Search Results

AutoAS is a software application intended to assist medical professionals in the assessment of moderate/severe aortic stenosis (AS). The software uses an artificial intelligence (AI) algorithm to process previously acquired two-dimensional transthoracic echocardiography (2D-TTE) images to provide a suggestion of moderate/severe aortic stenosis along with an associated confidence metric that can be a diagnostic aid to a physician in a point of care or similar setting in determining if further evaluation is needed, including whether a full echocardiogram (2D, Doppler) needs to be performed.

The results of AutoAS are not intended to be used on a stand-alone basis for clinical decision making and are not intended to supplement or replace a full echocardiographic examination. AutoAS results, along with the obtained ultrasound images, must be reviewed by a qualified physician. The AutoAS product is not intended to be used on patients who have prosthetic valves and/or have had prior valve repair or replacement.

AutoAS software is indicated for use in adult patients and is intended to be an accessory to compatible ultrasound systems in environments where healthcare is provided.

Automated Aortic Stenosis Software (AutoAS) is a breakthrough software product that assesses the presence and severity of aortic stenosis (AS) in B-mode cardiac ultrasound scans. The software can be integrated with a compatible ultrasound device in a headless manner. The AutoAS software is intended to be an accessory to compatible ultrasound systems. The AutoAS software is intended for use in adult patients undergoing transthoracic cardiac ultrasound examinations in whom assessment for aortic stenosis (AS) is clinically relevant. The indicated population includes patients who are being evaluated for the presence or likelihood of moderate to severe aortic stenosis as part of a routine or targeted echocardiographic study.

AutoAS processes relevant ultrasound images acquired from a concurrent and/or previously acquired ultrasound exam, employing advanced algorithms to generate AS predictions and supporting outputs for the user. The AutoAS software operates on B‑mode transthoracic cardiac ultrasound images acquired during a standard ultrasound examination using a compatible GE HealthCare ultrasound system. The reading protocol is designed to ensure that AutoAS outputs are used as adjunctive information and are interpreted within the context of a comprehensive clinical and echocardiographic evaluation by a qualified physician. The AS prediction, severity, and supporting outputs are summarized as a report that is available after the exam for the user to review. The report can also be exported to an archive with the ultrasound images ensuring seamless integration with the patient's record and facilitating downstream clinical workflows.

The software's algorithms process specific views obtained during an ultrasound study. These views may include the parasternal long axis (PLAX), parasternal short axis at the aortic valve level (PSAX-AV), and apical five-chamber (AP5). The AS predictions come in the form of a severity prediction: 1) Suggestive of moderate to severe AS or 2) Not suggestive of moderate to severe AS with associated information on the confidence of the algorithm's prediction.

The AutoAS results along with ultrasound images must be reviewed by a qualified physician as the AutoAS software does not diagnose Aortic Stenosis (AS) but rather indicates the likelihood of AS. Interpretation of AutoAS results must be performed by a qualified physician with training and experience in cardiac ultrasound and echocardiographic interpretation. The results of AutoAS are not intended to be used on a stand-alone basis for clinical decision making and are not intended to supplement or replace a full echocardiographic examination. The physician must review the AutoAS outputs in conjunction with the underlying ultrasound images and relevant clinical information. The user is responsible for determining the clinical relevance of the AutoAS findings and for integrating the software outputs into the overall diagnostic impression.

1. Table of Acceptance Criteria and Reported Device Performance

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

• Standalone Performance Assessment:
  • Sample Size: 401 studies from 401 unique patients.
  • Data Provenance: Retrospectively obtained from four different U.S. institutions. The dataset included echocardiographic studies from multiple ultrasound models from two different manufacturers (GE Healthcare and Philips Healthcare).
• Clinical Performance Assessment (MRMC Study):
  • Sample Size: A subset of 220 unique studies across 220 unique patients from the validation data used in the standalone performance assessment.
  • Data Provenance: Retrospectively obtained from three different U.S. institutions (implies these are also U.S. institutions, similar to the larger validation dataset).

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

• Number of Experts: Three (3) level-III echocardiographers.
• Qualifications of Experts: Described as "level-III echocardiographers," indicating a high level of expertise and experience in echocardiography interpretation.

4. Adjudication Method for the Test Set

• Adjudication Method: Majority vote of the 3 echocardiographers (also known as the statistical mode). Each echocardiographer assessed studies independently, blinded to the interpretations of the other two and the original study's AS interpretation.

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

• Yes, an MRMC study was done.
• Effect Size of Improvement with AI Assistance:
  • Sensitivity: A statistically significant improvement of +5.5% [95% CI: (1.5%, 9.5%)] for "Aided" readers compared to "Unaided" readers.
  • Partial AUROC: An 8.9% [95% CI: 1.2%, 20.5%] difference in partial AUROC, indicating superiority for the "Aided" group in the critical region of the ROC curve.
  • Inter-rater agreement: Aided readers demonstrated higher inter-rater agreement (89.0%) than unaided readers (81.9%), reflecting improved reader consistency.

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

• Yes, a standalone performance assessment was done. The results are presented in the "Standalone Performance Assessment" section and include metrics like Area Under the ROC Curve, Specificity, and Sensitivity.

7. The Type of Ground Truth Used

• Expert Consensus: For the clinical performance assessment (both standalone and MRMC), the reference standard was established by the majority vote of three independent level-III echocardiographers based on a full read of echocardiography studies, adhering to Aortic Valve Area (AVA) per clinical guidelines from the American Society of Echocardiography (ASE).

8. The Sample Size for the Training Set

• The document does not explicitly state the sample size for the training set. It mentions a "validation dataset" of 401 studies for standalone performance and a subset of 220 studies for the MRMC study, but no information is provided regarding the data used to train the AI algorithm.

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

• The document does not explicitly state how the ground truth for the training set was established. Information regarding the training data, including its provenance and annotation methodology, is not provided in this excerpt.

Ask a specific question about this device