Search Results

Precision Flow® HVNI is intended for use to add warm moisture to breathing gases from an external source for administration to a neonate/infant, pediatric and adult patients in the hospital and subacute institutions settings. It adds heat and moisture to a blended medical air/ oxygen mixture and assures the integrity of the precise air/oxygen mixture via an integral oxygen analyzer. The flow rates may be from 1 to 40 liters per minute via nasal cannula.

Precision Flow® HVNI provides high velocity nasal insufflation (HVNI) with simultaneous oxygen delivery to augment breathing of spontaneously breathing patients suffering from respiratory distress and/or hypoxemia in the hospital setting. Precision Flow® HVNI is not intended to provide total ventilatory requirements of the patient and not for use during field transport.

Device Description

The Precision Flow® HVNI system delivers high flow rates of heated and humidified blended breathing gas through high flow nasal cannulas. The Precision Flow® HVNI system can connect to air and oxygen source. The Precision Flow® HVNI may be operated with limited performance at gas inlet pressures as low as 4 psi (28 kPa). For the full specified range of gas flows and oxygen percentages, both gas inlet pressures must be at minimum 40 psi (276 kPa).The main unit contains an integrated blender that delivers the targeted gas mixture to the disposable patient circuit (DPC). The Disposable Patient Circuit contains:

Water path: tubing from sterile water supply to vapor transfer cartridge
. Vapor transfer (humidification) cartridge: 2 types; low flow (1-8 lpm) and high flow (5-40 lpm)
Delivery Tube: triple lumen tubing
. Nasal Cannula

The device automatically senses cartridge type. The available set temperature range is 33 °C to 39 °C. The device also contains a backup battery to provide power only for 15 minutes.

AI/ML Overview

Here's a breakdown of the acceptance criteria and the study information for the Precision Flow® HVNI, based on the provided text:

Acceptance Criteria and Device Performance

The acceptance criteria for the Precision Flow® HVNI are primarily non-clinical, focusing on safety, performance standards, software validation, and labeling requirements. The clinical studies establish safety and effectiveness by demonstrating non-inferiority or comparable outcomes to existing treatments.

Acceptance Criteria Category	Acceptance Criteria	Reported Device Performance/Evidence
Biocompatibility/Materials	Patient-contacting components must be demonstrated to be biocompatible (Special Control 1). Tests include Cytotoxicity, Sensitization, Intracutaneous Reactivity, Extractables and Leachables, Volatile Organic Compounds (VOC), and Particulate Matter (PM2.5).	All biocompatibility tests passed, demonstrating the biocompatibility of the device, in accordance with ISO 10993-1.
Shelf Life/Reprocessing/Sterility	Cleaning instructions for reusable components must be validated (Special Control 3). Acceptance criteria for cleaning validation: no visible soil, protein level <6.4 micrograms/cm², TOC <12 micrograms/cm². Low-level disinfection validated by ≥ 6 log reduction of specified test organisms.	Cleaning validation study met acceptance criteria: no visible soil, protein level <6.4 micrograms/cm², and TOC <12 micrograms/cm². Low-level disinfection demonstrated ≥ 6 log reduction of E. coli, S. aureus, K. pneumoniae, and P. aeruginosa.
EMC/Electrical/Mechanical/Thermal Safety	Electrical safety, thermal safety, mechanical safety, electromagnetic compatibility, and radiofrequency identification (RFID) testing must be performed (Special Control 4). Standards: IEC 60601-1-2:2014, AAMI / ANSI ES60601-1:2005/(R) 2012 and A1:2012, AIM Standard 7351731 (RFID reference).	Device passed tests in accordance with IEC 60601-1-2:2014, AAMI / ANSI ES60601-1:2005/(R) 2012 and A1:2012. Tested for RFID exposure using AIM Standard 7351731 as a reference.
Software	Software verification, validation, and hazard analysis must be performed (Special Control 5). Adequate software documentation consistent with a "Major" level of software concern (FDA Guidance, May 11, 2005).	Adequate software documentation provided. Software validation and verification testing demonstrated that the device met its design, implementation, and cybersecurity requirements.
Non-clinical Performance Testing	Device performs as intended under anticipated conditions for use (Special Control 2). Specific tests: a. Alarm testing b. Continuous use thermal stability testing c. Humidity output testing d. Blender performance (FiO2 blending accuracy) Additional tests: Battery performance, Cannula pressure drop (blocked tube alarm not initiated). Standards: AAMI / ANSI / IEC 60601-1-8:2006 & A1:2012, ISO 8185:2007, ISO 80601-2-74.	Blender performance: Verified set FiO2 at various flowrates (21%-100% FiO2) with high/low flow cartridges. Thermal Stability: Output temperature remained within specifications for 15 days across full range of settings, and another 15 days at max temp (43°C) and max flow (8/40 LPM). Battery performance: No change in set flow and FiO2 observed. Nasal cannula performance: Acceptable pressure drop at applicable flows, blocked tube alarm not initiated. Alarm testing, continuous use thermal stability, humidity output, and blender performance were implied to be successful as per "Additional testing".
Clinical Effectiveness (Adults)	Non-inferiority to NIPPV in avoidance of failure (including intubation/mechanical ventilation) within 72 hours of initiation (Primary Endpoints for Doshi et al. study).	Doshi et al. study: Intubation within first 72 hours: 7% (HVNI) vs. 13% (NIPPV). Demonstrated non-inferiority (Wald p<0.001) for intubation rates. While arm failures (crossover) were higher in HVNI (26% vs 17%), the intubation rate was lower, suggesting clinical usefulness. Secondary findings: HVNI favored for pCO2 normalization at 240 min (p=0.02), physician perception (respiratory response, comfort/tolerance, simplicity of use), and patient perception of dyspnea was similar.
Clinical Effectiveness (Neonates)	Safety and effectiveness comparable to CPAP/nCPAP/NIPPV for respiratory distress/hypoxemia, without increasing adverse outcomes (pneumothorax, infection, ROP surgery, extubation failure, BPD, air leaks, IVH, NEC, sepsis). Non-inferiority for intubation rates within 72 hours of therapy initiation (Lavizzari et al. study).	McQueen et al.: Compared to VON database (CPAP predominant), HFT did not increase pneumothorax (4.8% vs 4.4%) or nosocomial infection (14.8% vs 15.1%). Lower ROP surgery (2.2% vs 3.2%). Supported effectiveness and safety. Collins et al.: Non-inferior extubation failure rates for HFT (22%) vs nCPAP (34%). Similar clinical outcomes (BPD), trend for shorter supplemental oxygen in HFT. Improved nasal trauma score with HFT. Lavizzari et al.: Non-inferiority for primary endpoint (mechanical ventilation within 72 hrs). Failure rates: HFT (10.8%) vs nCPAP/BiPAP (9.5%). Lower bound of 95% CI (-6.0%) below zero met non-inferiority criteria. Similar secondary outcomes (duration of support, surfactant, air leaks, BPD). Kugelman et al.: Comparable failure to intubate rates for HFT (34.2%) vs NIPPV (31.6%). No clinically meaningful differences in secondary outcomes (air leak, nasal trauma, BPD, IVH, NEC, sepsis).
Labeling	Labeling must meet 21 CFR 801.109, include specific information (FiO2 ranges, flowrates, continuous monitoring warning, condensation warning, alarm description/function, not a CPAP device warning, aspiration risk warning) (Special Control 6).	Labeling meets 21 CFR 801.109 and includes all specified warnings and descriptions as per Special Control 6.
Adverse Events	No device-related serious adverse events.	In the Doshi et al. adult clinical study, there were no adverse events related to the use of the device. All adverse events were related to the underlying patient condition. The neonate studies also did not raise new safety concerns.

Study Information

Due to the nature of this particular device (high flow humidified oxygen delivery device) and the provided text, the "studies" are clinical trials and published literature, rather than an AI/ML-specific validation study with test sets, ground truth experts, or training sets in the typical sense. The regulatory submission leverages existing clinical evidence.

For the Adult Clinical Study (Doshi et al.):

Sample size used for the test set and the data provenance:
- Sample Size: 204 adult patients (104 in HVNI arm, 100 in NIPPV arm).
- Data Provenance: Multi-center, prospective, randomized clinical trial conducted at 5 sites. Country of origin not explicitly stated but implies US context for FDA submission.
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- This is a clinical trial where clinical outcomes (e.g., intubation, failure to tolerate device, physiological parameters like O2 saturation, PaO2, PCO2, relief of respiratory distress, deteriorating medical status) were the ground truth. These outcomes were assessed by treating clinicians (physicians, nurses, respiratory therapists) at the study sites. The "experts" are the medical professionals directly involved in patient care and decision-making within the clinical trial setting. Specific qualifications beyond "physician" are not detailed in the text.
Adjudication method for the test set:
- Not explicitly stated as a formal adjudication panel in the typical AI/ML context. Treatment failure was defined by clear clinical criteria (need for intubation, device intolerance, failure to oxygenate/ventilate, failure to relieve distress, deteriorating medical status). These assessments are inherent to clinical practice and were likely made by the treating clinicians in real-time, within the trial's protocol.
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:
- No, this was not an MRMC study comparing human readers with and without AI assistance. It was a comparative effectiveness study between two medical devices/therapies (HVNI vs. NIPPV).
If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:
- No, this device is a medical device for patient breathing support; it is not an AI algorithm performing a diagnostic task. Its performance is measured by its clinical impact on patients.
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- Outcomes Data: Primary ground truth was clinical outcomes data, specifically "treatment failure rate, defined as the need for intubation, and arm failure rate, defined as the decision for crossover to the alternate therapy, within 72 hours of initiation of assigned therapy." Other outcomes included vital signs, blood gas analysis (e.g., pCO2), physician perception, and adverse events.
The sample size for the training set:
- Not applicable in the context of this device being a medical hardware system, not an AI/ML algorithm requiring a specific training set of patient data.
How the ground truth for the training set was established:
- Not applicable for the same reason as above.

For the Neonate Patient Population (McQueen et al., Collins et al., Lavizzari et al., Kugelman et al.):

Sample size used for the test set and the data provenance:
- McQueen et al.: Retrospective cohort study. HFT group: Data from five centers, no specific n given for test set, but stated "used Vapotherm HVNI as their primary and predominant mode." Compared to Vermont Oxford Network (VON) database: 176,599 very low birth weight baby admissions (control/comparison data).
- Collins et al.: 132 low birth weight infants (< 2500g, <32 weeks gestation). Randomized (67 HFT, 65 nCPAP). Data provenance: Randomized Controlled Trial.
- Lavizzari et al.: Randomized trial, 298 infants (158 HFT, 158 nCPAP/BiPAP). Data provenance: Prospective, randomized clinical non-inferiority trial.
- Kugelman et al.: 76 infants (<35 week gestational age, >1000g). Randomized (38 HFT, 38 NIPPV). Data provenance: Randomized Pilot Study.
- Data Provenance for all neonate studies: Published literature, likely international based on author names and journal types, but not explicitly stated for each study.
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- Similar to the adult study, ground truth involved clinical outcomes assessed by treating medical professionals (neonatologists, nurses, etc.) within the context of their respective clinical trials or retrospective data collection. Specific expert qualifications beyond clinical roles are not detailed within the provided text.
Adjudication method for the test set:
- Not explicitly stated as a formal adjudication panel. Outcomes were generally defined by protocol-driven clinical events (e.g., extubation failure, need for intubation, physiological parameters).
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:
- No. These are clinical studies comparing therapies, not AI assistance for human readers.
If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:
- No.
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- Outcomes Data: Primary ground truths included extubation failure rates, rates of intubation/mechanical ventilation, occurrences of pneumothorax, nosocomial infection, ROP surgery, bronchopulmonary dysplasia (BPD), air leaks, IVH, NEC, sepsis, duration of respiratory support, need for surfactant, time to full feeds, length of stay, mortality, and nasal trauma scores.
The sample size for the training set:
- Not applicable.
How the ground truth for the training set was established:
- Not applicable.

Ask a Question

Ask a specific question about this device

Page 1 of 1