Search Results

The SafeBeat Rx App analyzes ECG data recorded in compatible formats. This ECG signal may originate from a full 12-lead ECG or a reduced lead set ECG. The device provides ECG signal processing and provisional analysis, including interval measurements (QT interval, QTc, QRS duration, heart rate, and RR interval), along with visible display of QRS onset (Q-start), R-peak, and QRS offset (S-end). The SafeBeat Rx App can electronically interface with, and perform analysis of, data transferred from other computer-based ECG systems, such as an ECG management system. The SafeBeat Rx App does not provide real-time ECG display, continuous monitoring, or alarm functions. The device is not for use in life-supporting or sustaining systems or ECG monitor and alarm devices. The SafeBeat Rx App ECG analysis is intended for adult patient populations (18 years and older).

SafeBeat Rx App provisional ECG analysis is not intended to be the sole means of diagnosis. The SafeBeat Rx App is not validated for use in lead I alone. The SafeBeat Rx App is intended to be used on an advisory basis only by qualified healthcare personnel to evaluate provisional ECG data. ECG data should be reviewed in conjunction with the patient's clinical history, symptoms, and/or other diagnostic tests, and the professional clinical judgement of the qualified healthcare provider.

The SafeBeat Rx App can be used in a professional healthcare environment such as a hospital, clinic or similarly equipped facility. The SafeBeat Rx App has an optional long term monitoring workflow intended for monitoring and evaluating a patient's home acquired ECG. The software workflow that is intended for use in the professional healthcare environment should not be used in the home environment to adjust QT prolonging medications as is contraindicated for applicable drugs.

The SafeBeat Rx App is a Software as a Medical Device (SaMD) that provides: (1) ML-based provisional ECG interval measurements of third-party ECG signals (e.g., HR, RR-interval variability, QT/QTc interval and QRS interval); and (2) optional non-device functions, including suggested antiarrhythmic drug (AAD) dosing consistent with manufacturer drug label for amiodarone, dofetilide, flecainide, sotalol and IV sotalol. The device analyzes ECG signals acquired by other ECG acquisition and storage devices. The device is only intended for traditional "wet" electrode inputs. The SafeBeat app does not directly acquire ECG data from patients. ECG data is obtained programmatically through an application programming interface (API) with the ECG acquisition and storage device, or manually via data upload through a secure web interface. The device is solely intended to analyze raw digital ECG data and does not allow the analysis of ECG signals imported by images.

Provisional ECG analysis is performed by the device. The device includes both beat-level feature identification and interval estimation. The beat-level parameters are:

• R-peak
• QRS onset
• ST onset
• T-wave offset

The interval estimation parameters:

• Heart rate
• RR interval variability
• QRS duration
• QT interval
• QT interval variability
• Heart rate corrected QT (e.g., QTcF)
• Heart rate corrected QT variability

Provisional ECG interval measurements are displayed on a user interface for review and interpretation by a qualified healthcare professional. The provisional ECG analysis can be viewed, edited, approved, or rejected by the qualified healthcare professional via the user interface.

The SafeBeat App does not provide continuous cardiac monitoring. The SafeBeat App does not provide rhythm interpretation or diagnosis cardiac arrhythmias (e.g. atrial fibrillation). The device does not include automated rhythm analysis. The device is intended for adult patient populations.

Here's an analysis of the acceptance criteria and the study proving the device meets them, based on the provided FDA 510(k) clearance letter for the SafeBeat Rx App:

1. Table of Acceptance Criteria and Reported Device Performance

The FDA letter does not explicitly state acceptance criteria in numerical targets (e.g., "accuracy must be > 90%"). Instead, the performance studies are designed to demonstrate "excellent agreement" with expert annotations and effective/accurate measurements. The key performance metrics reported are primarily related to agreement or error compared to a ground truth.

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

• SafeBeat Proprietary Validation Dataset: The exact sample size in terms of number of ECGs or patients is not explicitly stated, but it's described as a "retrospective testing using publicly available clinical datasets." It included a "broad spectrum of ECG morphologies" from "diverse patient populations collected across multiple geographically distinct locations, encompassing healthy individuals, patients in critical care settings, and patients with known arrhythmias."
  • Provenance: Retrospective, from "multiple independent sources" and "multiple geographically distinct locations." Race/ethnicity distribution included White (60.9%), Asian (3.8%), Black or African American (10%), Hispanic or Latino (4.8%), Other (4.1%), or Unknown (16.5%).
• Common Standards for Electrocardiography (CSE) Dataset: n=100 ECGs.
  • Provenance: Not explicitly stated, but it's a "CSE reference measurements" dataset.
• Standard ECG Test Databases (IEC 60601-2-47): Used the MIT-BIH Normal Sinus Rhythm dataset, AHA database, and MIT-BIH Noise Stress Test dataset.
  • Provenance: Well-known publicly available datasets.

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

• SafeBeat Proprietary Validation Dataset: Ground truth annotations were performed by an unspecified number of "board-certified cardiologists." Specific experience levels (e.g., "10 years of experience") are not provided.
• Common Standards for Electrocardiography (CSE) Dataset: "Manual CSE reference measurements" were used. The number and qualifications of the annotators for this reference dataset are not specified in the document.
• Standard ECG Test Databases (IEC 60601-2-47): These databases inherently contain established annotations, but the number and specific qualifications of the original annotators for these public datasets are not detailed within this FDA letter.

4. Adjudication Method for the Test Set

The document does not explicitly describe an adjudication method (e.g., 2+1, 3+1) for the SafeBeat proprietary dataset. It states annotations were "performed by board-certified cardiologists," which implies individual expert annotations were used as ground truth without further detail on how discrepancies (if multiple experts were involved) were resolved. For the CSE and IEC datasets, pre-existing reference measurements are used, so a separate adjudication method for the SafeBeat study is not mentioned.

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

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study comparing human readers with AI assistance versus without AI assistance was not conducted or reported in this 510(k) submission. The study focuses on the standalone performance of the AI algorithm in measuring ECG intervals against expert annotations and established reference databases.

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

Yes, the studies described are primarily standalone performance evaluations of the SafeBeat Rx App algorithm. The software's measurements (QTc, QRS, HR, R-R) are compared directly against expert annotations and reference measurements, indicating algorithm-only performance. The device is intended "to be used on an advisory basis only by qualified healthcare personnel," meaning a human-in-the-loop will review and potentially adjust the output, but the validation itself is of the algorithm's initial output.

7. The Type of Ground Truth Used

• Expert Consensus/Annotation: For the SafeBeat Proprietary Validation Dataset, ground truth was established by "board-certified cardiologists" annotations.
• Reference Measurements/Established Datasets: For the CSE and IEC 60601-2-47 datasets, pre-existing "reference measurements" or established annotations from these standard databases were used as ground truth.

8. The Sample Size for the Training Set

The document states, "The training dataset consisted of broad distribution of cardiac rhythms and less common supraventricular rhythms." However, the sample size (number of ECGs or patients) for the training set is not provided.

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

The document only states that "The training dataset consisted of broad distribution of cardiac rhythms and less common supraventricular rhythms. QRS and QTc morphology were diverse. The dataset ensured generalization across age, sex, rhythm classes and ECG waveform variations." It does not specify how the ground truth for the training set was established (e.g., by experts, automated methods).

Ask a specific question about this device