Search Results

The Masimo SafetyNet Opioid System is intended to monitor and alarm when a patient may be experiencing an opioid induced impairment of oxygenation.

The Masimo SafetyNet Opioid System is indicated for the non-invasive continuous monitoring of individuals 15 years and older for the identification of when they may be experiencing a substance induced impairment of oxygenation (e.g., opioid induced respiratory depression (OIRD) caused by oral or injectable opioids) in hospital and home environments.

The Masimo SafetyNet Opioid System (device) is a system intended for noninvasively and continuously monitoring opioid users 15 years and older to identify substance induced impairment of oxygenation that can lead to opioid induced respiratory depression (OIRD) caused by oral or injectable opioids in hospital or home use environments. The device functions using a pre-determined notification escalation policy that provides the ability for the user to receive alerts and to establish a network of emergency contacts who can be contacted based on the level of alert/alarm as determined by the device. These emergency contacts are notified after previous notifications sent to the device user do not resolve the desaturation condition. As a final action, the device can notify Emergency Medical Service (EMS) dispatch to trigger a wellness call. If assistance is needed or the device user does not respond, the dispatch will request that EMS (i.e., ambulance service) be sent to the device user's location.

The device consists of the following components:

• Bedside Station Device component that communicates monitoring data wirelessly from the medical technologies to provide visual/audible alerts.
• Masimo Sensor Wireless wearable pulse oximetry sensor that provides the monitoring data.
• Masimo SafetyNet Opioid App - Software application installed on a smart device that provides the graphical user interface to display live monitoring data (e.g., pulse rate (PR), pulse waveform) and alarm condition status.
• Opioid Halo Software that runs continuously to provide real-time detection of the severe OIRD risk based upon changes or patterns in oxygenation biomarker data (peripheral oxygen saturation (SpO2), pulse rate (PR), perfusion index (Pi)) found to be consistent with published understanding of physiological effects of OIRD.
• Notification Escalation Policy - Policy that is used to add levels of awareness through the notification of the device user, emergency contact, or contracted Emergency Responders.
• Masimo SafetyNet Cloud A server accessed over the internet that gathers and stores measured data communicated wirelessly from a Bedside Station.

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

Acceptance Criteria and Reported Device Performance

The clinical performance validation summarized in the document focuses on the device's ability to detect opioid-induced respiratory depression (OIRD) with improved specificity and reduced non-actionable alarms compared to a traditional pulse oximeter with a fixed threshold alarm.

• Standard Pulse Oximeter (Benchmark): 80.6% Specificity
• Halo Level 1: 93.9% Specificity
• Halo Level 2: 97.5% Specificity
• Halo Level 3: 99.5% Specificity
  Subpopulation Specificity (Halo Level 2):
• Naïve: 99.3%
• Chronic: 98.4%
• Hospital: 97.2%
• Illicit: 93.5% |
  | Equivalent Sensitivity in OIRD Detection (compared to traditional oximeter) | Overall:
• Standard Pulse Oximeter (Benchmark): 99.2% Sensitivity
• Halo Level 1: 100.0% Sensitivity
• Halo Level 2: 100.0% Sensitivity
• Halo Level 3: 79.2% Sensitivity (Not equivalent for Level 3 overall, but with mitigating factors discussed in benefit/risk determination)
  Subpopulation Sensitivity (Halo Level 2):
• Naïve: 100.0%
• Chronic: 100.0%
• Hospital: 100.0%
• Illicit: 100.0% |
  | Reduction in Non-Actionable Alarms (compared to traditional oximeter) | Overall:
• Halo Level 1: 75.0% Reduction
• Halo Level 2: 89.0% Reduction
• Halo Level 3: 96.7% Reduction
  Subpopulation Alarm Reduction (Halo Level 2):
• Naïve: 96.3%
• Chronic: 93.0%
• Hospital: 88.0%
• Illicit: 61.0%
• Sleeping Non-opioid users: 61.6% |
  | Responsiveness of Halo alarms (Time from opioid injection to alarm - TFO, and Time from alarm to intervention - TTI) | Illicit Users:
• Benchmark: TFO 3.14 min, TTI 2.69 min
• Halo Level 1: TFO 3.42 min, TTI 2.40 min
• Halo Level 2: TFO 4.60 min, TTI 1.22 min
• Halo Level 3: TFO 4.80 min, TTI 1.06 min (Halo Level 2 and 3 show improved TTI despite slightly longer TFO) |
  | Performance in varied skin pigmentation | No clinically significant effect on performance due to use of Halo software and reliance on trends in data, even with a hypothesized 2% positive SpO2 bias in illicit users data. Specificity for illicit users with bias was 88.7% (L1), 95.1% (L2), 99.6% (L3) compared to 85.8% (L1), 93.5% (L2), 99.6% (L3) without bias. |
  | SpO2 Accuracy (hardware performance) | ARMS (%) for Radius PPG Sensors:
• 90-100% SpO2: 1.73%
• 80-90% SpO2: 1.80%
• 70-80% SpO2: 1.73%
• 70-100% SpO2: 1.75%
  Bias between Light and Dark subjects:
• Light (13 subjects): Bias 0.05, ARMS 1.79
• Dark (9 subjects): Bias 0.03, ARMS 1.75 |

Study Details

1. Sample Size and Data Provenance:

   • Test Set (Clinical Performance Validation):
     • Algorithmic Performance (OIRD Detection, Sensitivity/Specificity): 40,322 data segments from 641 study participants.
       • 135 prescription home opioid users (naïve, chronic)
       • 242 hospitalized opioid users
       • 264 illicit opioid users
     • Alarm Reduction: Data from 936 cases (the same 641 opioid users + an additional 295 sleeping non-opioid users).
     • Timing Data (TFO/TTI): 17 illicit opioid rescue cases.
     • Skin Pigmentation Analysis: Data from 264 illicit users (original and modified with 2% SpO2 bias).
     • SpO2 Performance Validation (Hardware): 26 healthy male and female volunteer subjects with varying skin pigmentation (13 Light, 9 Dark for analysis, 4 excluded).
   • Data Provenance: Not explicitly stated (e.g., country of origin, retrospective/prospective), but the context of an FDA de novo submission typically implies a controlled clinical study environment, likely prospective. The reference to "prescription home opioid (i.e., naïve, chronic)", "hospitalized opioid", and "illicit opioid" suggests a mix of settings.
   • Training Set Sample Size: Not explicitly stated in the provided text.

2. Number of Experts for Ground Truth & Qualifications:

   • The document implies clinical experts were involved in establishing the "ground truth" for OIRD events, particularly mentioning "critical data were available to lead to development of a sufficiently rigorous ground truth comparator" and "for in-hospital patients who may be experiencing other physiologic perturbations not related to OIRD, it is uncertain what other data was available for the clinical experts to lay ground truth of an OIRD event."
   • However, the number of experts and their specific qualifications (e.g., "radiologist with 10 years of experience") are not specified in the provided text.

3. Adjudication Method for the Test Set:

   • The document does not explicitly describe an adjudication method (e.g., 2+1, 3+1) for establishing the ground truth of OIRD events from the study data. It refers to "clinical experts" establishing the ground truth, but the process (e.g., individual expert, consensus, adjudicated by a third expert in case of disagreement) is not detailed.

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

   • No, an MRMC comparative effectiveness study was not explicitly mentioned or performed in the context of human readers improving with AI vs. without AI assistance. The study assessed the device's algorithmic performance (Halo software) in detecting OIRD and reducing alarms compared to a traditional pulse oximeter (benchmark), not human readers. The clinical validation focused on the device's ability to improve detection and reduce nuisance alarms.

5. Standalone Performance (Algorithm Only):

   • Yes, a standalone (algorithm only without human-in-the-loop performance) was done. The published sensitivity and specificity values for the Halo software (Level 1, 2, 3) directly represent the algorithm's performance in identifying OIRD events and non-OIRD events without human interpretation or intervention affecting the classification, only alarm generation. The comparison to a "traditional pulse oximeter with a fixed threshold alarm" also implies an algorithmic comparison.

6. Type of Ground Truth Used:

   • The ground truth for OIRD detection appears to be based on expert consensus/clinical assessment (implied by "critical data were available to lead to development of a sufficiently rigorous ground truth comparator" and "clinical experts to lay ground truth of an OIRD event"). It is not stated to be pathology or direct outcomes data, but rather clinical determination of OIRD based on available patient data segments.

7. Sample Size for the Training Set:

   • The document does not provide the sample size for the training set. It only states: "For devices using algorithms based on machine learning, the clinical validation must be completed using a dataset that is separate from the training dataset". This confirms the test set described (40,322 data segments from 641 participants) was distinct from the training data, but the size of the training data is not given.

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

   • The document does not specify how the ground truth for the training set was established. It only mentions the requirement that "the clinical validation must be completed using a dataset that is separate from the training dataset." Based on the description for the test set, it's highly probable the training set's ground truth was established similarly through clinical expert assessment, but this is not explicitly confirmed.

Ask a specific question about this device