Search Results

ENLIGHT 2100 is a non-invasive, non-radiation medical device that provides information of local impedance variation within a cross-section of a patient's thorax. This information is presented to the clinician user as an adjunctive tool to other clinical information in order to support the user's assessment of variations in regional air content within a cross section of a patient's lungs.

It is intended for mechanically ventilated adult and pediatric patients in a hospital setting, whose thorax perimeter is within the range of 37.5 - 134cm.

ENLIGHT 2100 does not measure regional ventilation of the lungs.

ENLIGHT 2100 is a Ventilatory electrical impedance tomograph that uses several electrodes (usually between 16 and 32) placed around the patient's thorax to assess regional impedance variation in a lung slice (tomography). It provides a relative measurement, so it only provides information on variations in local impedance.

ENLIGHT 2100 estimates Local Impedance Variation, occurring in a cross section of the thorax during a respiratory cycles, and which are linearly related to Variations in Regional Air Content within the lung.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided FDA 510(k) clearance letter for the Timpel S.A. Enlight 2100 device:

Important Note: The provided document is a 510(k) clearance letter from the FDA. 510(k) clearance is based on demonstrating substantial equivalence to a previously legally marketed device (predicate device), not necessarily on rigorous clinical effectiveness studies or extensive AI model validation metrics typically seen for novel AI/ML devices. Therefore, the details around specific acceptance criteria, test sets, ground truth establishment, and MRMC studies for AI performance, as might be expected for an AI/ML software as a medical device (SaMD), are not explicitly detailed in this type of regulatory document. The focus here is on demonstrating that the new device (Enlight 2100) is sufficiently similar to the predicate (Enlight 1810) in terms of safety and performance.

Analysis of Acceptance Criteria and Device Performance

The provided document describes bench testing to demonstrate substantial equivalence, rather than clinical studies with human readers or AI-specific performance metrics like sensitivity/specificity for a diagnostic task. The acceptance criteria are implicit in the comparison table (Table 5.1) between the subject device (Enlight 2100) and the predicate device (Enlight 1810). The "Explanation of Differences" column directly indicates whether the performance attributes are similar or if one is slightly better, implying the new device meets or exceeds the predicate's performance.

1. Table of Acceptance Criteria and Reported Device Performance

Acceptance Criteria (Implied: Performance comparable or better than Predicate Device)

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

• Sample Size: The document does not specify a "test set" in terms of patient data or a specific number of unique tests for each performance characteristic. Instead, it refers to "bench testing" which implies testing on a controlled system or phantom, not human subjects, for these specific performance metrics. The comparison table focuses on instrument performance rather than AI model performance on a clinical dataset.
• Data Provenance: Not specified, as these are bench test results of the device itself rather than clinical data from a specific region or patient cohort. It is implicitly "prospective" bench testing of the new device.

3. Number of Experts Used to Establish Ground Truth and Qualifications

• Not applicable for this type of submission. The "ground truth" for the performance characteristics listed in the table comes from verifiable engineering measurements on the device, often using calibrated equipment or phantoms, not human expert consensus. This is a technical performance evaluation, not a clinical diagnostic assessment.

4. Adjudication Method for the Test Set

• Not applicable. As noted above, the "test set" refers to technical performance characteristics measured through bench testing, not a dataset requiring human adjudication of clinical findings.

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

• No, an MRMC comparative effectiveness study was not done as described for AI assistance. This submission is for an Electrical Impedance Tomograph (EIT) device, which provides "information of local impedance variation" as an "adjunctive tool." It does not involve AI for interpretation or diagnosis, nor does it directly assist human readers in image interpretation in the way an AI-powered diagnostic tool would. It provides raw data/images for clinicians to interpret.

6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study

• Yes, in spirit, the bench testing results represent a "standalone" evaluation of the device's technical performance. The reported performance characteristics (SNR, Voltage Accuracy, Drift, etc.) are inherent properties of the device and its internal algorithms, independent of human interaction or interpretation beyond setting up the test. However, this is not an "algorithm-only" study in the typical sense of a diagnostic AI algorithm that produces a specific output (e.g., classifying a disease). The device itself is the "algorithm" and hardware working together to produce data.

7. Type of Ground Truth Used

• The ground truth for the "Performance Characteristics" in Table 5.1 is based on engineering measurements and calibrations using established methods for evaluating electronic medical devices. It is not expert consensus, pathology, or outcomes data. For example, SNR is measured against a known signal input, and position error might be measured using phantoms with known, precisely located impedance changes.

8. Sample Size for the Training Set

• Not applicable. The Enlight 2100 is an EIT device that provides direct physical measurements and derived images based on those measurements, not a machine learning model that requires a "training set" of data in the typical sense. Its "algorithm" is based on the physics of electrical impedance tomography.

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

• Not applicable, as there is no "training set" for an AI model. The device's operation is based on established physical principles and engineering design.

Ask a specific question about this device