(180 days)
Tempus ECG-Low EF is software intended to analyze resting, non-ambulatory 12-lead ECG recordings and detect signs associated with having a low left ventricular ejection fraction (LVEF less than or equal to 40%). It is for use on clinical diagnostic ECG recordings collected at a healthcare facility from patients 40 years of age or older at risk of heart failure. This population includes but is not limited to patients with atrial fibrillation, aortic stenosis, cardiomyopathy, myocardial infarction, diabetes, hypertension, mitral regurgitation, and ischemic heart disease.
Tempus ECG-Low EF only analyzes ECG data and provides a binary output for interpretation. Tempus ECG-Low EF is not intended to be a stand-alone diagnostic tool for cardiac conditions, should not be used for patient monitoring, and should not be used on ECGs with paced rhythms. Results should be interpreted in conjunction with other diagnostic information, including the patient's original ECG recordings and other tests, as well as the patient's symptoms and clinical history.
A positive result may suggest the need for further clinical evaluation in order to establish a diagnosis of low LVEF. Patients receiving a negative result should continue to be evaluated in accordance with current medical practice standards using all available clinical information.
Tempus ECG-Low EF is a cardiovascular machine learning software intended for analysis of 12-lead resting ECG recordings using machine-learning techniques to detect signs of cardiovascular conditions for further referral or diagnostic follow-up. The software employs machine learning techniques to analyze ECG recordings and detect signs associated with a patient experiencing low left ventricular ejection fraction (LVEF), less than or equal to 40%. The device is designed to extract otherwise unavailable information from ECGs conducted under the standard of care, to help health care providers better identify patients who may be at risk for undiagnosed LVEF in order to evaluate them for further referral or diagnostic follow up.
As input, the software takes data from a patient's 12-lead resting ECG (including age and sex). It is only compatible with ECG recordings collected using 'wet' Ag/AgCl electrodes with conductive gel/paste, and using FDA authorized 12-lead resting ECG machines manufactured by GE Medical Systems or Philips Medical Systems with a 500 Hz sampling rate. It checks the format and quality of the input data, analyzes the data via a trained and 'locked' machine-learning model to generate an uncalibrated risk score, converts the model results to a binary output (or reports that the input data are unclassifiable), and evaluates the uncalibrated risk score against pre-set operating points (thresholds) to produce a final result. Uncalibrated risk scores at or above the threshold are returned as 'Low LVEF Detected,' and uncalibrated risk scores below the threshold are returned as 'Low LVEF Not Detected.' This information is used to support clinical decision making regarding the need for further referral or diagnostic follow-up. Typical diagnostic follow-up could include transthoracic echocardiogram (TTE) to detect previously undiagnosed LVEF, as described in device labeling. Results should not be used to direct any therapy against LVEF itself. Tempus ECG-Low EF is not intended to replace other diagnostic tests.
Tempus ECG-Low EF does not have a dedicated user interface (UI). Input data comprising ECG tracings, tracing metadata (e.g., sample count, sample rate, patient age/sex), is provided to Tempus ECG-Low EF through standard communication protocols (e.g., file exchange) with other medical systems (e.g., electronic health record systems, hospital information systems, or other data display, transfer, storage, or format-conversion software). Results from Tempus ECG-Low EF are returned to users in an equivalent manner.
Here's a detailed breakdown of the acceptance criteria and the study that proves the Tempus ECG-Low EF device meets them, based on the provided FDA 510(k) clearance letter:
Acceptance Criteria and Reported Device Performance
| Criteria | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| Sensitivity (for LVEF ≤ 40%) | ≥ 80% (lower bound of 95% CI) | 86% (point estimate); 84% (lower bound of 95% CI) |
| Specificity (for LVEF > 40%) | ≥ 80% (lower bound of 95% CI) | 83% (point estimate); 82% (lower bound of 95% CI) |
Study Details
2. Sample Size Used for the Test Set and Data Provenance
- Test Set Sample Size: Greater than 15,000 ECGs (specifically, 14,924 patient records are detailed in Table 1, with each patient having one ECG).
- Data Provenance: Retrospective observational cohort study. The data was derived from 4 geographically distinct US clinical sites.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
The document does not explicitly state the number of experts used or their qualifications for establishing the ground truth. It mentions that a clinical diagnosis of Low EF (LVEF ≤ 40%) was determined by a Transthoracic Echocardiogram (TTE), which is considered the gold standard for LVEF measurement. The interpretation of these TTE results to establish the ground truth would typically be done by cardiologists or trained echocardiography specialists, but the specific number and qualifications are not provided in this document.
4. Adjudication Method for the Test Set
The document does not explicitly state an adjudication method (such as 2+1 or 3+1) for the ground truth of the test set. The ground truth was established by correlating ECGs with TTEs to determine the presence or absence of a clinical diagnosis of low EF. It is implied that the TTE results themselves, as the gold standard, served as the definitive ground truth without a further adjudication process by multiple human readers for the TTE results in the context of this AI device validation.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, and the Effect Size of How Much Human Readers Improve with AI vs. Without AI Assistance
The document does not indicate that an MRMC comparative effectiveness study was performed, nor does it provide an effect size for human reader improvement with AI assistance. The study focuses on the standalone performance of the AI device.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done
Yes, a standalone study was done. The described clinical performance validation evaluated the device's ability to "detect signs associated with a clinical diagnosis of low LVEF" and provided sensitivity and specificity metrics for the algorithm's output. The device "only analyzes ECG data and provides a binary output for interpretation," indicating a standalone performance assessment.
7. The Type of Ground Truth Used
The ground truth used was established by Transthoracic Echocardiogram (TTE), specifically used to determine the presence or absence of a clinical diagnosis of Low EF (LVEF ≤ 40%). This is a form of outcomes data / reference standard as TTE is the established clinical diagnostic method for LVEF.
8. The Sample Size for the Training Set
- Training Set Sample Size: More than 930,000 ECGs (specifically, 930,689 ECGs are detailed in Table 1).
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. However, given that the model was trained to "detect signs associated with having a low left ventricular ejection fraction (LVEF less than or equal to 40%)" and the validation set used TTE for ground truth, it is highly probable that the training set also used LVEF measurements (likely from echocardiograms) as the ground truth. The description states the model was trained "on data from more than 930,000 ECGs," but does not detail the specific methodology for establishing the LVEF ground truth for each of these training examples.
N/A