Search Results

The Pulse Oximeter is a non-invasive device indicated for use in measuring, displaying, storing and transmitting functional oxygen saturation of arterial hemoglobin (SpO2) and pulse rate for adult patients. It is intended for spot- check and/or continuous data collection, and not continuous monitoring. It can be used in sleep labs, long-term care, hospitals and home use.

The Pulse Oximeter is a lightweight, portable health finger ring oximeter for home or healthcare facilities. SpO2 measurement technology is based on the developed photoelectron method and circuit design, and Shenzhen Viatom Technology Co., Ltd. developed calculation software. The SpO2 sensor receives the optical signal from the red light and infrared light through the finger into the oximeter. Two emitting tubes (red light diodes and infrared diodes) are located on the inner right side of the sensor, and they can emit red light and infrared. The receiving end is located on the inner left side of the sensor, and it can receive the red light and infrared through the finger. The MCU receives the pulse signal, gets the frequency signal by counting, processes its digital signal, and finally gets the measured SpO2 value. The PR is calculated on average by the above peak intervals of the PR waveform.

The provided text is a 510(k) summary for a Pulse Oximeter (Models PO2, PO2A, PO2B) and does not contain information about an AI/ML-driven device. Therefore, it is not possible to describe acceptance criteria or a study related to an AI/ML device based on this document.

The document primarily focuses on demonstrating the substantial equivalence of the subject pulse oximeter to a predicate pulse oximeter (K191088 Checkme O2 Pulse Oximeter) by meeting established performance standards for pulse oximeters, such as ISO 80601-2-61.

Here's an analysis based on the information available in the document, demonstrating why it doesn't fit the AI/ML framework you've described:

General Device Performance (Pulse Oximeter):

1. A table of acceptance criteria and the reported device performance:
   The document does not present a formal "acceptance criteria table" in the context of an AI/ML model. Instead, it provides a comparison table of the subject device's specifications and performance metrics against a predicate device.

   Characteristic	Acceptance Criteria (Implicit from Predicate & Standards)	Reported Device Performance (Subject Device)
   SpO2 Accuracy (70-100%)	±2%	1.77% ARMS
   Pulse Rate Accuracy	±2bpm or ±2% (whichever is greater)	±2bpm or ±2% (whichever is greater)
   SpO2 Measurement Accuracy (ARMS)	≤ 2% (from ISO 80601-2-61)	1.77%
   Work Mode	Spot-check and continuous data collection (not continuous monitoring)	Spot-check and continuous data collection (not continuous monitoring)
   Intended Application Site	Finger	Finger

2. Sample sized used for the test set and the data provenance:

   • Test Set Description: The "clinical validation testing of the SpO2 performance" was conducted on "healthy adult volunteers."
   • Sample Size: The exact number of healthy adult volunteers is not specified in the document.
   • Data Provenance: The document states "Measured values are from a controlled lab study in healthy volunteers." It does not specify the country of origin. The study appears to be prospective, as it involved actively testing subjects with the device.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
   This question is not applicable to this device. For a pulse oximeter, "ground truth" for SpO2 and pulse rate is established using a co-oximeter and ECG (or similar reference standard) on induced hypoxia studies, as per ISO 80601-2-61. There are no "experts" in the AI/ML sense establishing ground truth labels for images or other complex data. The "ground truth" clinical values are collected by qualified personnel trained in conducting such studies.

4. Adjudication method (e.g. 2+1, 3+1, none) for the test set:
   This question is not applicable. Adjudication methods like 2+1 or 3+1 are used in studies involving human interpretation (e.g., radiology reads) to resolve discrepancies and establish a consensus ground truth. For a pulse oximeter, the reference measurements from the co-oximeter are the ground truth, and human adjudication is not part of the process.

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:
   This question is not applicable. The device is not an AI-assisted diagnostic tool. No MRMC study was performed or is relevant for this type of device.

6. If a standalone (i.e. algorithm only without human-in-the loop performance) was done:
   This question is not applicable in the context of an AI/ML algorithm. The performance of the pulse oximeter itself (which contains an algorithm to calculate SpO2 and PR from optical signals) was evaluated in a standalone manner against reference standards in the clinical study. The device's performance is inherently "standalone" in its measurement function, as it provides direct numerical outputs.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
   For SpO2 accuracy, the ground truth was arterial oxygen saturation measured by a co-oximeter (from blood samples) during induced hypoxia, as per ISO 80601-2-61. For pulse rate, the ground truth would typically be derived from an electrocardiogram (ECG) monitor. This is a physiological reference standard, not expert consensus or pathology.

8. The sample size for the training set:
   This question is not applicable. The document describes a traditional medical device (pulse oximeter) that uses a pre-defined algorithm based on physical principles (absorption of red and infrared light). It does not mention any "training set" for an AI/ML model. The "algorithm" for SpO2 calculation is based on principles of spectroscopy and physiological models, not machine learning from a dataset in the way a deep learning model would be trained.

9. How the ground truth for the training set was established:
   This question is not applicable for the same reasons as #8. There is no training set for an AI/ML model described.

In summary, the provided document details the regulatory clearance for a traditional medical device (pulse oximeter) based on established performance standards, and therefore, the questions tailored for AI/ML device evaluations are generally not relevant.

Ask a specific question about this device