Search Results

Bunkerhill ECG-EF is software intended to aid in screening for Left Ventricular Ejection Fraction (LVEF) less than or equal to 40% in adults at risk for, but not already diagnosed with low LVEF.

Bunkerhill ECG-EF is not intended to be a stand-alone diagnostic device for cardiac conditions, should not be used for monitoring of patients, and should not be used on ECGs with a paced rhythm. A positive result may suggest the need for further clinical evaluation in order to establish a diagnosis of Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%.

Additionally, if the patient is at high risk for the cardiac condition, a negative result should not rule out further non-invasive evaluation.

Bunkerhill ECG-EF is adjunctive and must be interpreted in conjunction with the clinician's judgment, the patient's medical history, symptoms, and additional diagnostic tests. For a final clinical diagnosis, further confirmatory testing, such as echocardiography, is required.

ECG-EF is a software-only medical device that employs deep learning algorithms to analyze 12-lead ECG data for the detection of low left ventricular ejection fraction (LVEF < 40%). The algorithm processes 10-second ECG waveform snippets, providing predictions to assist healthcare professionals in the early identification of patients at risk for heart failure.

ECG-EF algorithm receives digital 12-lead ECG data and processes it through its machine learning model. The output of the analysis is transmitted to integrated third-party software systems, such as Electronic Medical Records (EMR) or ECG Management Systems (EMS). The results are displayed by the third-party software on a device such as a smartphone, tablet, or PC.

ECG-EF algorithm produces a result indicating "Low EF Screen Positive - High probability of low ejection fraction based on the ECG", " Low EF Screen Negative - Low probability of low ejection fraction based on the ECG" or "Error – device input criteria not met " for cases that do not meet data input requirements. These results are not intended to be definitive diagnostic outputs but rather serve as adjunctive information to support clinical decision-making. A disclaimer accompanies the output, stating: "Not for diagnostic use. The results are not final and must be reviewed alongside clinical judgment and other diagnostic methods."

The Low Ejection Fraction AI-ECG Algorithm device is intended to address the unmet need for a point-of-care screen for LVEF less than or equal to 40% and is expected to be used by cardiologists, front-line clinicians at primary care, urgent care, and emergency care settings, where cardiac imaging may not be available or may be difficult or unreliable for clinicians to operate. Clinicians will use the Low Ejection Fraction AI-ECG Algorithm to aid in screening for LVEF less than or equal to 40% and making a decision for further cardiac evaluation.

Here's a breakdown of the acceptance criteria and the study details for the Bunkerhill ECG-EF device, based on the provided FDA 510(k) Clearance Letter:

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size for the Test Set and Data Provenance

• Sample Size for Test Set: 15,994 patient records.
• Data Provenance:
  • Country of Origin: United States.
  • Source: Two health systems.
  • Type: Retrospective study.
  • Diversity: Representative of the U.S. population (65.5% White, 18.8% Hispanic, 5.7% American Indian or Alaska Native, 3.9% Asian, 3.0% Black/African American, 2.8% Other; 53% Male, 47% Female).
  • Geographical Distribution: Curated from 5 geographically distributed sites throughout the United States.

3. Number of Experts and Qualifications for Ground Truth

The document does not explicitly state the number of experts used or their specific qualifications for establishing the ground truth. It only mentions that the ground truth was established from echocardiograms.

4. Adjudication Method for the Test Set

The document does not specify an adjudication method (e.g., 2+1, 3+1, none) for the test set. The ground truth was derived directly from echocardiogram measurements.

5. MRMC Comparative Effectiveness Study

The document does not mention a Multi-Reader Multi-Case (MRMC) comparative effectiveness study or any effect size of human readers improving with AI vs. without AI assistance. The study focuses solely on the standalone performance of the AI algorithm.

6. Standalone Performance Study (Algorithm Only)

Yes, a standalone study evaluating the algorithm's performance without human-in-the-loop was conducted. The performance metrics (Sensitivity, Specificity, PPV, NPV) and the confusion matrix presented are for the algorithm's direct output.

7. Type of Ground Truth Used

The ground truth used was Transthoracic Echocardiogram (TTE) with disease, specifically using the Simpson's Biplane measurement method to determine Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. The echocardiogram was taken less than 15 days apart from the ECG scan.

8. Sample Size for the Training Set

The document does not explicitly state the sample size used for the training set. It only mentions the retrospective study for validation involved 15,994 patient records.

9. How Ground Truth for the Training Set Was Established

The document states that the "Ground Truth for Model Training" was Transthoracic echocardiogram (TTE) with disease. It can be inferred that this same method (TTE, likely Simpson's Biplane) was used to establish ground truth for the training data, similar to the test set, but specific details on the process for the training set are not provided.

Ask a specific question about this device

The Hypertension Notification Feature (HTNF) is a software-only mobile medical application that analyzes photoplethysmography (PPG) data opportunistically collected by Apple Watch to identify patterns that are suggestive of hypertension and provides a notification to the user.

The feature is intended for over-the-counter (OTC) use by adults age 22 and over who have not been previously diagnosed with hypertension. It is not intended to replace traditional methods of diagnosis, to monitor hypertension treatment effect, or to be used as a method of blood pressure surveillance. It is not intended for use during pregnancy. The absence of a notification does not indicate the absence of hypertension.

The Hypertension Notification Feature (HTNF) is an over-the-counter mobile medical application that is intended to analyze data collected from the PPG sensor of the Apple Watch (a general purpose computing platform), over multiple days to surface a notification to users who may have hypertension. The feature is intended for adults who have not been previously diagnosed with hypertension. The feature is not intended for use during pregnancy. The feature is not intended to replace traditional methods of diagnosis, to monitor hypertension treatment effect, or to be used as a method of blood pressure surveillance.

Absence of a notification does not indicate the absence of hypertension. HTNF cannot identify every instance of hypertension. In addition, HTNF will not surface a notification if insufficient data is collected.

HTNF comprises the following features:
• A software feature on the Apple Watch ("Software Feature on Watch"), and
• A pair of software features on the iOS device ("Software Feature on iPhone" and "Software Feature on iPad")

On the Apple Watch, HTNF uses PPG data and qualification information from the watch platform. The Software Feature on watch incorporates a machine-learning model that gives each qualified PPG signal a score associated with risk of hypertension.

On the iPhone, HTNF incorporates an algorithm that aggregates qualified hypertension risk scores and identifies patterns suggestive of hypertension. If hypertension patterns are identified, the feature surfaces a notification to users that they may have hypertension. The feature includes a user interface (UI) framework to enable user on-boarding and display educational materials and hypertension notification history in the Hypertension Notification room in the Health app.

On the iPad, HTNF provides a data viewing framework to display hypertension notification history in the Hypertension Notification room in Health app.

Here's a summary of the acceptance criteria and the study that proves the Apple Hypertension Notification Feature (HTNF) meets them, based on the provided FDA 510(k) clearance letter:

Apple Hypertension Notification Feature (HTNF) - Acceptance Criteria and Study Summary

1. Table of Acceptance Criteria and Reported Device Performance

Note: The document states that the feature "met all pre-determined primary endpoints" for overall sensitivity and specificity, but the specific numerical targets for these endpoints are not directly listed as "acceptance criteria" in the provided text. The reported performance values are the results from the clinical study that met these implicit criteria.

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

• Test Set Sample Size:

  • Clinical Validation Study: 2,229 enrolled subjects, with 1,863 subjects providing at least 15 days of usable data for the primary endpoint analysis.
  • Longitudinal Performance Evaluation: 187 non-hypertensive subjects.

• Data Provenance: The document does not explicitly state the country of origin for the data. However, it indicates subjects were "enrolled from diverse demographic groups" and "representative of the intended use population." The study described is a prospective clinical validation study where subjects wore an Apple Watch and measured blood pressure.

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

The document does not specify the use of "experts" to establish the ground truth for the test set.

• Ground Truth Method: Hypertension status was defined based on objective measurements from an FDA-cleared home blood pressure monitor. Specifically, "Hypertension is established as average systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 80 mmHg by America Heart Association." Therefore, expert consensus was not the primary method for ground truth determination in the principal clinical study.

4. Adjudication Method for the Test Set

Not applicable, as the ground truth was based on objective blood pressure monitor readings against established guidelines, not expert review requiring adjudication.

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

No, an MRMC comparative effectiveness study was not conducted. The HTNF is an "algorithm only" device designed to provide notifications to lay users, not an assistive tool for human readers in a diagnostic setting.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done

Yes, the primary clinical validation study assessed the standalone performance of the HTNF algorithm. The device "analyzes photoplethysmography (PPG) data... to identify patterns that are suggestive of hypertension and provides a notification to the user," without human intervention in the interpretation of the PPG data for notification generation.

7. The Type of Ground Truth Used

The ground truth used for the clinical validation study was objective outcome data (blood pressure measurements). Specifically, "Hypertension is established as average systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 80 mmHg by America Heart Association" using an FDA-cleared home blood pressure monitor as the reference.

8. The Sample Size for the Training Set

The document describes the algorithm development dataset as follows:

• Self-supervised learning for deep-learning (DL) model: "large-scale unlabeled data... included Apple Watch sensor data collected over 86,000 participants."
• Linear model training for classification: "included Apple Watch sensor data and home blood pressure reference measurements collected over 9,800 participants."

These datasets were pooled and split into Training, Train Dev, Test Dev, and Test sets for model development.

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

For the linear model that provides specific hypertension classifications (hypertensive vs. non-hypertensive), the ground truth for the training set was established using home blood pressure reference measurements. For the self-supervised deep learning model, it used "large-scale unlabeled data" where ground truth for hypertension status wasn't required for pre-training.

Ask a specific question about this device

The ECG-AI LEF 12-Lead algorithm is software intended to aid in earlier detection of Left Ventricular Ejection Fraction (LVEF) less than or equal to 40% in adults at risk for heart failure. This population includes, but is not limited to:

• patients with cardiomyopathies
• patients who are post-myocardial infarction
• patients with aortic stenosis
• patients with chronic atrial fibrillation
• patients receiving pharmaceutical therapies that are cardiotoxic, and
• postpartum women.

The ECG-AI LEF 12-Lead algorithm is not intended to be a stand-alone diagnostic device for cardiac conditions, should not be used for monitoring of patients, and should not be used on ECGs with a paced rhythm.

A positive result may suggest the need for further clinical evaluation in order to establish a diagnosis of Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. Additionally, if the patient is at high risk for the cardiac condition, a negative result should not rule out further non-invasive evaluation.

The ECG-AI LEF 12-Lead Algorithm should be applied jointly with clinician judgment.

The ECG-AI LEF 12-Lead algorithm interprets 12-lead ECG voltage times series data using an artificial intelligence-based algorithm. The device analyzes 10 seconds of a single 12-lead ECG acquisition, and within seconds provides likelihood of LVEF (ejection fraction less than or equal to 40%) to third party software. The results are displayed by the third party software on a device such as a smartphone, tablet, or PC. The ECG-AI LEF 12-Lead algorithm was trained to detect Low LVEF using positive and control cohorts, and the detection of Low LVEF in patients is generated using defined conditions and covariates.

The ECG-AI LEF 12-Lead algorithm device is intended to address the unmet need for a point-of-care screen for LVEF less than or equal to 40% and is expected to be used by cardiologists, frontline clinicians at primary care, urgent care, and emergency care settings, where cardiac imaging may not be available or may be difficult or unreliable for clinicians to operate. Clinicians will use the ECG-AI LEF 12-Lead algorithm to aid in earlier detection of LVEF less than or equal to 40% and making a decision for further cardiac evaluation.

The software module can be integrated into a client application to be accessed by clinicians and results viewed through an Electronic Medical Record (EMR) system or an ECG Management System (EMS) accessed via a PC, mobile device, or another medical device. In each case, the physician imports 12-lead ECG data in digital format. The tool analyzes the 10 seconds or longer duration of voltage data collected during a standard 12-lead ECG and outputs a binary result of the likelihood of low ejection fraction as an API result.

The provided text is a 510(k) clearance letter and summary for the Anumana, Inc. ECG-AI Low Ejection Fraction (LEF) 12-Lead Algorithm ([K250652](https://510k.innolitics.com/search/K250652)). While it describes the device, its intended use, and substantial equivalence to a predicate device, it does not contain the detailed performance study results, acceptance criteria tables, sample sizes, or ground truth establishment methods that would typically be found in the clinical study section of a full 510(k) submission.

The document discusses a "Predetermined Change Control Plan (PCCP)" which mentions future performance enhancement validation studies, but it doesn't present the specific results of the validation study that led to this clearance ([K250652](https://510k.innolitics.com/search/K250652)). It only states that "The performance characteristics for the ECG-AI LEF 12-Lead algorithm were evaluated through software verification and labeling verification," which refers to non-clinical data.

Therefore, many of the requested details cannot be extracted from the provided text. I will populate the table and answer the questions based only on the information available in the given document.

Acceptance Criteria and Device Performance Study (Extracted from provided 510(k) Summary)

The provided 510(k) summary (K250652) serves as an update to a previously cleared device (K232699). It focuses on expanding compatibility and minor changes, asserting substantial equivalence based on the predicate's performance rather than detailing a new, comprehensive clinical study for this specific submission. The document emphasizes "software verification and labeling verification" as the evaluation methods for performance characteristics for this particular submission, rather than a clinical performance study with specific metrics for acceptance criteria.

The Predetermined Change Control Plan (PCCP) section alludes to future performance enhancements and their validation, stating: "To be implemented, a modified version must demonstrate improved performance by meeting pre-specified acceptance criteria. These criteria require the new version's sensitivity and specificity point estimates to be greater than or equal to the previous version, with an improvement shown by either an increased point estimate or a tighter confidence interval lower bound for at least one of these metrics." However, these are future criteria for updates, not the current acceptance criteria for the clearance of K250652 based on a new clinical study.

Therefore, the specific quantitative acceptance criteria and reported device performance for the clinical study supporting the K250652 clearance are not explicitly stated in the provided text. The clearance is largely based on demonstrating substantial equivalence to the predicate (K232699) and software/labeling verification.

Based on the provided text, the specific details regarding the clinical performance study (including acceptance criteria, reported performance values, sample sizes, expert details, adjudication methods, MRMC studies, standalone performance, and ground truth establishment for the test set) are NOT available.

1. A table of acceptance criteria and the reported device performance

As noted above, the provided text does not contain a table of explicit quantitative acceptance criteria or reported device performance metrics (e.g., sensitivity, specificity, AUC) from a clinical study for K250652. The document claims "The performance characteristics for the ECG-AI LEF 12-Lead algorithm were evaluated through software verification and labeling verification" for this submission, indicating that a new, detailed clinical performance study with such metrics was not the basis for this specific clearance. The PCCP section specifies criteria for future updates, but not for this clearance.

2. Sample size used for the test set and the data provenance (e.g., country of origin of the data, retrospective or prospective)

• Sample Size for Test Set: Not specified in the provided document.
• Data Provenance: Not specified in the provided document. The PCCP mentions "multi-center retrospective clinical study" for future validations, but this isn't linked to the original clearance's test set.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g., radiologist with 10 years of experience)

• Not specified in the provided document.

4. Adjudication method (e.g., 2+1, 3+1, none) for the test set

• Not specified in the provided document.

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

• Not specified in the provided document. The current indication is "to aid in earlier detection" and "applied jointly with clinician judgment," which implies human-in-the-loop, but an MRMC study comparing performance with and without AI assistance is not detailed.

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

• The document states: "The ECG-AI LEF 12-Lead algorithm is not intended to be a stand-alone diagnostic device for cardiac conditions," and "should be applied jointly with clinician judgment." This implies the device is not intended for standalone use in practice. However, whether a standalone performance study was conducted to assess its raw diagnostic capability (e.g., area under the curve) is not explicitly stated. The statement "outputs a binary result of the likelihood of low ejection fraction as an API result" suggests a standalone algorithm output, but the FDA's clearance is for an "aid," not a primary diagnostic tool.

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

• The document mentions the device "was trained to detect Low LVEF using positive and control cohorts." For LVEF, the common ground truth is often echocardiography (measuring ejection fraction), but the specific method used for ground truth (e.g., echocardiography, MRI, or a combination/adjudication) is not specified.

8. The sample size for the training set

• Not specified in the provided document.

9. How the ground truth for the training set was established

• The document states the device "was trained to detect Low LVEF using positive and control cohorts," but it does not describe how the ground truth was established for these training cohorts (e.g., type of diagnostic test, clinical adjudication process).

Ask a specific question about this device

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

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.

Ask a specific question about this device

Tempus ECG-AF is intended for use to analyze recordings of 12-lead ECG devices and detect signs associated with a patient experiencing atrial fibrillation and/or atrial flutter (collectively referred to as AF) within the next 12 months. It is for use on resting 12-lead ECG recordings collected at a healthcare facility from patients:

• · 65 years of age or older,
• · without pre-existing or concurrent documentation of atrial fibrillation and/or atrial flutter,
• · who do not have a pacemaker or implantable cardioverter defibrillator, and
• · who did not have cardiac surgery within the preceding 30 days.

Performance of repeated testing of the same patient over time has not been evaluated and results SHOULD NOT be used for patient monitoring.

Tempus ECG-AF only analyzes ECG data. 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. Tempus ECG-AF is not for use in patients with a history of AF, unless the AF occurred after a cardiac surgery procedure and resolved within 30 days of the procedure. It is not for use to assess risk of occurrence of AF related to cardiac surgery.

Results do not describe a person's overall risk of experiencing AF or serve as the sole basis for diagnosis of AF, and should not be used as the basis for treatment of AF.

Results are not intended to rule out AF follow-up.

Tempus ECG-AF is a cardiovascular machine learning-based notification software intended to analyze recordings of 12-lead ECG devices from patients 65 years of age and older. The software employs machine learning techniques to analyze ECG recordings and detect signs associated with a patient experiencing atrial flutter (collectively referred to as AF) within the next 12 months. 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 AF in order to evaluate them for referral of further diagnostic follow up and address the unmet need of reducing the number of undiagnosed AF patients.

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 and 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 'increased risk' information; uncalibrated risk scores below the threshold are returned as 'no increased risk' information is used to support clinical decision making regarding the need for further referral or diagnostic follow-up. Typical diagnostic follow-up could include ambulatory ECG monitoring to detect previously undiagnosed AF, as described in device labeling. Results should not be used to direct any therapy aqainst AF itself, including anticoagulation therapy.

Tempus ECG-AF does not have a dedicated user interface (UI). Input data comprising ECG tracing metadata (sample count, sample rate, etc.), patient age and patient sex, will be provided to Tempus ECG-AF through standard communication protocols (e.g. AP), file exchange) with other medical systems (e.g., electronic health record systems, hospital information systems, or other medical device data display, transfer, storage, or format-conversion software). Results from Tempus ECG-AF will be returned to users in an equivalent manner.

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

1. Table of Acceptance Criteria and Reported Device Performance

The FDA clearance document does not explicitly state "acceptance criteria" in a table format with specific thresholds that the device had to meet. However, it presents the results of the clinical performance validation study, which implicitly represent the device's performance against expected clinical utility. The study endpoints of sensitivity and specificity were "met," indicating they were within an acceptable range for the intended use.

Here's a table based on the provided performance metrics:

Note: While the document states "Study endpoints of sensitivity and specificity were met," it does not explicitly define what specific numerical thresholds were considered "met" for acceptance.

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

• Test Set Sample Size: 4017 patients, with one ECG analyzed per patient. (Page 6)
• Data Provenance: Retrospective observational cohort study. Data was collected from 3 geographically distinct clinical sites (real-world data). (Page 6)

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

The document states that the AF status of each patient in the test set was determined based on "duplicate manual chart review" (Page 6). It does not specify the number of experts or their specific qualifications (e.g., "radiologist with 10 years of experience"). This suggests the ground truth was derived from existing medical records interpreted by qualified personnel, but the specific details of those adjudicators are not provided.

4. Adjudication Method for the Test Set

The document mentions "duplicate manual chart review" (Page 6) for establishing the ground truth. This implies at least two reviewers independently reviewed charts to determine AF status. It does not explicitly state a 2+1, 3+1, or other specific adjudication method if there were discrepancies between the duplicate reviews.

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

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. The study described is a standalone performance validation of the AI algorithm. The document focuses on the algorithm's performance in detecting signs of AF risk rather than how human readers' performance might improve with AI assistance.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done

Yes, a standalone performance study was done. The "Summary of Clinical Studies" section describes the evaluation of "the Tempus ECG-AF model" and its observed performance metrics (sensitivity, specificity, PPV, NPV). This is a direct evaluation of the algorithm's performance without the explicit involvement of human readers in the loop as part of the study design for device performance. The device is then intended to provide information to clinicians, but the study described is an algorithm-only evaluation against ground truth.

7. The Type of Ground Truth Used

The ground truth used was based on "a clinical diagnosis of AF in the next 12 months" (Page 6), determined through "duplicate manual chart review" (Page 6) and "sufficient pre- and post-ECG data available to determine that the patient was part of the intended use patient population and to enable at least 1 year of follow-up to determine the presence of a clinical diagnosis of AF." This suggests a combination of medical record review and outcomes data (clinical diagnosis of AF within 12 months based on follow-up).

8. The Sample Size for the Training Set

• Training Set Sample Size: > 1,500,000 ECGs and > 450,000 patients. (Page 6)

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

The document states that the "Tempus ECG-AF model was trained on data from > 1,500,000 ECGs and > 450,000 patients, with 80% of data used for training and 20% of the data used for model tuning." (Page 6)

While it doesn't explicitly detail the methodology for establishing ground truth for the training set, it can be inferred that it involved similar processes to the test set, likely leveraging existing clinical diagnoses of AF from electronic health records or other forms of medical documentation, given the large scale of the dataset. The text does not provide specific details on manual review or expert involvement for the training set's ground truth beyond "data from" ECGs and patients.

Ask a specific question about this device

The CorVista® System analyzes sensor-acquired physiological signals of patients presenting with cardiovascular symptoms (such as chest pain, dyspnea, fatigue) to provide a binary output indicating the likelihood of elevated mean pulmonary arterial pressure (mPAP), an indicator of pulmonary hypertension. The analysis is presented for interpretation by healthcare providers in conjunction with their clinical judgment, the patient's signs, symptoms, and clinical history as an aid in diagnosis.

The CorVista® System is a non-invasive medical device system comprised of several hardware and software components that are designed to work together to allow a physician to evaluate the patient for the presence of cardiac disease indicators, using a static detection algorithm.

The CorVista System has a modular design, where disease-specific "Add-On Modules" will integrate with a single platform, the CorVista Base System, to realize its intended use. The Cor Vista Base System is a combination of hardware, firmware, and software components with the functionality to acquire, transmit, store, and analyze data, and to generate a report for display in a secure web-based portal. The architecture of the CorVista Base system allows for integration with indication-specific "Add-Ons" which perform data analysis using a machine learned detection algorithm to indicate the likelihood of specific diseases at point of care. The PH Add-On indicates the likelihood of elevated mean pulmonary arterial pressure (mPAP), an indicator of pulmonary hypertension. The analysis is presented for interpretation by healthcare providers in conjunction with their clinical judgment, the patient's signs, symptoms, and clinical history as an aid in diagnosis.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria are implied by the reported performance metrics, which the document states "passed the pre-specified secondary endpoint."

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

• Sample Size: N = 386 subjects. These subjects were divided into Population A (elevated mPAP population for Sensitivity Testing) and Population B (non-elevated mPAP population for Specificity Testing).
• Data Provenance: The study was a prospective, multicenter, non-randomized, repository study. The document does not explicitly state the country of origin, but "Analytics for Life, Inc." is located in Toronto, ON, Canada, which might suggest Canadian or North American centers.

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

• Ground Truth Establishment: Ground truth was established via "invasive right heart catheterization (RHC)" and "core lab adjudicated Transthoracic echocardiogram (TTE)."
• Number and Qualifications of Experts:
  • For RHC: Not specified, but generally performed by interventional cardiologists or pulmonologists.
  • For TTE: "Core lab adjudicated." The number of experts involved in the core lab adjudication and their specific qualifications (e.g., "radiologist with 10 years of experience") are not specified in the provided text.

4. Adjudication Method for the Test Set

The document states "core lab adjudicated TTE" was used. The specific adjudication method (e.g., 2+1, 3+1) is not specified.

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

No, an MRMC comparative effectiveness study comparing human readers with AI vs. without AI assistance was not explicitly mentioned or described in the provided text. The study focused on the standalone performance of the CorVista System.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done

Yes, the described clinical testing evaluates the standalone performance of the CorVista System with PH Add-On. The results (sensitivity, specificity, NPV, AUC-ROC) are for the algorithm's prediction compared to the established ground truth. The device is intended to be used "in conjunction with their clinical judgment," but the performance metrics provided are for the algorithm's direct output.

7. The Type of Ground Truth Used

The ground truth used was guideline-driven ground truth via invasive catheterization (Right Heart Catheterization for elevated mPAP) or core-lab adjudicated Transthoracic echocardiogram (TTE) for non-elevated mPAP. This combines a definitive invasive measure with an expert-adjudicated non-invasive imaging modality.

8. The Sample Size for the Training Set

The sample size for the training set is not specified in the provided text. The document refers to model training and validation but only provides details about the clinical validation (test set) population.

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

The document states: "Guideline-driven ground truth via invasive catheterization or core-lab adjudicated TTE." This method described for ground truth establishment for the test set is also stated as the method for "Ground Truth for Model Training and Validation," implying the same approach was used for the training data, although specific details on how this was applied to the training set are not elaborated upon.

Ask a specific question about this device

Eko Low Ejection Fraction Tool (ELEFT) is a software intended to aid clinicians in identifying individuals with Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. ELEFT takes as input ECG and heart sounds and is intended for use on patients at risk for heart failure. This population includes, but is not limited to, patients with: coronary artery disease; diabetes mellitus; cardiomyopathy; hypertension; and obesity.

The interpretations of heart sounds and ECG offered by the software are meant only to assist healthcare providers in assessing Left Ventricular Ejection Fraction ≤ 40% , who may use the result in conjunction with their own evaluation and clinical judgment. It is not a diagnosis or for monitoring of patients diagnosed with heart failure. This software is for use on adults (18 years and older).

Eko Low Ejection Fraction Tool (ELEFT) is an algorithm that is intended to aid clinicians to identify individuals with Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. ELEFT takes as input ECG and heart sounds from patients at risk for heart failure. The software uses signal processing as well as machine learning algorithms, to analyze the electrocardiogram (ECG) and heart sound/phonocardiogram (PCG) recording signals generated by FDA-cleared Eko Stethoscopes and saved as .WAV file recordings in the Eko Cloud. ELEFT is a machine learning based notification software which employs machine learning techniques to suggest the likelihood of LVEF < 40% for further referral or diagnostic follow-up. It is intended as the basis for further testing and is not intended to provide diagnostic quality output. As an integral part of a physical assessment, clinician's interpretations of this data can help identify previously undiagnosed left ventricular dysfunction in a patient.

The ELEFT consists of the following algorithm components:
· Eko Low Ejection Fraction Tool API
· Waveform Analysis:

The Eko Low Ejection Fraction Tool (ELEFT) is a software intended to aid clinicians in identifying individuals with Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. The device takes ECG and heart sound inputs and processes them using signal processing and machine learning algorithms.

Here's an analysis of its acceptance criteria and the study proving its performance:

1. Table of Acceptance Criteria and Reported Device Performance

The provided document doesn't explicitly state "acceptance criteria" in a numerical target format (e.g., "Sensitivity must be >= X%"). However, the clinical performance results presented demonstrate the device's capability to detect Low EF. The acceptance effectively hinges on the presented sensitivity and specificity values.

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

• Test Set Sample Size: 3,456 unique subjects. After excluding 307 recordings due to poor ECG quality, the performance analysis was based on the remaining suitable recordings.
• Data Provenance: Retrospective data collected from:
  • US, 5 sites: 2,960 patients.
  • India, 1 site: 496 patients.

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

• Number of Experts: Not explicitly stated as a number, but the ground truth for ejection fraction was "overread by a board-certified cardiologist." This implies at least one, and potentially multiple, board-certified cardiologists were involved in reviewing the echocardiogram results.
• Qualifications of Experts: Board-certified cardiologists.

4. Adjudication Method for the Test Set

The document does not explicitly describe an adjudication method like 2+1 or 3+1 for resolving discrepancies in ground truth establishment. It states that the "subject's true ejection fraction was measured by the echocardiogram machine's integrated cardiac quantification software at the echocardiogram and then overread by a board-certified cardiologist." This suggests a single expert review after automated measurement, with no mention of multiple reviewers or a formal reconciliation process if initial measurements or interpretations differed.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not conducted. The study focuses solely on the standalone performance of the ELEFT algorithm without a human-in-the-loop component or evaluating the improvement of human readers with AI assistance.

6. Standalone (Algorithm Only) Performance

Yes, a standalone (algorithm only) performance study was conducted. The results for sensitivity, specificity, PPV, and NPV presented in Table 2 and the subsequent text (page 9) are for the ELEFT algorithm's performance in differentiating between Low EF (≤40%) and Normal EF (>40%).

7. Type of Ground Truth Used

The type of ground truth used was expert consensus / pathology based on instrumental measurements and expert review:

• Echocardiogram (Instrumental Measurement): The true ejection fraction was measured by the echocardiogram machine's integrated cardiac quantification software.
• Expert Overread: This measurement was "overread by a board-certified cardiologist."
• Categorization: Ejection status (Low EF or Normal EF) was then assigned based on these measured and reviewed values.

8. Sample Size for the Training Set

The sample size for the training set was 1,852 patients. This data was contributed from:

• US, 7 sites: 1,515 patients.
• India, 1 site: 337 patients.

9. How Ground Truth for the Training Set Was Established

The document does not explicitly detail the exact process for establishing ground truth for the training set. However, given the consistency in the data description and the validation methodology, it is highly probable that the ground truth for the training set was established using the same methodology as the test set: gold standard echocardiogram measurements, subsequently overread by board-certified cardiologists, and then categorized into Low EF (≤40%) or Normal EF (>40%).

Ask a specific question about this device

The Anumana Low Ejection Fraction AI-ECG Algorithm is software intended to aid in screening for Left Ventricular Ejection Fraction (LVEF) less than or equal to 40% in adults at risk for heart failure. This population includes, but is not limited to:
· patients with cardiomyopathies

• patients who are post-myocardial infarction
• · patients with aortic stenosis
• · patients with chronic atrial fibrillation
• · patients receiving pharmaceutical therapies that are cardiotoxic, and
  • postpartum women.

Anumana Low Ejection Fraction Al-ECG Algorthm is not intended to be a stand-alone diagnostic device for cardiac conditions, should not be used for monitoring of patients, and should not be used on ECGs with a paced rhythm.

A positive result may suggest the need for further clinical evaluation in order to establish a diagnosis of Left Ventricular Ejection Fraction (LVEF) less than or equal to 40%. Additionally, if the patient is at high risk for the cardiac condition, a negative result should not rule out further non-invasive evaluation.

The Anumana Low Ejection Fraction AI-ECG Algorithm should be applied jointly with clinician judgment.

The Low Ejection Fraction AI-ECG Algorithm interprets 12-lead ECG voltage times series data using an artificial intelligence-based algorithm. The device analyzes 10 seconds of a single 12lead ECG acquisition, and within seconds provides a prediction of likelihood of LVEF (ejection fraction less than or equal to 40%) to third party software. The results are displayed by the third-party software on a device such as a smartphone, tablet, or PC. The Low Ejection Fraction AI-ECG Algorithm was trained to predict Low LVEF using positive and control cohorts, and the prediction of Low LVEF in patients is generated using defined conditions and covariates. The Low Ejection Fraction AI-ECG Algorithm device is intended to address the unmet need for a point-of-care screen for LVEF less than or equal to 40% and is expected to be used by cardiologists, front-line clinicians at primary care, urgent care, and emergency care settings, where cardiac imaging may not be available or may be difficult or unreliable for clinicians to operate. Clinicians will use the Low Eiection Fraction AI-ECG Algorithm to aid in screening for LVEF less than or equal to 40% and making a decision for further cardiac evaluation.

Here's a breakdown of the acceptance criteria and the study proving the device meets those criteria, based on the provided FDA 510(k) clearance letter for the Low Ejection Fraction AI-ECG Algorithm:

Low Ejection Fraction AI-ECG Algorithm: Acceptance Criteria and Performance Study

1. Table of Acceptance Criteria and Reported Device Performance

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

• Sample Size for Test Set: The clinical validation study included 16,000 patient records initially, though 2,040 records were excluded due to quality checks, resulting in a final analysis sample of 13,960 patient-ECG pairs.
• Data Provenance: The data was retrospective, collected from 4 health systems across the United States.

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

The document does not specify the number of experts or their qualifications used to establish the ground truth for the clinical validation test set. The ground truth (LVEF <= 40% or > 40%) was derived from transthoracic echocardiogram (TTE) measurements. While TTE interpretation requires expertise, the document doesn't detail the method of expert review or consensus for these TTE results themselves for the test set.

4. Adjudication Method for the Test Set

The document does not specify an adjudication method (e.g., 2+1, 3+1) for the ground truth for the test set. The ground truth was established by TTE measurements.

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

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. The study evaluated the standalone performance of the AI algorithm against a ground truth without human readers in the loop.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done

Yes, a standalone performance study was done. The reported sensitivity and specificity values are for the algorithm's performance alone in detecting low LVEF.

7. The Type of Ground Truth Used

The type of ground truth used for both training and validation was objective clinical measurements from Transthoracic Echocardiogram (TTE), specifically the Left Ventricular Ejection Fraction (LVEF) measurement. An LVEF of $\le$ 40% was defined as the disease cohort, and > 40% as the control cohort.

8. The Sample Size for the Training Set

The training set for the algorithm development consisted of 93,722 patients with an ECG and TTE performed within a 2-week interval. These were split into:

• Training dataset: 50% of the 93,722 patients.
• Tuning dataset: 20% of the 93,722 patients.
• Set-aside testing dataset: 30% of the 93,722 patients (used for internal validation during development, distinct from the independent clinical validation study).

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

The ground truth for the training set was established using LVEF measurements obtained from transthoracic echocardiograms (TTE). Specifically, for each patient, the LVEF measurement from the earliest TTE within a 2-week interval of an ECG was paired with the closest ECG recording. LVEF $\le$ 40% defined the disease cohort, and LVEF > 40% defined the control cohort. This data was identified from a research-use authorized clinical database from Mayo Clinic.

Ask a specific question about this device

The CorVista® System analyzes sensor-acquired physiological signals of patients presenting with cardiovascular symptoms (such as chest pain, dyspnea, fatigue) to indicate the likelihood of significant coronary artery disease. The analysis is presented for interpretation by healthcare providers in conjunction with their clinical judgment, the patient's signs, symptoms, and clinical history as an aid in diagnosis.

The CorVista® System is a non-invasive medical device system comprised of several hardware and software components that are designed to work together to allow a physician to evaluate the patient for the presence of cardiac disease, or cardiac disease indicators, using a static detection algorithm. The CorVista System has a modular design, where disease-specific "Add-On Modules" will integrate with a single platform, the CorVista Base System, to realize its intended use. The CorVista Base System is a combination of hardware, firmware, and software components with the functionality to acquire, transmit, store, and analyze data, and to generate a report for display in a secure web-based portal. The architecture of the CorVista Base system allows for integration with indication-specific "Add-Ons" which perform data analysis using a machine learned detection algorithm to indicate the likelihood of specific diseases at point of care. The CAD Add-On indicates the likelihood of significant Coronary Artery Disease (CAD). The analysis is presented for interpretation by healthcare providers in conjunction with their clinical judgment, the patient's signs, symptoms, and clinical history as an aid in diagnosis.

Here's a breakdown of the acceptance criteria and the study proving the CorVista® System meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The text does not explicitly state pre-defined acceptance criteria in a quantitative format (e.g., "Sensitivity >= X%"). Instead, it presents the device's performance results and implies that these results were deemed acceptable for substantial equivalence to the predicate device. The comparison to CCTA's "rule out performance" suggests a benchmark, but not a strict acceptance criterion.

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

• Test Set Sample Size: N = 1,816 subjects.
  • Population A (CAD+ for Sensitivity Testing): Number not specified, but this population was evaluated for sensitivity.
  • Population B (CAD- for Specificity Testing): Number not specified, but this population was evaluated for specificity.
• Data Provenance: Prospective, multicenter, non-randomized, repository study. The text does not explicitly state the country of origin, but given the FDA submission, it implicitly refers to data collected in the US.

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

The text states that the ground truth for CAD was established via "invasive catheterization (ICA)" or "core-lab adjudicated CTA."

• Number of Experts: Not explicitly stated for ICA or CTA adjudication.
• Qualifications of Experts: It implies that medical professionals performed the ICA, and a core-lab performed the CTA adjudication. The specific qualifications (e.g., number of years of experience for radiologists or cardiologists performing these procedures/adjudications) are not detailed.

4. Adjudication Method for the Test Set

The adjudication method for the ground truth was:

• For ICA: Clinical outcome from invasive coronary angiography. This is a direct, invasive diagnostic procedure.
• For CTA: "Core-lab adjudicated CTA." This implies a standardized process by a specialized lab, likely involving multiple readers or a defined quality control process, but the specific multi-reader method (e.g., 2+1, 3+1) is not provided.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done

• No, an MRMC comparative effectiveness study was not explicitly stated to have been done for human readers with and without AI assistance to assess improvement. The study described focuses on the standalone performance of the CorVista System compared to established diagnostic methods (ICA/CTA). The device is intended to be an "aid in diagnosis" used "in conjunction with their clinical judgment," but the study design presented does not evaluate the human-AI interaction in a comparative effectiveness study setting.

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

• Yes, a standalone performance study was done. The described clinical testing focuses on the algorithm's performance in indicating the likelihood of significant CAD by comparing its predictions to objective ground truth (ICA/CTA results). The reported sensitivity, specificity, and AUC-ROC are measures of the algorithm's standalone performance.

7. The Type of Ground Truth Used

• Objective Clinical Data / Outcomes Data: The ground truth for the test set was established by:
  • Invasive Coronary Angiography (ICA): This is considered a gold standard for diagnosing CAD.
  • Core-lab Adjudicated Coronary Computed Tomography Angiography (CTA): This is another strong diagnostic imaging modality, with the "core-lab adjudicated" aspect indicating a high level of rigor in interpretation.
    These methods directly determine the patient's actual CAD classification (CAD+ or CAD-).

8. The Sample Size for the Training Set

• The text states the ground truth for the "Model Training and Validation" was "Guideline-driven ground truth via invasive catheterization or core-lab adjudicated CTA." However, the specific sample size for the training set is not provided. The N=1,816 refers to the validation population (test set) used for performance testing.

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

• The ground truth for model training (and validation) was established using "Guideline-driven ground truth via invasive catheterization or core-lab adjudicated CTA." This implies that the same rigorous, objective diagnostic methods used for the test set's ground truth were also used to label the data utilized during the training and internal validation phases of the algorithm development.

Ask a specific question about this device

Viz HCM is intended to be used in parallel to the standard of care to analyze recordings of 12-lead ECG made on compatible ECG devices. Viz HCM is capable of analyzing the ECG, detecting signs associated with hypertrophic cardiomyopathy (HCM), and allowing the user to view the ECG and analysis results. Viz HCM is indicated for use on 12-lead ECG recordings collected from patients 18 years of age or older. Viz HCM is not intended for use on patients with implanted pacemakers. Viz HCM is limited to analysis of ECG data and should not be used in-lieu of full patient evaluation or relied upon to make or confirm diagnosis. Viz HCM identifies patients for further HCM follow-up and does not replace the current standard of care methods for diagnosis of HCM. The results of the device are not intended to rule-out HCM follow-up.

The Viz HCM ECG Analysis Algorithm (HCM Algorithm) is a machine learning-based software algorithm that analyzes 12-lead electrocardiograms (ECGs) for characteristics suggestive of hypertrophic cardiomyopathy (HCM). The mobile software module enables the end user to receive and toggle notifications for ECGs determined by the Viz HCM ECG Analysis Algorithm to contain signs suggestive of HCM.

The Viz HCM is a Software as a Medical Device (SaMD) intended to analyze ECG signals collected as part of a routine clinical assessment, independently and in parallel to the standard of care. Viz HCM is a combination of software modules that consists of an ECG analysis software algorithm and mobile application software module.

Here's a breakdown of the acceptance criteria and the study proving the Viz HCM device meets them, based on the provided text:

Acceptance Criteria and Device Performance

The core acceptance criteria for the Viz HCM device are implicitly defined by the sponsor's performance metrics and the explicit special controls outlined by the FDA. The performance testing section provides the evidence that the device meets these criteria.

1. Table of Acceptance Criteria and Reported Device Performance

Given that this is a De Novo request, specific pre-defined quantitative acceptance criteria (e.g., "Sensitivity must be > X%") are often not explicitly stated upfront in the narrative. Instead, the "Performance Testing" section presents the demonstrated performance as evidence for acceptance. The FDA then evaluates if this performance is acceptable given the device's intended use and risks.

Based on the provided text, the key performance metrics and their reported values are:

Implicit Acceptance Criteria (from Special Controls and Risk Analysis):

• Clinical Performance Testing (Special Control 1):
  • Device performs as intended under anticipated conditions of use.
  • Clinical validation uses a test dataset of real-world data from a representative patient population.
  • Data is representative of sources, quality, and encountered conditions.
  • Test dataset is independent from training/development data.
  • Sufficient cases from important cohorts (demographics, confounders, comorbidities, hardware/acquisition characteristics) are included for subgroup analysis.
  • Study protocols include ground truth adjudication processes.
  • Consistency of output demonstrated over the full range of inputs.
  • Performance goals justified in context of risks.
  • Objective performance measures reported with descriptive/developmental measures.
  • Summary-level demographic and subgroup analyses provided.
  • Test dataset includes a minimum of 3 geographically diverse sites (separate from training).
• Software Verification, Validation, and Hazard Analysis (Special Control 2):
  • Model description, inputs/outputs, patient population.
  • Integration testing in intended system.
  • Impact of sensor acquisition hardware on performance.
  • Input signal/data quality control.
  • Mitigations for user error/subsystem failure.
• Human Factors Assessment (Special Control 3):
  • Evaluates risk of misinterpretation of device output.
• Labeling (Special Control 4):
  • Summary of performance testing, hardware, patient population, results, demographics, subgroup analyses, minimum performance.
  • Device limitations/subpopulations where performance may differ.
  • Warning against ruling out follow-up based on negative finding.
  • Statement that output shouldn't replace full clinical evaluation.
  • Warnings on sensor acquisition factors impacting results.
  • Guidance for interpretation and typical follow-up.
  • Type of hardware sensor data used.

Study Details for Proving Acceptance

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

• Test Set Sample Size: 3,196 ECG cases (291 HCM-Positive and 2905 HCM-Negative).
• Data Provenance: Retrospective study. Data collected from 3 hospitals in the US (Boston, Massachusetts area - 2 sites; Salem, Massachusetts - 1 site). The Boston sites are described as racially and ethnically diverse, while the Salem site was predominantly Caucasian or Latino. Data was collected between July 1, 2017, and June 30, 2022.

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

• Number of Experts: A single cardiologist performed the initial chart and imaging review for each HCM-Positive or HCM-Negative case to establish the ground truth.
• Qualifications of Experts: Described as "cardiologist." No further details on their years of experience or specific board certifications are provided in the excerpt. A "second cardiologist" was used for a secondary assessment on a subset of cases to check agreement/consistency.

4. Adjudication Method for the Test Set

• Method: A single cardiologist established the ground truth for each case through chart and imaging review based on predefined guidelines (Cornell criteria or Sokolow-Lyon criteria).
• Consistency Check: A "secondary assessment" was performed on a selection of 60 cases (30 HCM-Positive, 30 HCM-Negative) where a second cardiologist independently truthed the cases to perform an analysis of agreement/consistency. The results of this agreement analysis are not detailed, but the method was a 1+1 adjudication for a subset. For the main test set, it was effectively a "none" (single expert review) or rather an individual expert labeling.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done

• No MRMC Study was described. The provided text focuses on the standalone performance of the algorithm and does not include a comparative effectiveness study involving human readers with and without AI assistance. The device is intended to be used "in parallel to the standard of care," suggesting it provides an additional signal, not necessarily assistance to human readers interpreting ECGs.

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

• Yes, a standalone performance study was done. The entire "PERFORMANCE TESTING" section, especially "SUMMARY OF CLINICAL INFORMATION," describes the performance of the Viz HCM algorithm in identifying suspected HCM from ECGs compared directly to the clinical ground truth established by cardiologists. The reported sensitivity, specificity, and PPV are all "algorithm-only" performance metrics.

7. The Type of Ground Truth Used

• Expert Consensus/Clinical Records Review: The ground truth for the test set was established by a cardiologist (single expert for primary truth, with a second expert for consistency check on a subset) who performed a chart and imaging review for each patient. This was based on "predefined guidelines using either the Cornell criteria or the Sokolow-Lyon criteria." ICD-10 codes were used for initial sampling, but the definitive ground truth was established by clinical review. This is a form of expert consensus/clinical documentation ground truth.

8. The Sample Size for the Training Set

• Training Set Sample Size: 301,106 patients, encompassing 831,329 ECG exams.
  • HCM positive patients: 4,470
  • HCM negative patients: 298,394

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

• The text states: "The data for algorithm development was collected from different US and Non-US (OUS) sources. The data contains both HCM Positive (obstructive and nonobstructive) and HCM Negative examples including random ECG samples (random control) and enrichment for conditions differential for and associated with HCM (negative controls)."
• It further clarifies that for HCM-Negative cases in the development (training and internal validation) dataset, absence of HCM was determined by the "lack of ICD-9/10 code for HCM."
• For HCM-Positive and HCM-Negative cases with available imaging, "additional chart review and review of imaging provided more confidence into the label."

In summary, for the training set, the ground truth was established primarily through ICD-9/10 codes, supplemented by chart review and imaging review where available. This suggests a semi-automated, large-scale labeling approach for the training data, potentially with manual review for confirmation or difficult cases.

Ask a specific question about this device