Search Results

The Hemolung Respiratory Assist System is indicated for respiratory support that provides extracorporeal carbon dioxide (CO2) removal from the patient's blood for up to 5 days in adults with acute, reversible respiratory failure for whom ventilation of CO2 cannot be adequately or safely achieved using other available treatment options and continued clinical deterioration is expected.

The Hemolung RAS provides low blood flow, veno-venous extracorporeal carbon dioxide removal (ECCO2R) using a single, 15.5 French dual lumen catheter inserted percutaneously in the femoral or jugular vein. The Hemolung RAS is not intended to provide therapeutic levels of oxygenation. During Hemolung therapy, blood passing through the circuit is oxygenated via room air sweep gas; however, at ultra-low extracorporeal blood flows, the limited oxygen carrying capacity of blood precludes meaningful oxygenation of mixed venous blood. The Hemolung RAS consists of three main components: C. Hemolung Controller, A. Hemolung Cartridge, and B. Hemolung Catheter.

The acceptance criteria for the Hemolung Respiratory Assist System are not explicitly listed in a single table or section, but rather are distributed across sections describing Nonclinical/Bench Studies, Performance Testing - Animal, and Summary of Clinical Information, and further consolidated in the RISKS TO HEALTH and SPECIAL CONTROLS sections. The "device performance" is therefore interpreted as the results from the various tests conducted.

Here is a summary of the acceptance criteria and the device's reported performance based on the provided text:

Hemolung Respiratory Assist System Acceptance Criteria and Reported Performance

2. De-escalation of mechanical ventilatory support while preventing respiratory acidosis.
   Demonstrate clinically beneficial response (correction of respiratory acidosis).
   Avoidance of intubation/IMV in NIV patients. | Study: Analysis of data from 234 patients from prospective clinical trials, real-world use, and the Hemolung Post-market Registry. Patients included if they met 4 criteria: original patient-level efficacy data, pre-Hemolung pH/PaCO2, data for at least one additional time point during first 35 hours, received therapy for at least 6 hours. Data stratified by baseline pH (=7.35 for non-acidotic). Acute timepoint (first day) analyzed.
   Performance:

• Demonstrated clinically and statistically significant correction of respiratory acidosis and de-escalation of mechanical ventilatory support.
• 91% of Hemolung patients showed a clinically beneficial response (correction in respiratory acidosis after one day of therapy).
• 92% of non-invasively ventilated (NIV) patients avoided intubation and invasive mechanical ventilation (IMV) with Hemolung therapy.
• For acidotic patients failing NIV (n=43), Hemolung corrected respiratory acidosis (pH =7.24 -> pH=7.36). |
  | Safety | Low incidence of therapy-related, procedural, operational, and component-related complications. Most complications should not result in patient harm, rarely permanent harm or death. Demonstrated characterization and mitigation of risks (bleeding, thrombocytopenia, hemolysis, thrombosis, infection, adverse tissue reaction, mechanical failure, hemodynamic instability, hypothermia, mechanical injury, inadequate gas exchange, hemodilution, gas embolism). | Study: Analysis of clinical complication data from 1,034 patients receiving Hemolung therapy (prospectively monitored AE data, US Expanded Access/EUA, OUS post-market Registry, OUS post-market surveillance).
  Performance:
• Therapy-Related Complications (172 total in 121 patients): Bleeding (7.0%), Hemolysis (2.6%), Thrombocytopenia (2.1%), Thrombosis/coagulation disorder (1.3%), Hemodynamic instability (1.6%), Other (2.1%). 66 had no patient impact, 96 required medical intervention, 10 resulted in death. No unanticipated adverse device events.
• Procedural, Operational, and Component-Related Complications (128 total):
  • Procedural: 62 (6.0%); 7 required intervention, 4 resulted in death, 51 no impact.
  • Operational: 33 (3.2%); 0 required intervention or death, 33 no impact.
  • Component-related: 33 (3.2%); 1 required intervention, 0 resulted in death, 32 no impact.
• Overall, incidences were low, most did not result in patient harm, and rarely permanent harm or death. Risks are characterized and mitigated via non-clinical testing, animal studies, and special controls. |

Additional Information on Studies:

1. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)

   • Clinical Effectiveness (Test Set): 234 patients. Data provenance included prospective clinical trials, real-world use, and the Hemolung Post-market Registry. It is implied to be a mix of retrospective and prospective data, but predominantly based on existing patient-level data. The text doesn't specify country of origin for the 234 patients, but mentions "OUS post-market Hemolung Registry" and "OUS post-market surveillance", suggesting international data.
   • Clinical Safety (Test Set): 1,034 patients. Data derived from monitored and independently adjudicated adverse event data from prospective clinical trials, US Expanded Access or Emergency Use Authorization data collection and surveillance, the OUS post-market Hemolung Registry, and OUS post-market surveillance. This is a mix of prospective and retrospective (post-market surveillance) data, with both US and OUS (outside US) origins.

2. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)

   • For the clinical safety data (1,034 patients), adverse event data from prospective clinical trials were "monitored and independently adjudicated." The number and qualifications of these adjudicators (experts) are not specified.
   • For clinical effectiveness data, the "ground truth" was established by measured clinical outcomes (pH, PaCO2, ventilatory support status). This does not explicitly involve human experts creating a ground truth label in the same way as, for example, image interpretation. The interpretation of these physiological parameters is standard medical practice carried out by treating physicians.

3. Adjudication method (e.g. 2+1, 3+1, none) for the test set

   • For clinical safety, adverse events from prospective clinical trials were "independently adjudicated." The specific method (e.g., 2+1, 3+1) is not provided. For other data sources (real-world use, registries, surveillance), the text does not mention formal adjudication of events by a panel, but rather refers to data collection from treating physicians and surveillance procedures.

4. 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 MRMC comparative effectiveness study was done. The device is not an AI-assisted diagnostic tool for human readers; it is a medical device providing extracorporeal CO2 removal. Therefore, questions regarding "human readers improve with AI vs without AI assistance" are not applicable.

5. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

   • The Hemolung Respiratory Assist System is a device with hardware, software, and disposables that directly interacts with the patient's blood. Its performance is inherently "standalone" in the sense that its primary function (CO2 removal) is performed by the device itself, not by an algorithm providing information for a human to act on. The "clinical performance" study (234 patients) and "animal study" (8 calves) assess the device's standalone performance in a clinical or simulated clinical setting. The software within the device has undergone validation and verification.

6. The type of ground truth used (expert consensus, pathology, outcomes data, etc)

   • Animal Study: Ground truth was established through direct physiological measurements (blood chemistry, hematology, etc.), veterinary observations, clinical scoring (abnormality grading), and detailed necropsy with histopathology. This is a combination of objective measurements and expert veterinary evaluation of animal health status and pathology.
   • Clinical Effectiveness: Ground truth for effectiveness was based on physiological outcomes data, specifically pH and PaCO2 measurements, and the need for or de-escalation of mechanical ventilatory support. These are objective clinical parameters.
   • Clinical Safety: Ground truth for safety was based on reported and, in some cases, adjudicated adverse events documented during clinical trials, real-world use, and post-market surveillance. This involves a combination of physician diagnoses, clinical observations, and in some cases, independent adjudication.

7. The sample size for the training set

   • The document explicitly describes the studies conducted to support the de novo request (bench, animal, and clinical performance/safety). It does not mention a "training set" in the context of machine learning. The studies listed are for validation of the device's safety and performance. The device itself is not described as an AI/ML product developed with training data but rather a direct therapeutic device.

8. How the ground truth for the training set was established

   • Not applicable, as no machine learning training set is mentioned for the device itself.

Ask a specific question about this device