FDA 510(k) Search - By Innolitics (SaMD and AI Experts)

Patients with clinical syndromes or situations at increased risk of cardiac arrhythmias
Patients who experience transient symptoms that may suggest a cardiac arrhythmia
The device has not been tested for and is not intended for pediatric use

Device Description

The BioMonitor 2 is a small, leadless, implantable device that uses two electrodes on the body of the device to monitor continuously the patient's subcutaneous ECG. The BioMonitor 2 is designed to record automaticallythe occurrence of arrhythmias in a patient. Recordings can also be triggered by use of the associated Remote Assistant. Arrhythmia may be classified as atrial fibrillation, bradycardia, asystole, or high ventricular rate. The device memory can automatically store a maximum of 55 separately recorded SECG-episodes of 40 seconds each, and 4 patient triggered SECG-episodes of 7.5 minutes.

AI/ML Overview

The provided text describes the BioMonitor 2, an implantable cardiac monitor intended to detect various cardiac arrhythmias. However, the document is a 510(k) premarket notification summary from the FDA, and as such, it focuses on demonstrating substantial equivalence to a predicate device rather than providing a detailed technical "acceptance criteria" table with specific metrics (like sensitivity, specificity, accuracy) for arrhythmia detection algorithms or a direct "study that proves the device meets the acceptance criteria" in the way one might expect for a novel AI/ML-based diagnostic device.

The document mentions a clinical study and software verification/validation, but it does not provide the granular details requested in your prompt regarding:

Specific performance metrics (e.g., sensitivity, specificity) for arrhythmia detection.
Quantitative results showing the device performance against specific acceptance criteria.
Details about the ground truth establishment (number/qualifications of experts, adjudication methods).
Sample sizes for training sets or the provenance of training data.
Any MRMC studies or human-in-the-loop performance improvements.
Standalone algorithm performance.

The "acceptance criteria" presented are primarily functional (what the device does and what arrhythmias it detects) and safety/technological (MRI compatibility, software validation, meeting standards), rather than detailed performance metrics of its arrhythmia detection capability against a ground truth dataset.

Given these limitations of the source text, I will construct a response based on what is provided and explicitly state where information is not available in the document.

Acceptance Criteria and Device Performance for BioMonitor 2

Based on the provided FDA 510(k) summary for the BioMonitor 2, the acceptance criteria are implicitly defined by its intended use and functional capabilities, and its demonstrated substantial equivalence to a predicate device. The document does not provide specific quantitative acceptance criteria for arrhythmia detection performance (e.g., specific sensitivity or specificity thresholds) in the way one might see for a diagnostic AI algorithm. Instead, the "performance" demonstrated is primarily about the device's ability to consistently perform its intended functions and its safety.

Here's a table based on the functional aspects the device is indicated to detect:

1. Table of Acceptance Criteria (Implicit) and Reported Device Performance (Functional)

Functional/Performance Aspect	Acceptance Criteria (Implicit from Indications for Use and Device Description)	Reported Device Performance (as stated in the document)
Arrhythmia Detection:	The device must be capable of detecting the specified arrhythmias.	Indicated to detect:
Atrial Fibrillation	Capable of detecting Atrial Fibrillation.	Detected by BioMonitor 2.
Bradycardia	Capable of detecting Bradycardia.	Detected by BioMonitor 2.
Sudden Rate Drop	Capable of detecting Sudden Rate Drop.	Detected by BioMonitor 2.
High Ventricular Rate (HVR)	Capable of detecting HVR.	Detected by BioMonitor 2.
Asystole	Capable of detecting Asystole.	Detected by BioMonitor 2.
Recording Capability:	Must accurately record subcutaneous ECG (sECG) episodes.	Records automatically and via patient trigger.
Auto-activated events	Stores auto-activated sECG episodes.	Up to 55 episodes of 40 seconds each.
Patient-triggered events	Stores patient-triggered sECG episodes.	4 episodes of 7.5 minutes each.
Storage Duration	Sufficient sECG storage capacity.	>66 minutes total (30 min patient-triggered, 36.7 min auto-activated).
Safety/Usability:	Must be safe and function as intended in MRI environments.	Cleared for 1.5T and 3.0T full-body MRI scans.
Software Integrity	Software must be verified and validated to high standards.	Software considered "major" level of concern; verification/validation conducted per FDA guidance.
Compliance with Standards	Must comply with relevant medical device standards.	Tested in accordance with EN 45502, ISO 11135, ISO 14708, ISO/TS 10974.
Clinical Implantation	Implantation procedure and sensing quality must be demonstrated.	Data from 30 patients demonstrating implantation and sensing quality.

Study Details Proving Device Meets Acceptance Criteria

The document refers to a "Clinical Study" and "Software Verification and Validation Testing" as evidence supporting substantial equivalence, which implicitly means meeting the functional "acceptance criteria" described above and ensuring safety.

2. Sample Size and Data Provenance:

Test Set Sample Size: 30 patients were included in the primary clinical study.
Data Provenance: Data sourced from 5 Australian clinical sites.
Retrospective/Prospective: The study duration (December 18, 2014, through July 06, 2015) suggests it was a prospective clinical data collection. An additional "Post Market Observation (PMO)" commenced in Europe with 15 implantations as of October 9, 2015, which would also be prospective.

3. Number of Experts and Qualifications for Ground Truth:

The document does not specify the number of experts used to establish the ground truth for arrhythmia detection, nor their qualifications (e.g., electrophysiologists, cardiologists, years of experience). The study objective was on implantation procedure and sensing quality, not explicitly on algorithmic arrhythmia detection accuracy against a human expert reference standard.

4. Adjudication Method for the Test Set:

The document does not specify any adjudication method for establishing ground truth for arrhythmia events. Given the focus on "sensing quality," it's plausible that sensed events were reviewed, but the process is not detailed.

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

No, an MRMC comparative effectiveness study was not mentioned. The study's objective was to demonstrate the sensing quality and implantation procedure of the BioMonitor 2, not to compare human reader performance with and without AI assistance. This device is an implanted monitor, not an AI-assisted diagnostic workstation for human readers.

6. Standalone (Algorithm Only) Performance Study:

The document does not explicitly describe a standalone algorithm-only performance study with quantitative metrics like sensitivity, specificity, or accuracy for arrhythmia detection. The "Clinical Study" assesses general sensing quality in implanted patients. The device itself is an algorithm-driven monitor, so its in-vivo performance is its "standalone" performance in the clinical context. However, the study results do not include specific performance metrics for individual arrhythmia types.

7. Type of Ground Truth Used:

The document does not explicitly state how the ground truth for arrhythmia events was established (e.g., expert consensus, independent ECG reviews, pathology, outcomes data). The study assessed "sensing quality," which implies that the device's ability to sense and record heart activity was evaluated, but the method for verifying the true occurrence and classification of arrhythmias against which the device's output would be measured is not detailed. It's likely that adjudicated clinical events from the patient's records or an independent review of the collected sECG data constituted the ground truth but this is not confirmed.

8. Sample Size for the Training Set:

The document does not provide any information about the sample size used for training the device's algorithms. As an implanted cardiac monitor, its algorithms for arrhythmia detection are likely developed and validated using extensive physiological data, but these details are not part of this 510(k) summary.

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

The document does not provide any information on how the ground truth for the training set (if applicable for algorithm development) was established.

In summary, the FDA 510(k) process for the BioMonitor 2 focused on its substantial equivalence to an existing predicate device based on similar indications, technological characteristics, and safety data. While a clinical study was performed, its primary reported outcomes were related to the implantation process and "sensing quality" in a small patient cohort, rather than a detailed, quantitative analysis of its arrhythmia detection performance against a rigorously established ground truth using metrics common for AI/ML diagnostics.

Summary

Regulation Number and Section

§ 870.1025 Arrhythmia detector and alarm (including ST-segment measurement and alarm).

(a)
Identification. The arrhythmia detector and alarm device monitors an electrocardiogram and is designed to produce a visible or audible signal or alarm when atrial or ventricular arrhythmia, such as premature contraction or ventricular fibrillation, occurs.(b)
Classification. Class II (special controls). The guidance document entitled “Class II Special Controls Guidance Document: Arrhythmia Detector and Alarm” will serve as the special control. See § 870.1 for the availability of this guidance document.