LogoFDA 510(k) Search
Search Filters

Search Results

Found 1 results

510(k) Data Aggregation

    K Number
    K181288
    Device Name
    Athelas One
    Manufacturer
    Athelas Inc.
    Date Cleared
    2018-11-05

    (173 days)

    Product Code
    GKZ
    Regulation Number
    864.5220
    Type
    Traditional
    Panel
    Hepatology / Hepatic (HE)
    Reference & Predicate Devices
    K112605,K071967
    Why did this record match?
    Device Name :

    Athelas One

    AI/MLSaMDIVD (In Vitro Diagnostic)TherapeuticDiagnosticis PCCP AuthorizedThirdpartyExpeditedreview
    Intended Use

    Athelas One is indicated for use for quantitative determination of white blood cells (WBC) and Neutrophil percentages (NEUT%) in capillary or K2EDTA venous whole blood. The Athelas One system is for In Vitro Diagnostic use only. The Athelas One is only to be used with Athelas One Test Strips. The Athelas One is indicated for use in clinical laboratories and for point of care settings. The Athelas One is only indicated for use in adult populations (aged 21 and older).

    Device Description

    The Athelas One is an automated diagnostic device intended to perform tests on whole blood samples collected in K2EDTA or capillary finger stick samples collected directly into the Athelas One test strip. The system is intended to be placed within the Point of Care, and Clinical Laboratory sites. Athelas One returns WBC and Neut% metrics from the blood sample. A clinician places a sample on the Athelas One test strip either directly from finger or via pipette from K2EDTA whole blood tube. The Athelas One test strip stains and creates a monolayer of the blood sample within the chamber. The strip is inserted into the test strip slot of the Athelas One device and the device returns results by conducting image analysis of cells present. Result-viewing and device control are conducted through a companion tablet/mobile application.

    AI/ML Overview

    Here's a breakdown of the acceptance criteria and study details for the Athelas One device based on the provided document:

    Acceptance Criteria and Reported Device Performance

    ParameterAcceptance Criteria (Target Evaluation)Reported Device PerformanceStudy Section
    WBC Precision7.5% CV above 2K/µL WBC. 0.25 K/µL SD below 2K/µL WBC.Within-run Precision: All samples met the 7.5% CV target. For example, a mean value of 2.20 K/µL had a total CV of 5.87%, and a mean value of 23.33 K/µL had a total CV of 5.81%.
    Between-run Precision (Reproducibility): Total CV for WBC Low (2.746 K/µL) was 5.583%. Total CV for WBC Medium (7.546 K/µL) was 5.726%. Total CV for WBC High (15.246 K/µL) was 6.305%. All met the acceptance criteria.Bench: Within-run Precision/Reproducibility, Precision Reproducibility (between run)
    WBC Bias/Error±7.5% error above 2K/µL WBC. ±0.25 K/µL error below 2K/µL WBC.Method Comparison: Mean Bias for WBC (1.1 - 23 K/µL) was -0.151 K/µL (-2.31%).
    Specimen Stability: Met ± 7.5% Bias for WBC for samples tested within 24 hours.Clinical: Method Comparison, Specimen Stability Studies
    Neutrophil % Precision5% SD OR 15% CV (updated based on Sysmex XW-100 Neut% CV criteria).Within-run Precision: Not explicitly stated in the WBC Summarized table for Neutrophils.
    Between-run Precision (Reproducibility): Total CV for NEUT % Low (50.781%) was 6.780%. Total CV for NEUT % Medium (49.990%) was 6.689%. Total CV for NEUT % High (50.823%) was 6.525%. All met the acceptance criteria (which allows for 15% CV if 5% SD is not met).Bench: Precision Reproducibility (between run)
    Neutrophil % Bias±10% bias or ±5% Neut% total error (whichever larger).Method Comparison: Mean Bias for Neutrophil % (8 - 92.89%) was 0.636% (1.18%).
    Specimen Stability: Met ± 10% Bias for Neutrophils for samples tested within 24 hours.Clinical: Method Comparison, Specimen Stability Studies
    Linearity (WBC)Not a single target, but demonstrated to be linear with acceptable max % diff for each interval.R² = 0.997, Slope = 1.013, Intercept = 0.0449, CVr = 5.08%. Demonstrated to be linear from lower limit to upper limit and within measured allowable max % diff.Bench: Linearity
    Flagging Accuracy (Morphological Flags)≥ 90% accuracyPositive Agreement (Sensitivity) = 90.91%. Negative Agreement (Specificity) = 96.71%. Overall Agreement = 94.87%. Met the specification.Clinical: Flagging Comparison
    Matrix Comparability (Venous vs. Capillary)No statistical difference and met evaluation criteria for regression parameters and 95% CI of bias.WBC: Slope 1.026 (CI: 1.000, 1.055), Intercept -0.145 (CI: -0.319, 0.018), Mean Bias 0.056 K/µL (-0.588%). R = 0.996.
    Neutrophil %: Slope 0.999 (CI: 0.938, 1.058), Intercept 0.457 (CI: -2.502, 3.614), Mean Bias 0.162 Percentage Points (-0.333%). R = 0.969. Concluded no statistical difference.Clinical: Matrix Comparison

    Study Details

    1. Sample size used for the test set and the data provenance:

      • Within-run Precision: 9 whole blood samples, 90 tests per sample (total 810 tests). Provenance not explicitly stated beyond "K2EDTA whole blood samples" and "normal and abnormal samples."
      • Precision/Reproducibility (between run): 3 levels of quality control material, 80 readings per level at each of 3 sites (total 720 readings for controls). Provenance not explicitly stated.
      • Linearity: 10 samples run in 4 replicates across 4 devices (total 160 measurements + varying replicates/devices for different points in the linearity study, but "10 samples" is the core). Provenance of samples not explicitly stated, but "pooling together one low WBC concentration fresh whole blood sample one high WBC concentration sample."
      • Interfering Substances: "Various substances... found naturally occurring in patient samples, or were spiked in whole blood." Specific sample size not provided.
      • Reference Intervals: 120 healthy donors. Provenance not explicitly stated.
      • Limit of Blank (LoB): 120 total repeated measurements of blank samples.
      • Limit of Detection (LoD): 5 low-level samples, 2 replicates per sample over 3 days (total ~30 measurements per lot for low-level samples, plus 60 blank measurements per lot).
      • Limit of Quantification (LoQ): 4 independent low-level whole blood samples, 3 replicates per sample.
      • Specimen Stability: 9 different venous blood samples (low, normal, high WBC levels).
      • Test Strip Stability: Not a patient sample study, but involved testing control fluid.
      • Method Comparison: 312 patient samples. Provenance: "taken at 3 point of care sites in the US." The document implies prospective collection as they were "run ... on the XE-5000" and then analyzed by Athelas One.
      • Flagging Comparison: The same 312 patient samples used in Method Comparison.
      • Matrix Comparison: 59 patients who provided both capillary finger-prick and venous whole blood samples.

    2. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:

      • The document does not explicitly state the number of experts or their qualifications used to establish ground truth.
      • For the Method Comparison and Flagging Comparison, the "ground truth" was the Sysmex XE-5000 predicate analyzer. This is a legally marketed device and generally considered a gold standard in this context.
      • For the Linearity study, "original concentrations were obtained from the predicate analyzer (Sysmex XE-5000)."

    3. Adjudication method for the test set:

      • Not applicable as the ground truth was primarily established by a predicate device (Sysmex XE-5000) or by CLSI-recommended methods for bench studies. There was no mention of human expert adjudication for discrepancies.

    4. 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 MRMC comparative effectiveness study was done. This document describes a standalone performance study of the Athelas One device against a predicate device, not a human-in-the-loop study comparing human performance with and without AI assistance.

    5. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:

      • Yes, a standalone performance study was done. The Athelas One device, which uses "computer vision based image analysis" (i.e., its algorithm), was compared directly against a predicate automated hematology analyzer (Sysmex XE-5000) for WBC and Neutrophil % measurement.

    6. The type of ground truth used (expert consensus, pathology, outcomes data, etc.):

      • The primary ground truth for clinical comparisons (Method Comparison, Flagging Comparison) was the Sysmex XE-5000 automated hematology analyzer, a legally marketed predicate device.
      • For bench studies like Linearity and LoQ, concentrations were often referenced against the Sysmex XE-5000 as well.
      • For precision studies, the device's own measurements were evaluated for consistency against predefined statistical targets (SD, CV).
      • For interfering substances, the impact on bias and precision performance was evaluated, likely against expectations from predicate devices or known clinical impact.

    7. The sample size for the training set:

      • The document does not provide information on the training set size or how the algorithm was developed. This submission focuses on the validation of the finished device for regulatory approval, not its development process.

    8. How the ground truth for the training set was established:

      • Since information on the training set itself is not provided, how its ground truth was established is also not described in this document.
    Ask a Question

    Ask a specific question about this device

    Page 1 of 1