(72 days)
The AI-ECG Tracker is intended to be used by qualified healthcare professionals in hospitals and healthcare facilities for the assessment of arrhythmias using ECG data acquired from adults (age 22 and older) without pacemakers. The product supports downloading and analyzing data recorded from electrodes with conductive paste/gel placed on standard location in compatible formats from any device used for the arrhythmia diagnostics such as Holter, event recorder, 12-lead ambulatory ECG devices when assessment of the rhythm is necessary. The AI-ECG Tracker provides ECG signal processing and analysis on a beat by beat basis, QRS detection, Supraventricular Ectopic Beat detection, QRS feature extraction, interval measurement, and rhythm analysis. The AI-ECG Tracker is not for use in life supporting or sustaining systems or ECG monitoring and Alarm devices.
The AI-ECG Tracker interpretation results are not intended to be the sole means of diagnosis for any abnormal ECG. It is offered to physicians and clinicians on an advisory basis only in conjunction with the physician's knowledge of ECG patterns, patient background, clinical history, symptoms, and other diagnostic information.
The AI-ECG Tracker is a distributed ECG auto analysis system designed to assist physicians and qualified healthcare professionals in measuring and interpreting ambulatory ECG data. The interpretations by the analysis program can be confirmed, edited, modified, or deleted by the physician and qualified healthcare professionals. The program is intended for use by qualified healthcare professionals in hospitals and other healthcare facilities for the assessments of common cardiac arrhythmias using ECG data acquired from adults (age 22 and older) without pacemakers.
The AI-ECG Tracker receives ECG waveform data uploaded by a user, analyzes ECG data and automatically interprets on a computer server. The ECG measurement, interpretation and waveform data are then downloaded to the Physician Diagnostic Client for a physician to review, modify, confirm the analysis statements, and print the report. The original ECG waveform data is stored permanently in the user's server computer securely.
The system uses a machine learning based process (convolutional neural network or CNN) only for development of the AI ECG algorithm. The AI ECG algorithm is only used for beat classification. After the AI ECG algorithm is developed, the AI-based beat classification model is locked in the released product which means the marketed device doesn't have active machine learning or selflearning features.
Here's a breakdown of the acceptance criteria and study information based on the provided document:
1. Acceptance Criteria and Reported Device Performance
The document primarily focuses on demonstrating substantial equivalence to a predicate device rather than explicitly stating acceptance criteria with specific performance metrics for the AI-ECG Tracker. However, it does reference compliance with performance standards.
| Acceptance Criteria (Implied / Referenced Standard) | Reported Device Performance (Compliance) |
|---|---|
| Basic safety and essential performance | Comply with IEC 60601-2-25 |
| Measurement performance (Cardiac Rhythm and ST-Segment algorithms) | Comply with AAMI/ANSI EC57 and IEC 60601-2-47 |
Note: The document states that "Bench tests were conducted to verify that the subject device met all design specifications, as was Substantially Equivalent (SE) to the predicate device." This suggests that the device's performance was evaluated against the predicate device's capabilities and relevant standards to establish substantial equivalence. Specific quantitative performance values (e.g., sensitivity, specificity for arrhythmia detection) are not provided in this document.
2. Sample Size for Test Set and Data Provenance
The document does not explicitly state the sample size used for the test set or the data provenance (e.g., country of origin, retrospective/prospective nature of the data). It mentions "bench tests," which typically cover technical performance and compliance with standards rather than clinical validation on a specific dataset.
3. Number of Experts and Qualifications for Ground Truth
The document does not provide details on the number of experts used or their qualifications for establishing ground truth for the test set.
4. Adjudication Method for the Test Set
The document does not specify any adjudication method (e.g., 2+1, 3+1, none) used for the test set.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
There is no indication that a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was conducted or reported in this document. The focus is on the device's standalone performance and its equivalence to a predicate device.
6. Standalone (Algorithm Only) Performance Study
Yes, a standalone (algorithm only without human-in-the-loop performance) study was performed. The document states:
- "The AI-ECG Tracker provides ECG signal processing and analysis on a beat by beat basis, QRS detection, Supraventricular and Ventricular Ectopic Beat detection, QRS feature extraction, interval measurement, heart rate measurement, and rhythm analysis."
- "Bench tests were conducted to verify that the subject device met all design specifications..."
- The "AI-ECG Tracker" is presented as an automated analysis system that "automatically interprets on a computer server."
This indicates that the algorithm's performance in analyzing ECG data was evaluated independently.
7. Type of Ground Truth Used
The specific type of ground truth used for performance evaluation is not explicitly stated. However, given the context of "bench tests" and compliance with standards like AAMI/ANSI EC57, it is highly likely that the ground truth for algorithm performance was established through:
- Reference annotated ECG databases: Standardized ECG databases with expert-adjudicated annotations are commonly used for validating ECG analysis algorithms.
- Expert consensus: For specific cases or discrepancies, expert cardiologists would establish the true rhythm or beat classification.
8. Sample Size for the Training Set
The document does not provide the sample size used for the training set. It mentions that the "system uses a machine learning based process (convolutional neural network or CNN) only for development of the AI ECG algorithm."
9. How Ground Truth for the Training Set Was Established
While the document states that a CNN was used for "development of the AI ECG algorithm" (for beat classification), it does not explicitly detail how the ground truth for this training set was established. In typical machine learning development for medical devices, this would involve:
- Expert annotation: Cardiologists or trained technicians meticulously annotating ECG recordings to label various beat types and rhythms.
- Validated databases: Leveraging existing, publicly available, or proprietary ECG databases that come with expert-verified annotations.
§ 870.2340 Electrocardiograph.
(a)
Identification. An electrocardiograph is a device used to process the electrical signal transmitted through two or more electrocardiograph electrodes and to produce a visual display of the electrical signal produced by the heart.(b)
Classification. Class II (performance standards).