Search Results

ArteraAI Prostate is a software only device intended to analyze scanned histopathology whole slide images (WSIs) from treatment-naïve prostate core needle biopsies prepared from formalin fixed paraffin-embedded (FFPE) tissue and stained using Hematoxylin & Eosin (H&E) stains. ArteraAI Prostate provides 10-year risks of distant metastasis and prostate cancer specific mortality and is intended to assist physicians with prognostic risk-based decisions along with other clinicopathological factors in non-metastatic prostate cancer patients.

ArteraAI Prostate is intended for males 55 years of age or older without clinically or pathologically defined metastases, and who are candidates for curative intent management (surgery, radiation therapy with or without systemic therapy, or active surveillance).

ArteraAI Prostate is intended to utilize WSIs acquired from an FDA-cleared interoperable scanner with which ArteraAI Prostate has been authorized for use or a 510(k)-cleared scanner that has been assessed in accordance with the Predetermined Change Control Plan (PCCP) for qualifying additional interoperable scanners.

ArteraAI Prostate is a software only device that utilizes deep learning algorithms developed with WSI of H&E-stained prostate needle biopsies to assess risk of distant metastasis and prostate cancer specific mortality. The software performs an algorithmic assessment of features extracted from WSIs using self-supervised learning. ArteraAI Prostate consists of one algorithm, comprised of multiple building block models that intake image data to calculate ArteraAI raw scores. These raw scores are used to estimate risk categories (High, Intermediate, Low) for 10-year risk of DM and PCSM. The algorithm is locked; it is not a continuous learning (continual machine learning model) algorithm.

ArteraAI Prostate includes the ArteraAI Platform (which includes the sub-components ArteraAI Web Portal and ArteraAI Back-End) and ArteraAI Prostate (which includes the sub-components ArteraAI Image Converter and the ArteraAI AI Engine). The ArteraAI Prostate interacts with the ArteraAI Platform. The functions of the above components are provided in the Table 1 below.

Here's a breakdown of the acceptance criteria and the study that proves the ArteraAI Prostate device meets them, based on the provided document:

1. Acceptance Criteria and Reported Device Performance

The document doesn't explicitly state a table of pre-defined "acceptance criteria" with numerical targets for clinical performance. Instead, it describes the desired clinical significance of the risk categories produced by the device, and then presents the results of a retrospective clinical study to demonstrate that the device achieved these desired outcomes.

Therefore, I've constructed a table based on the desired clinical significance statements mentioned in the "Summary of Clinical Validation Results" section, and mapped them to the reported performance.

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

• Sample Size (Test Set/Clinical Validation Set): 886 patients.
• Data Provenance: Retrospective clinical study across three sites in the US. Patients were diagnosed since 2005.

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

The document indicates that pathologists were involved in the initial diagnosis and identification of the highest Gleason score, which is a key input to the device. However, it does not specify the number of experts explicitly used to establish the ground truth for the prognostic outcomes (distant metastasis and prostate cancer specific mortality) in the clinical validation test set. The ground truth for these outcomes would typically be derived from long-term follow-up clinical data, not directly from expert review of the images for the purpose of the study. Pathologists' initial diagnoses determined the patient cohort for the study.

4. Adjudication Method for the Test Set

The document does not describe an adjudication method for establishing the ground truth of the outcomes in the clinical validation study. The outcomes (distant metastasis and prostate cancer specific mortality) are factual events observed during patient follow-up, not subjective interpretations requiring adjudication.

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

No, a MRMC comparative effectiveness study was not done as described in this document. The study evaluated the standalone performance of the ArteraAI Prostate device in predicting patient outcomes, not its improvement of human reader performance with AI assistance.

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

Yes, a standalone study was done. The clinical performance evaluation focused on the ArteraAI Prostate device's ability to predict 10-year risks of DM and PCSM based on its algorithmic analysis of WSIs, without human readers "in-the-loop" during the prognosis prediction phase for the study purpose. The device output is a report for physician interpretation, but the core clinical performance evaluation was of the algorithm's prognostic capability itself.

7. The Type of Ground Truth Used

The ground truth used for the clinical validation study was outcomes data. Specifically:

• Actual distant metastasis (DM) events experienced by the patients.
• Actual prostate cancer specific mortality (PCSM) events experienced by the patients.
  These events were tracked over a 10-year period (and up to 18.7 years of follow-up for censored patients) using Kaplan-Meier survival analysis.

8. The Sample Size for the Training Set

The document states: "There were 1,133 distant metastasis events and 931 prostate cancer specific mortality events in the dataset used to train the model." While it provides the number of events, it does not explicitly state the total sample size (number of patients or WSIs) used for the training set. It refers to "multiple, multi-center, prospective randomized controlled clinical trials and clinical studies" for the training data (Table 2 describes the characteristics of this training data but does not provide a total N).

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

The ground truth for the training set was established using outcomes data. The model was trained on "the ground truth of actual metastasis or prostate cancer specific mortality events experienced by the patients in the trials from which the data is extracted." This implies that the historical clinical follow-up data from these trials served as the ground truth for training the AI model to predict these events.

Ask a specific question about this device