | Linearity/Reportable Range| Coefficient of determination (R²) close to 1.00, and slopes close to 1.00, across the measuring range. CLSI EP6-A guidelines followed. | Phadia 250:
• R² values: 1.00 for all three dilution ranges.
• Slopes: 0.95, 1.00, 1.00.
  Phadia 2500E:
• R² values: 0.99, 1.00, 0.99 for the three dilution ranges.
• Slopes: 1.03, 1.00, 1.02.
  "Linearity was shown for the entire measuring range." |
  | Detection Limit | LoB, LoD, and LoQ to be established according to CLSI EP17-A2 guidelines (false positives and false negatives 10 EliA U/mL |
  | Comparison Studies | | |
  | Method Comparison (vs. Predicate Device) | High Positive Percent Agreement (PPA), Negative Percent Agreement (NPA), and Total Agreement with the predicate device. CLSI EP09c-ED3 followed. | EliA Rib-P (equivocal considered negative):
• PPA: 94.7% (95% CI: 82.3% - 99.4%)
• NPA: 98.6% (95% CI: 96.4% - 99.6%)
• Total Agreement: 98.1% (95% CI: 96.0% - 99.3%)
  EliA Rib-P (equivocal considered positive):
• PPA: 100% (95% CI: 90.7% - 100%)
• NPA: 88.1% (95% CI: 83.7% - 91.6%)
• Total Agreement: 89.5% (95% CI: 85.6% - 92.6%) |
  | Instrument Comparison | Strong correlation and agreement between different Phadia instrument series. | Regression analysis showed a slope of 0.94 (95% CI: 0.93 - 0.98) and an intercept of -0.76 (95% CI: -1.20 - -0.48) between Phadia 250 and Phadia 2500E. |
  | Clinical Studies | | |
  | Clinical Sensitivity and Specificity | Acceptable sensitivity for SLE and high specificity against other autoimmune and infectious diseases. | EliA Rib-P (equivocal considered positive):
• Sensitivity (for SLE): 34.9% (95% CI: 27.2% - 43.3%)
• Specificity (against disease controls): 99.3% (95% CI: 97.9% - 99.9%)
  EliA Rib-P (equivocal considered negative):
• Sensitivity (for SLE): 28.1% (95% CI: 21.0% - 36.1%)
• Specificity (against disease controls): 99.8% (95% CI: 98.7% - 100%) |

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

• Precision/Reproducibility:
  • Phadia 250: 5 samples, tested in 252 replicates each across 3 lots and 3 instruments over 7 days.
  • Within-Lab Imprecision: 4 samples, tested in 80 replicates each on 1 instrument over 20 days.
  • Phadia 2500/5000 (E-module): 5 samples, tested in 84 replicates each on 3 instruments over 7 days.
  • Provenance: Not explicitly stated, typically laboratory-prepared controls or banked human serum samples.
• Linearity/Assay Reportable Range: 3 serum samples (diluted). Provenance not explicitly stated.
• Detection Limit: 4 blank and 4 low-level samples (depleted IgG sera and prepared low-level samples) tested in 5-fold determination across 3 runs on Phadia 250 and Phadia 2500E.
• Analytical Specificity (Interference): 3 serum samples (one negative, one equivocal, one high positive) spiked with interfering substances. Provenance not explicitly stated.
• Assay Cut-Off: A cohort of 70 apparently healthy blood donors and 30 samples from SLE patients. Provenance not explicitly stated.
• Method Comparison with Predicate Device: 323 patient samples. Provenance not explicitly stated, but implies diverse clinical samples covering the measuring range.
• Instrument Comparison: 47 positive, 10 equivocal, and 28 negative samples. Provenance not explicitly stated.
• Clinical Sensitivity and Specificity: 560 clinically defined serum samples from various diagnostic groups (146 SLE, 414 disease controls). Provenance not explicitly stated, but these are patient samples with confirmed diagnoses.
• Expected Values/Reference Range: 638 apparently healthy subjects, equally distributed by age and gender from Caucasian, African American, Hispanic, and Asian populations obtained from a blood bank.

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

The document does not mention the use of experts to establish ground truth for the analytical performance characteristics. For clinical studies, the "ground truth" for patient samples was based on "clinically defined serum samples with a diagnosis" (e.g., systemic lupus erythematosus, Celiac disease, etc.). This typically implies diagnosis by medical professionals (e.g., rheumatologists, gastroenterologists, etc.) based on established diagnostic criteria, but the number and specific qualifications of these experts are not explicitly stated in this summary.

4. Adjudication Method (e.g. 2+1, 3+1, none) for the Test Set

No explicit adjudication method is mentioned. The ground truth for clinical samples appears to be based on pre-existing clinical diagnoses.

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

This section is not applicable as the device is a standalone in vitro diagnostic immunoassay, not an AI-enabled device or an assist device for human readers. No MRMC study was conducted.

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

This device is a standalone algorithm/device in the sense that it performs automated semi-quantitative measurement of antibodies without real-time human interpretation impacting the measurement result. The assay directly measures the amount of antibody via fluorescence. The interpretation of results (negative, equivocal, positive) is based on predefined cut-offs.

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

• Analytical ground truth: Based on reference materials, spiked samples, and dilution series with known concentrations or characteristics.
• Clinical ground truth: "Clinically defined serum samples with a diagnosis" (e.g., SLE patients, various disease controls). This implies established clinical diagnoses, likely based on standard diagnostic criteria, which could involve expert clinical assessment, pathology, and other laboratory findings.

8. The Sample Size for the Training Set

This document does not specify a separate "training set" in the context of machine learning or AI. For a traditional immunoassay, method development and optimization would involve various experiments and sample sets, but these are not explicitly termed "training sets" here. The "Assay Cut-Off" study used 70 healthy blood donors and 30 SLE patients to define the cut-off, which could be considered a form of "training" for the interpretive criteria.

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

As noted above, a formal "training set" in the AI sense is not described. For the cut-off determination, the ground truth was established by using "apparently healthy blood donors" and "samples from SLE patients," indicating that these samples had known clinical statuses (healthy or diagnosed with SLE).

Ask a specific question about this device

EliA M2 is intended for the in vitro semi-quantitative measurement of IgG antibodies directed to M2 in human serum and plasma (Li-heparin, EDTA) to aid in the clinical diagnosis of primary biliary cirrhosis in conjunction with other laboratory and clinical findings. EliA M2 uses the EliA IgG method on the instrument Phadia 2500/5000.

The method-specific reagents are identical with K141375 (EliA M2 on Phadia 250), but are filled in containers specific for the Phadia 2500/5000 instrument. Each device consists of:

• Test Wells: -EliA M2 Wells are coated with native pyruvate dehydrogenase complex from mitochondria and recombinant M2-antigen - 4 carriers (12 wells each), ready to use:
• EliA Sample Diluent: PBS containing BSA, detergent and 0.095% (w/v) sodium azide - 6 bottles, 48 mL each, ready to use; or 6 bottles, 400 mL each, ready to use;
• EliA IgG Conjugate 50 or 200: ß-Galactosidase labeled anti-IgG (mouse monoclonal antibodies) in PBS containing BSA and 0.06% (w/v) sodium azide – 6 wedge shaped bottles, 5 mL each, ready to use; or 6 wedge shaped bottles, 19 mL each, ready to use
• EliA IgG Calibrator Strips: Human IgG (0, 4, 10, 20, 100, 600 µg/L) in PBS containing BSA, detergent and 0.095% (w/v) sodium azide - 5 strips, 6 singleuse vials per strip, 0.3 mL each, ready to use;
• -EliA IgG Curve Control Strips: Human IgG (20 ug/L) in PBS containing BSA, detergent and 0.095% (w/v) sodium azide - 5 strips, 6 single-use vials per strip, 0.3 mL each, ready to use;
• EliA IgG Calibrator Well: Coated with mouse monoclonal antibodies 4 carriers (12 wells each), ready to use.
  The Phadia EliA Immunodiagnostic System is an automated system for immunodiagnostic testing. The EliA reagents are available as modular packages, each purchased separately. All packages are required to carry out EliA M2 tests.

The provided text describes the 510(k) premarket notification for the EliA™ M2 Immunoassay, specifically focusing on its performance on new instrument platforms (Phadia® 2500/5000) compared to a previously cleared device (EliA M2 on Phadia 250 instrument, K141375).

This document is for an in vitro diagnostic (IVD) device, which measures IgG antibodies to M2 to aid in the clinical diagnosis of primary biliary cirrhosis. The "device" in this context is the immunoassay system, which includes the reagents and the instrument platforms.

Therefore, the acceptance criteria and study details discussed pertain to the analytical performance of this IVD device. The concepts of "human readers," "expert consensus for image interpretation," "MRMC," and "standalone algorithm performance" are not applicable here, as this is not an AI/ML imaging device or a device requiring human interpretation of complex visual data. The "ground truth" for an IVD diagnostic test is typically established through reference methods, established clinical diagnoses, or highly characterized samples.

Here's a breakdown of the acceptance criteria and study proof based on the provided text, focusing on the analytical performance of the immunoassay system.

Acceptance Criteria and Reported Device Performance

The core of the "acceptance criteria" for this 510(k) submission is demonstrating substantial equivalence to the predicate device. This is achieved by showing that the analytical performance characteristics of the EliA M2 immunoassay on the Phadia 2500/5000 instruments are comparable and acceptable. The document highlights various analytical performance parameters with their respective results.

1. Table of Acceptance Criteria and Reported Device Performance

Since this is an IVD device, the "acceptance criteria" aren't explicitly stated as pass/fail thresholds in a table, but rather implied by the data presented for substantial equivalence. The document presents the study results for various analytical performance metrics.

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

• Precision/Reproducibility:
  • Sample Size: Each sample was tested in four replicates/run, over 7 days, on 3 instruments. This totals 84 replicates per sample (4 replicates * 7 days * 3 instruments = 84 replicates). The document doesn't specify the number of distinct "samples" (control materials or patient samples) used to establish precision at different concentration levels, but it shows results for 5 different mean concentration levels (1.7, 4.0, 5.9, 74.8, 175.9 U/mL). Typically, these would be control materials or pooled patient samples.
  • Data Provenance: Not explicitly stated, but clinical studies for cut-off determination (K141375) and reference ranges mentioned a "Caucasian population obtained from a blood bank." For analytical studies like precision, standard reference materials or well-characterized internal controls are commonly used. The study was performed over 7 days.
• Linearity/Assay Reportable Range:
  • Sample Size: Four patient serum samples were diluted.
  • Data Provenance: Not explicitly stated regarding country of origin or retrospective/prospective nature for these specific samples.
• Detection Limit (LoB, LoD, LoQ):
  • Sample Size:
    • LoB: One blank sample, measured in twelve replicates in each of six runs (72 determinations total).
    • LoD: Five low-level samples, measured in twelve replicates in each of six runs (360 low level replicates total).
    • LoQ: 360 determinations.
  • Data Provenance: Not explicitly stated.
• Method Comparison (Instrument Comparison):
  • Sample Size: More than 100 samples.
  • Data Provenance: Not explicitly stated regarding country of origin or retrospective/prospective nature. Samples were run in "single replicates" on one Phadia 250 and one Phadia 2500/5000 instrument.

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

• Not applicable. This is an in vitro diagnostic (IVD) immunoassay system, not an AI/ML imaging device that requires human expert interpretation of visual data. The "ground truth" for IVD laboratory tests is typically established by:
  • Quantitative values: Directly measured by reference methods or highly characterized (e.g., gravimetric, spectrophotometric) standards traceability.
  • Qualitative results (positive/negative): Defined by cut-off values derived from clinical studies, often comparing against a gold standard diagnostic method (e.g., biopsy, established clinical diagnosis of PBC).

4. Adjudication Method for the Test Set

• Not applicable. This concept (e.g., 2+1, 3+1) relates to consensus reading in imaging studies. For an IVD assay, results are quantitative values or interpreted based on pre-defined cut-offs.

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

• Not applicable. This is an IVD immunoassay device, not an AI-assisted diagnostic imaging system. There are no "human readers" in the context of image interpretation improving with AI assistance. The device directly measures antibody levels.

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

• Partially applicable, but in a different sense. The device itself performs the measurement and provides a quantitative result. The "performance" demonstrated (precision, linearity, LoD, method comparison) is inherently "standalone" in that it's the instrument and reagents producing the values. The "human-in-the-loop" would be the clinician interpreting the numerical result in conjunction with other clinical findings for diagnosis. There isn't an "algorithm" making a diagnostic call independently like in an imaging AI; rather, it's a measurement system.

7. The Type of Ground Truth Used

• Analytical Performance:
  • Reference materials and statistical methods: For precision, linearity, and detection limit, the ground truth is based on highly characterized control materials, serial dilutions, and established statistical methodologies (e.g., CLSI guidelines EP05-A3, EP06-A, EP17-A).
• Clinical Performance (Cut-off determination and comparison):
  • Clinical diagnosis and predicate device comparison: The clinical cut-off values for "Negative," "Equivocal," and "Positive" were "derived from the clinical studies (s. K141375)." This implies that the cut-offs were established using patient samples with confirmed clinical diagnoses of primary biliary cirrhosis (or healthy controls) as the ground truth in the predicate device's original studies.
  • For the current submission, the "ground truth" for the method comparison study (comparing Phadia 2500/5000 to Phadia 250) is the result generated by the predicate device (EliA M2 on Phadia 250 instrument). The goal is to show the new instrument produces substantially equivalent results to the established predicate.

8. The Sample Size for the Training Set

• Not applicable in the AI/ML sense. This is an immunoassay system; it does not have a "training set" in the context of machine learning model development. The development process would involve formulation optimization, reagent stability testing, and calibration studies, but not "training data" for an algorithm that learns patterns. The calibration curve is established using calibrators provided with the kit.

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

• Not applicable. See point 8. For IVDs, the "ground truth" for establishing a calibration curve (which is somewhat analogous to "training" the system to interpret raw signals into quantitative units) is defined by the known concentrations of the calibrator materials provided in the kit. These calibrators are rigorously manufactured and assigned values traceable to reference standards.

Ask a specific question about this device