(294 days)
The Navigator™ is indicated for the acquisition, processing, and display of hemodynamic parameters, in order to assist the clinician in achieving and maintaining a prescribed target hemodynamic stability.
Mean Systemic Filling Pressure (Pms) and Heart Efficiency (En) parameters are calculated by the Navigator from Cardiac Output (CO), Mean Arterial Pressure (MAP) and Right Atrial Pressure (RAP) data, obtained directly from patient monitoring equipment, and other data obtained by manual clinician entry. The Navigator system provides a visual indication of the patient's circulatory status in relation to predetermined goals.
The Navigator is a physiologically integrated system designed to assist clinicians in the management of the systemic circulatory state in the critically ill patient.
Here's an analysis of the provided text regarding the Navigator device's acceptance criteria and study, structured to include the requested information:
Acceptance Criteria and Device Performance Study for Navigator™
The Navigator™ device is a computer, diagnostic, pre-programmed, single-function system designed to assist clinicians in managing the systemic circulatory state in critically ill patients.
1. Table of Acceptance Criteria and Reported Device Performance
The primary efficacy endpoint for the Navigator™ device was its ability to demonstrate non-inferiority to conventional care with respect to maintaining hemodynamic stability.
| Acceptance Criteria | Reported Device Performance |
|---|---|
| Efficacy: Non-inferiority to conventional care regarding the average distance (mean standardized) to the central point of the target cardiovascular zone. | The study "demonstrated non-inferiority to Navigator over conventional care." There was "no statistically significant difference between the two treatment arms with regards to the mean standardized distance to the central point of the cardiovascular zone." |
| Safety: Incidence of adverse events (AEs) and serious adverse events (SAEs) comparable to conventional care, and no AEs/SAEs considered device-related. | The incidence of AEs and SAEs was comparable across the two study arms ("no trends in the type, frequency or severity of adverse events per study arm"). "No adverse or serious adverse events [were] considered to be related to Navigator™." Minor device incidents occurred but were not associated with clinical consequences or patient safety impact. |
2. Sample Size Used for the Test Set and the Data Provenance
- Test Set Sample Size: 112 patients were initially enrolled (Intent-To-Treat population). 107 patients completed the study as planned. The Modified Intent-To-Treat (MITT) population, defined as patients connected to Navigator™ for three hours or more of data, comprised 105 patients (57 in the Navigator™ arm and 48 in the control arm).
- Data Provenance: The study was a "multi-center, open, randomized, controlled human clinical trial." The document does not explicitly state the country of origin for the data, but the applicant (Applied Physiology Pty Ltd) is based in Sydney, NSW, Australia, suggesting the study likely took place in Australia, and possibly other international sites if it was a true "multi-center" global study. The study was prospective, as it involved enrolling patients and randomizing them to receive care guided by Navigator™ or conventional care.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and the Qualifications of Those Experts
The study was a comparative effectiveness trial aiming to show non-inferiority to conventional care, rather than validating the device's output against an independent "ground truth" established by experts. The "ground truth" for evaluating the device's effectiveness was based on the clinical outcome of maintaining patients within a target cardiovascular zone by clinicians using the device or conventional care.
Therefore, there were no specific experts explicitly mentioned as establishing a "ground truth" for the test set in the way one might with diagnostic accuracy studies. The "ground truth" was essentially the patient's physiological state and clinical course as monitored and managed by the attending clinicians in both the Navigator and control arms, with the primary endpoint being a physiological measure (distance to the central point of the target cardiovascular zone).
4. Adjudication Method for the Test Set
The study design describes a randomized controlled trial where patients were randomized to either receive care guided by Navigator™ or conventional care. The primary endpoint "the average distance (mean standardized) to the central point of the target cardiovascular zone" was a quantitative measure.
There is no specific adjudication method (e.g., 2+1, 3+1) described for establishing a "ground truth" of individual cases by expert review. Given the nature of the primary endpoint (a statistical comparison of a numerical physiological outcome), such an adjudication method would not be directly applicable. The data collection would have been direct physiological measurements from monitoring equipment.
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 was not a Multi-Reader Multi-Case (MRMC) comparative effectiveness study in the typical sense of evaluating human reader performance on diagnostic images. Instead, it was a randomized controlled clinical trial comparing the impact of a computer-assisted decision support system (Navigator™) versus conventional care on patient management and physiological outcomes.
Therefore, the concept of "human readers improve with AI vs without AI assistance" as a direct effect size (e.g., AUC improvement) is not applicable to this study design. The study demonstrated that care guided by Navigator™ was non-inferior to conventional care in achieving and maintaining hemodynamic stability. The benefit is implied as enabling clinicians to achieve comparable outcomes with the aid of the system.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done
No, a standalone (algorithm-only) performance study was not described or performed. The Navigator™ is explicitly described as a system "designed to assist clinicians" and its indications for use state it is "to assist the clinician." The clinical trial evaluated the device with a human-in-the-loop (i.e., care guided by Navigator™). The device calculates parameters and provides visual indications, which clinicians then use for management.
7. The Type of Ground Truth Used
The "ground truth" used for efficacy was the patient's physiological state and clinical course, specifically measured by the "average distance (mean standardized) to the central point of the target cardiovascular zone." This is an outcomes data / physiological measurement type of ground truth, rather than expert consensus on a diagnostic finding or pathology.
For safety, the "ground truth" was established through the monitoring and reporting of adverse events by study personnel and clinicians, with subsequent assessment of relatedness to the device or study procedures.
8. The Sample Size for the Training Set
The document does not provide information on a training set sample size. This device is a pre-programmed diagnostic computer mentioned as having "Software Verification and Validation testing" rather than a machine learning/AI model that typically requires a distinct "training set." While it's a "single-function pre-programmed diagnostic computer," the "computer-assisted decision support" implies pre-defined physiological models and algorithms, not necessarily a data-driven machine learning model requiring a large training dataset as understood in modern AI.
9. How the Ground Truth for the Training Set Was Established
Since no training set and associated machine learning algorithm development are described, there is no information provided on how ground truth for a training set was established. The device relies on "physiologically integrated systems" and algorithms that calculate parameters from direct patient monitoring data and clinician input, based on established physiological principles.
§ 870.1435 Single-function, preprogrammed diagnostic computer.
(a)
Identification. A single-function, preprogrammed diagnostic computer is a hard-wired computer that calculates a specific physiological or blood-flow parameter based on information obtained from one or more electrodes, transducers, or measuring devices.(b)
Classification. Class II (performance standards).