Search Results

The VitalFlow Console controls the speed of the VitalFlow Centrifugal blood pump during extracorporeal cardiopulmonary life support for adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent. The VitalFlow Centrifugal pump is driven by the VitalFlow Motor Drive or the VitalFlow Emergency Handcrank.

The VitalFlow Console provides control of blood pumping through an extracorporeal circuit during extracorporeal membrane oxygenation (ECMO) procedures. The console powers the VitalFlow motor drive unit which provides rotation of the VitalFlow Centrifugal pump. Pump motor speed (RPM) can be adjusted by the user and flow and bubble detection is provided by an ultrasonic flow probe and displayed on the touchscreen. The touchscreen display allows users to set alarm limits for all measured parameters. The device will alarm visually and audibly when limits are exceeded. Status indicators, power/battery life and secondary RPM indicator are provided. Data download and data streaming from the console is available for ECMO circuit data only; no patient data is stored for output.

The provided FDA 510(k) clearance letter for the VitalFlow Console (K250199) generally describes the device and its indications for use, and makes a case for substantial equivalence to a predicate device. However, it does not provide specific details about acceptance criteria or the study that proves the device meets those criteria, particularly regarding AI or algorithm performance.

The core of this submission is about a software upgrade to an existing hardware device (VitalFlow Console) which enables software upgrades via USB. The FDA clearance is based on demonstrating that this change does not raise new questions of safety or effectiveness and maintains the existing performance characteristics. Therefore, the information typically requested in your prompt (e.g., sample size for test sets, number of experts for ground truth, MRMC studies, AI effect size, etc.) would primarily be relevant if the upgrade introduced a new AI/algorithmic component or significantly altered a measurement or diagnostic function that required such validation.

Based on the document, here's what can be extracted and what cannot:

Analysis of the Provided Document Regarding Acceptance Criteria and Performance Study:

The document explicitly states:
"Software verification (which included Cybersecurity) was used to verify the performance characteristics of the subject device with the upgrade over USB change."
And:
"Substantial equivalence of the performance characteristics is demonstrated through regression testing. The VitalFlow Console (Upgrade Over USB) continues to meet international standards for safety and has demonstrated effectiveness at maintaining the device performance."

This indicates that the performance study focused on regression testing to ensure the new software upgrade feature (upgrade over USB) did not negatively impact the existing performance of the device or introduce new risks, especially related to cybersecurity. It was not a de novo study to establish new performance metrics for an AI or algorithmic component, as the device's fundamental function (controlling a blood pump) remains unchanged, and the cleared modification is about the method of software updating.

Therefore, most of the specific questions about AI/algorithm performance studies (e.g., number of experts, ground truth type, MRMC study) are not applicable to the scope of this particular 510(k) submission.

Extracted Information:

1. A table of acceptance criteria and the reported device performance:

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

• Sample Size: Not explicitly stated. The testing mentioned is "Software verification" and "regression testing." For software changes, the "test set" would typically refer to the test cases used in the verification and validation (V&V) activities. The document does not provide the number of test cases run or the duration/number of runs for regression testing.
• Data Provenance: Not explicitly stated. Given that it's regression testing for a software update on an existing device, it would likely involve internal testing data and potentially real-world data from the existing predicate device's operational environment. It's implicitly a combination of retrospective (based on existing device performance) and prospective (new tests for the updated software) testing, but the document doesn't specify. Country of origin not mentioned.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:

• Not Applicable / Not Stated. This type of information is typically required for studies establishing diagnostic accuracy or clinical effectiveness, often involving human readers/interpreters. This submission focuses on a software update for a control console, where "ground truth" would relate to the correct functioning of the software and hardware rather than clinical interpretation by experts. "Software verification" and "regression testing" would involve engineering and software quality assurance expertise to establish whether the device performed as specified.

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

• Not Applicable / Not Stated. Adjudication methods are relevant for subjective interpretations (e.g., medical images). For software verification and regression testing, failures are objective (e.g., a test case passes or fails, a function behaves as expected or not).

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:

• No. An MRMC study was not conducted as this submission is not about a diagnostic AI/CAD device or a device that assists human interpretation in a new way. It's about a software update for a control console.

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

• Not Applicable / Not Stated in this context. The "algorithm" here is the control software for the blood pump. Its "standalone" performance would be measured by its ability to precisely control the pump's RPM, detect flow, and manage alarms, which is what "software verification" and "regression testing" would assess. The document confirms these existing performance characteristics are maintained.

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

• Not Applicable in the traditional sense. For software verification and regression testing, the "ground truth" is the established functional and performance requirements of the device. This is confirmed through testing against design specifications, system requirements, and pre-existing performance benchmarks of the predicate device.

8. The sample size for the training set:

• Not Applicable. This is not an AI/Machine Learning device where a "training set" is relevant. The "software update" refers to a traditional software engineering change, not an AI model retraining.

9. How the ground truth for the training set was established:

• Not Applicable. As above, no training set for an AI model.

Summary of Device and Changes:

• Device: VitalFlow Console, controls a centrifugal blood pump for extracorporeal cardiopulmonary life support.
• Predicate: VitalFlow Console (K230364).
• Change in K250199: Enables software upgrades to be uploaded through the USB port (previously required Medtronic service personnel to open the console and connect laptops to individual boards).
• Regulatory Focus: Ensuring the new USB update feature does not compromise the safety or effectiveness of the device, particularly regarding cybersecurity and maintaining existing performance.
• Performance Study Type: Software verification and regression testing, indicating an engineering-focused validation rather than a clinical outcome or AI performance study.

Ask a specific question about this device

The CentriMag Extracorporeal Blood Pumping System is a non-roller-type cardiopulmonary and circulatory bypass blood pump used to pump a patient's blood through an extracorporeal circuit for periods lasting less than 6 hours for the purpose of providing either: i. Full or partial cardiopulmonary bypass (i.e., circuit includes an oxygenator) during open surgical procedures on the heart or great vessels: or ii. Temporary circulatory bypass for diversion of flow around a planned disruption of the circulatory pathway necessary for open surgical procedures on the aorta or vena cava The CentriMag™ Blood Pump for use with the CentriMag™ Acute Circulatory Support System (Motor, Monitor, Console, and Flow Probes) is indicated for controlling blood flow as part of an extracorporeal membrane oxygenation (ECMO) circuit. ECMO is intended to provide assisted extracorporeal circulation and physiologic gas exchange of the patients' blood for adult patients with acute respiratory failure and/or acute cardiopulmonary failure. where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent.

The CentriMag™ Acute Circulatory Support System, hereafter referred to as the CentriMag System, was designed to provide temporary mechanical circulatory support. To date, the CentriMag system in the United States (US) is indicated as a component of an extracorporeal bypass circulatory support circuit for use during cardiopulmonary bypass (CPB) surgery (up to 6 hours), it is also indicated for controlling blood flow as part of an extracorporeal membrane oxygenation (ECMO) circuit. The CentriMag System provides circulatory assistance for patients in acute hemodynamic compromise, a population whose treatment options are limited. The CentriMag System is composed of: - CentriMag Primary Console - CentriMag Motor - CentriMag Blood Pump - CentriMag Flow Probe - · Mag Monitor The CentriMag Motor is a reusable, non-sterile component of the CentriMag System. The CentriMag Motor holds the blood pump and drives the impeller inside the blood pump. The motor turns the magnet (and impeller) within the blood pump (Full MagLev™ technology) at a speed that is set on the console by the user. When the impeller is rotated, a pressure gradient develops between the center and outside edge of the pump, causing blood to flow from the inflow to the outflow port of the pump. The amount of flow through the pump depends on the speed of the impeller, and the difference between the inlet and outlet pressures.

The provided documents describe a 510(k) submission for the Abbott CentriMag™ Acute Circulatory Support System. This submission is a "Special 510(k)," which means it addresses a modification to an already cleared device. The primary change is an update to the CentriMag Motor, specifically a new "snap-in motor locking feature" replacing a "screw locking feature."

Therefore, the acceptance criteria and performance study described are focused on demonstrating that this change does not negatively impact the safety and effectiveness of the device and that the new design is substantially equivalent to the predicate device.

Here's the breakdown of the requested information based on the provided text:

1. A table of acceptance criteria and the reported device performance

Since this is a Special 510(k) for a design change (motor locking mechanism), the "performance" is about demonstrating that the new design is as safe and effective as the previous one, and that it conforms to requirements for its intended use. The document doesn't list specific numerical acceptance criteria (e.g., a "failure rate must be less than X%"). Instead, it describes categories of testing performed to ensure the new design performs adequately.

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

The document does not specify the sample sizes for any of the verification or validation tests (e.g., number of motors subjected to durability testing, number of users in human factors testing).
The testing described (Mechanical, Durability, Cleaning, Operating Manual Verification, Human Factors Validation) are typically prospective engineering and usability tests conducted by the manufacturer, not retrospective studies on patient data. The provenance for such tests would be the manufacturer's testing facilities, likely in the US (given the submitter's address in California).

3. 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)

This section is not applicable in the context of this 510(k) submission. The "ground truth" concept, often associated with clinical diagnostic studies requiring expert review of medical images or patient outcomes, is not relevant here. The studies described are hardware and human factors engineering tests. For human factors, users (not necessarily "experts" in the medical sense, but representative users of the device) would participate to validate the user interface changes. Their qualifications are not specified but would typically align with the intended users (e.g., perfusionists, surgeons, nurses).

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

This is not applicable to the type of engineering and human factors testing described. Adjudication methods like "2+1" are used in clinical studies where multiple experts independently review cases and discrepancies are resolved by a third party.

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 is not applicable. The device is a mechanical circulatory support system (pump, motor, console, flow probe), not an AI-assisted diagnostic or clinical decision support tool that would involve human readers interpreting data.

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

This is not applicable. The device is a physical medical device (mechanical support system), not a software algorithm or AI. Its operation inherently involves a human in the loop (a clinician operating the system).

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

"Ground truth" as traditionally understood in clinical validation studies (e.g., pathology for cancer detection, long-term outcomes for treatment efficacy) is not directly applicable here. For the engineering tests, the "ground truth" is simply the documented design specifications and functional requirements of the device. For human factors, the "ground truth" or success criteria would be based on validated usability principles and the ability of representative users to safely and effectively operate the device with the new feature.

8. The sample size for the training set

This is not applicable. This is a hardware modification to an existing device, not a machine learning or AI algorithm that requires a training set.

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

This is not applicable, as there is no training set for a hardware modification.

Ask a specific question about this device

Extracorporeal Membrane Oxygenation

The LifeSPARC System is a centrifugal blood pump system intended to assist in circulation of the patient's blood when part of an extracorporeal circuit including physiologic gas exchange of the patient's blood in adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent. These may include:

· Failure to wean from cardiopulmonary bypass following cardiac surgery in adult patients;

· ECMO-assisted cardiopulmonary resuscitation in adults.

Cardiopulmonary Bypass

The LifeSPARC System is intended to pump the blood through an extracorporeal circuit for periods lasting less than 6 hours for the purpose of providing either: (i) Full or partial cardiopulmonary bypass (i.e., circuit includes an oxygenator) during open surgical procedures on the heart or great vessels; or (ii) Temporary circulatory bypass for diversion of flow around a planned disruption of the circulatory pathway necessary for open surgical procedures on the aorta or vena cava.

The LifeSPARC Pump is a sterile, single-use, low prime volume centrifugal pump with an integrated motor and a single-point, pivot bearing. It is sterilized using ethylene oxide (EO) and sized to fit in the palm of the hand or to secure to the patient.

The LifeSPARC Controller provides the interface between pump and user, as well as the power and electrical signals to drive the pump. It is a microprocessor-based electromechanical pump drive system designed to operate on standard AC current (100/240 VAC, 50/60 Hz) or on internal, rechargeable batteries for intra-hospital transport.

The provided document is a 510(k) premarket notification summary for the CardiacAssist, Inc. LifeSPARC System. This submission appears to be for a software update to an already cleared device, not for a new device requiring extensive clinical trials to prove its performance against specific acceptance criteria.

The document does not contain the information requested regarding acceptance criteria and a study that proves the device meets those criteria, as such a study would typically be conducted for a novel device or a significant change in intended use, which is not the case here.

Here's why the requested information is absent and what the document does indicate:

1. A table of acceptance criteria and the reported device performance: This is not present. The submission focuses on demonstrating "substantial equivalence" to previously cleared versions of the LifeSPARC System, specifically due to a software update. This means the performance is assumed to be equivalent to the predicate devices, rather than being re-evaluated against new, explicit acceptance criteria.

2. Sample size used for the test set and the data provenance (e.g., country of origin of the data, retrospective or prospective): Not applicable in the context of this submission. The testing mentioned is "Software Verification and Validation testing (including Cybersecurity)," which typically does not involve human patient data or "test sets" in the clinical sense.

3. 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): Not applicable. "Ground truth" established by clinical experts is relevant for diagnostic or AI-driven devices that interpret complex data (like medical images). This device is a mechanical pump system with a controller, and its "performance" is assessed through engineering and software validation.

4. Adjudication method (e.g., 2+1, 3+1, none) for the test set: Not applicable for the reasons stated above.

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: Not applicable. This device is not an AI diagnostic tool and does not involve "human readers" interpreting data with or without AI assistance.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done: The document mentions "Software Verification and Validation testing." This type of testing evaluates the algorithm's performance in isolation (standalone), but it's in the context of a software update for a medical device's control system, not a diagnostic algorithm.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.): Not applicable. For software validation, the "ground truth" would be the expected functional behavior of the software as defined by its requirements and specifications.

8. The sample size for the training set: Not applicable. This is not an AI/machine learning model that requires a training set.

9. How the ground truth for the training set was established: Not applicable.

Summary of what the document does state regarding performance and testing:

• Device Change: The submission is for an update to the software version of a currently cleared device (LifeSPARC System). All other aspects of the device are identical to the predicate and reference devices.
• Testing Performed: "Software Verification and Validation testing (including Cybersecurity) was completed for the Subject device with the new software."
• Conclusion: "Testing described in this notification demonstrates that the Subject LifeSPARC System performance is substantially equivalent to the legally marketed Predicate LifeSPARC System (K211830) and Reference device (K232132)." The document explicitly states that "no concerns regarding safety and effectiveness result from the software changes."

In essence, this FDA 510(k) clearance is based on demonstrating that the software update does not change the safety or effectiveness of the device compared to its previously cleared versions, and therefore, it is "substantially equivalent." It is not a submission for a new device requiring extensive new clinical performance data against pre-defined acceptance criteria.

Ask a specific question about this device

The VitalFlow™ Centrifugal Pump is intended to pump blood through the extracorporeal circuit for circulatory support up to 48 hours, in adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent.

The VitalFlow™ Centrifugal Pump is driven by the VitalFlow™ Console and Drive Motor, or the Emergency Handcrank.

The VitalFlow Centrifugal Pump is a sterile, single-use centrifugal blood pump. It is a non-invasive, nonpyrogenic device designed to move blood through the extracorporeal circuit by centrifugal force. The pump is the disposable portion of the pumping system and it is electromagnetically coupled to an instrument that monitors and displays the flow and pressure of the blood. Blood enters the inlet port of the pump, where a cone with impeller blades within the pump housing rotates and the blood is gently accelerated toward the outlet of the pump.

The VitalFlow Centrifugal Pump can be driven through magnetic coupling by an External Drive Motor or the Emergency Handcrank.

The provided document is a 510(k) Summary for the VitalFlow™ Centrifugal Pump, a Class II medical device. It details the device's indications for use, description, and studies conducted to demonstrate its substantial equivalence to predicate and reference devices.

Here's an analysis of the acceptance criteria and study information:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly present a table of "acceptance criteria" with quantitative targets in the way one might expect for a diagnostic AI device (e.g., target specificity, sensitivity). Instead, the acceptance criteria are framed within the context of demonstrating substantial equivalence to predicate devices and meeting specific "Special Controls" outlined in 21 CFR 870.4100. The performance is assessed through various bench, animal, and real-world clinical data.

Therefore, the table below summarizes the Special Controls as acceptance criteria and how the device's performance, as reported, addresses them.

2. Sample Size Used for the Test Set and Data Provenance

The document describes several types of studies:

• Bench Performance Evaluations: No specific sample size is given for individual tests, but it states "Design verification studies were performed by the original manufacturer... and supplemented by MC3 bench performance testing, including long-term characterization." This implies multiple samples were tested for each characteristic.
• Animal Studies (In vivo evaluation): A sample size of 13 sheep was used for the 96-hour study. The data provenance is a controlled research environment.
• Clinical Performance (Real-world evidence): This utilized 1048 reports from the ELSO.org registry for the Affinity CP Centrifugal Blood Pump (AP40 group). The document states this is a "summary of real-world evidence," implying retrospective data extraction from an international registry. The country of origin for the data isn't explicitly stated but the ELSO registry collects data internationally. The comparison group ("All Other Pumps") involved 51,032 reports.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

This device is not an AI diagnostic device that relies on expert interpretation of images or signals to establish a "ground truth" in the traditional sense. The "ground truth" for its performance is established through objective measures in bench testing (e.g., flow, pressure, mechanical integrity), physiological parameters in animal studies (e.g., presence/absence of clots, functionality), and reported clinical outcomes in the ELSO registry.

Therefore, there is no mention of "experts" establishing ground truth in the context of adjudication for a test set. Design verification and animal study results would be assessed by engineers, veterinarians, and researchers involved in those studies. Clinical outcomes in the ELSO registry are reported data.

4. Adjudication Method for the Test Set

Given that this is not an AI diagnostic device evaluating, for example, medical images requiring consensus on findings, there is no adjudication method (e.g., 2+1, 3+1) mentioned or applicable for the test sets described.

5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was done

No, an MRMC comparative effectiveness study was not done. This type of study typically assesses the performance of human readers (e.g., radiologists) with and without AI assistance for tasks like diagnosis or detection. The VitalFlow™ Centrifugal Pump is a mechanical device, not an AI diagnostic tool, so such a study is not relevant.

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 done, as the VitalFlow™ Centrifugal Pump is a physical medical device. The "standalone" performance here would refer to the device's function outside of a human-in-the-loop context, which is effectively what the bench and animal studies demonstrate regarding its mechanical integrity and ability to pump blood.

7. The Type of Ground Truth Used

The "ground truth" for evaluating the VitalFlow™ Centrifugal Pump's performance is multi-faceted:

• Bench Testing: Engineering specifications and performance metrics (e.g., pressure-flow curves, durability, wear, integrity, blood trauma data).
• Animal Studies: Direct observation of device functionality, physiological parameters, and absence of adverse events like clot formation in a living system over a defined period (96 hours).
• Clinical Performance (Real-world evidence): Reported clinical outcomes and complication rates from a large patient registry (ELSO.org), providing real-world data on pump failure, hemolysis, circuit changes, and other adverse events. This relies on the accuracy of data submitted to the registry.

8. The Sample Size for the Training Set

The document is for a traditional medical device (centrifugal pump), not an AI/Machine Learning device. Therefore, there is no "training set" in the context of AI model development that would typically have a distinct sample size. The design verification, bench testing, and animal studies serve as part of the overall development and validation process.

9. How the Ground Truth for the Training Set Was Established

As there is no AI training set, this question is not applicable. The "ground truth" for the device's design and engineering would be based on established medical and engineering principles, material science, and regulatory requirements, which are then verified through rigorous testing, as mentioned above.

Ask a specific question about this device

The VitalFlow™ Console controls the VitalFlow™ Centrifugal blood pump during extracorporeal cardiopulmonary life support for adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent. The VitalFlow Centrifugal pump is driven by the VitalFlow Motor Drive or the VitalFlow Emergency Handcrank.

The VitalFlow Console provides control of blood pumping through an extracorporeal circuit during extracorporeal life support (ECLS) procedures. The console powers the VitalFlow motor drive unit which provides rotation of the VitalFlow Centrifugal pump. Pump motor speed (RPM) can be adjusted by the user and flow and bubble detection is provided by an ultrasonic flow probe and displayed on the touchscreen. The touchscreen display allows users to set alarm limits for all measured parameters. The device will alarm visually and audibly when limits are exceeded. Status indicators, power / battery life and secondary RPM indicator is provided. Data download and data streaming from the console is available for ECLS circuit data only; no patient data are stored or exported.

The VitalFlow Emergency Handcrank (drive unit) can be used in emergency situations to manually drive the centrifugal pump in the case of console and or motor drive failure.

The provided text describes the VitalFlow™ Console, an extracorporeal circuit component, and its FDA 510(k) clearance (K230364). The document outlines the device's indications for use, description, and the performance evaluations conducted to demonstrate substantial equivalence to a predicate device.

However, the document does not contain the detailed information necessary to answer all parts of your request, specifically regarding acceptance criteria for an AI/ML model, sample sizes for test sets in an AI context, expert ground truth establishment, MRMC studies, or training set details for an AI model. This is likely because the VitalFlow™ Console is a mechanical/electronic medical device, not an AI/ML-driven diagnostic or assistive device that would typically undergo such evaluations.

The "Bench Performance Evaluations" section lists various studies (Hydraulic performance, Reliability, EMC, Electrical Safety, Software Validation, Usability Studies), but these are standard engineering and regulatory tests for hardware and software functionality, not AI model performance.

Therefore, I cannot provide a table of acceptance criteria for an AI model or details about AI-specific testing from this document.

If you are looking for information regarding an AI/ML-driven medical device, this document does not seem to pertain to such a device.

From the provided text, I can infer the following about the device and its testing:

• Device Type: The VitalFlow™ Console is an extracorporeal circuit and accessories for long-term respiratory/cardiopulmonary failure (Product Code: QNR, Regulation Number: 21 CFR 870.4100). It controls features like pump motor speed (RPM), flow, and bubble detection.
• Purpose of Studies: The studies were conducted to demonstrate "substantial equivalence" of the VitalFlow™ Console to a predicate device (TandemHeart Escort Controller, K202751) and to meet the "Special Controls" described in 21 CFR 870.4100.
• Nature of Studies: The studies were primarily "Bench Performance Evaluations" and included aspects like hydraulic performance, reliability, EMC, electrical safety, software validation, and usability studies. These are typical engineering and functional performance tests for a physical device.
• Ground Truth: For a physical device like this, "ground truth" would be established through calibrated instruments and established engineering principles for parameters like flow rate, RPM, pressure, and electrical safety. It's not based on expert consensus on image interpretation or pathology.
• No AI Component: There is no mention of an AI/ML component or algorithm within the VitalFlow™ Console. Therefore, concepts like training sets, test sets for AI models, expert adjudication, or MRMC studies are not applicable to the information provided.

In summary, the provided document does not contain the information requested about AI/ML device acceptance criteria and study details because the VitalFlow™ Console appears to be a physical medical device, not an AI-driven one.

Ask a specific question about this device

The CentriMag™ Blood Pump for use with CentriMag™ Acute Circulatory Support System (Motor, Monitor, Console, and Flow Probes) is indicated for controlling blood flow as part of an extracorporeal membrane oxygenation (ECMO) circuit. ECMO is intended to provide assisted extracorporeal circulation and physiologic gas exchange of the patients' blood for adult patients with acute respiratory failure and/or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent.

The CentriMag™ Blood pump for use with CentriMag™ Acute Circulatory Support System (hereafter referred to as the CentriMag System) is designed to provide assisted extracorporeal circulation and physiologic gas exchange of patients' blood for adult patients with acute respiratory and/or acute cardiopulmonary failure. The CentriMag System was designed to provide temporary mechanical circulatory support. The CentriMag System provides circulatory assistance for patients in acute hemodynamic compromise, a population whose treatment options are limited. The system includes: CentriMag 2nd Generation Primary Console, CentriMag Motor, CentriMag Blood Pump, Flow Probe, and Mag Monitor (optional). The CentriMag System features a centrifugal flow pump with inflow and outflow ports that are at right angles to one another, and a magnetically levitated impeller (Full MagLev™ technology). The CentriMag Motor is a reusable, non-sterile component of the CentriMag Acute Circulatory Support System. The CentriMag Motor holds the blood pump and drives the impeller inside the blood pump. When the pump is inserted into the motor and activated, the internal impeller is electromagnetically levitated and centered, eliminating the need for shafts, seals, and bearings in the pump. Utilizing magnetic levitation technology (Full MagLev™ technology) to suspend and spin the impeller eliminates bearing and seal friction, resulting in minimal heat generation and wear of the pump components. the console is used to control pump speed, the resultant blood flow, and monitor the operation of the system. A cable connects the console to the motor, allowing flexibility in the pump motor and pump positioning.

The provided text describes the CentriMag™ Blood Pump for use with the CentriMag™ Acute Circulatory Support System. While it details extensive performance testing and claims substantial equivalence, it does not present acceptance criteria in terms of specific quantitative metrics (e.g., sensitivity, specificity, accuracy) that would apply to an AI/ML device, nor does it provide a study that explicitly demonstrates the device meets these types of criteria using methods common for AI/ML performance evaluation (like those involving expert consensus for ground truth).

Instead, the document focuses on the engineering and clinical performance of a mechanical circulatory support system. The "acceptance criteria" appear to be implicit in meeting the special controls outlined in 21 CFR 870.4100(b) and demonstrating substantial equivalence to a predicate device through various non-clinical and clinical evaluations.

Here's a breakdown of the information that is present, interpreted in the context of the prompt, and highlighting what is not present but would be typical for AI/ML device evaluation:

1. A table of acceptance criteria and the reported device performance

The document does not provide a table with explicit acceptance criteria (e.g., target percentages for accuracy, sensitivity, specificity) and corresponding reported device performance for an AI/ML component. The "acceptance criteria" for this device seem to be implicitly tied to meeting regulatory requirements for extracorporeal circuits and accessories, and demonstrating performance comparable to a predicate device through a range of engineering and clinical tests.

Reported Device Performance (based on the provided text):

The performance of the CentriMag System was demonstrated through a comprehensive set of tests:

2. Sample size used for the test set and the data provenance

• Sample Size for Test Set: Not explicitly stated as a test set in the traditional AI/ML sense. The "clinical summary" refers to a "pre-specified statistical analysis plan from the ELSO Registry," which is a real-world clinical data source. The number of patients included in this propensity-matched analysis is not provided.
• Data Provenance: The data comes from the ELSO Registry, which is a registry for Extracorporeal Life Support Organization, suggesting a multi-center, international scope. The analysis was retrospective (utilizing existing registry data).

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

Not applicable in the typical AI/ML context. The clinical study used the ELSO Registry, which records actual clinical outcomes and complications. The "ground truth" here is the aggregated, documented patient outcomes from the registry, not an expert-annotated dataset for an algorithm.

4. Adjudication method for the test set

Not applicable. As the "ground truth" is derived from a clinical registry of patient outcomes, there's no mention of an adjudication method used by experts for a test set. The registry likely has its own data entry and validation protocols.

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

No such study was conducted or mentioned, as the device is a mechanical circulatory support system, not an AI-assisted diagnostic or decision support tool for human readers.

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

This question is not applicable to the CentriMag Blood Pump, as it is a mechanical device, not an algorithm.

7. The type of ground truth used

For the clinical evaluation, the "ground truth" consists of outcomes data from the ELSO Registry, specifically ECMO-related clinical complications and other patient outcomes.

8. The sample size for the training set

Not applicable. The CentriMag Blood Pump is a hardware device; it does not involve a training set for an AI/ML algorithm.

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

Not applicable. There is no training set mentioned for this device.

Ask a specific question about this device

The LifeSPARC System is a centrifygal blood pump system intended to assist in circulation of the patient's blood when part of an extracorporeal circuit including physiologic gas exchange of the patient's blood in adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent. These may include:

• Failure to wean from cardiopulmonary bypass following cardiac surgery in adult patients
• ECMO-assisted cardiopulmonary resuscitation in adults

The LifeSPARC Pump is a sterile, single-use, low prime volume centrifugal pump with an integrated motor and a single-point, pivot bearing. It is sterilized using ethylene oxide (EO) and sized to fit in the palm of the hand or to secure to the patient.

The LifeSPARC Controller provides the interface between pump and user, as well as the power and electrical signals to drive the pump. It is a microprocessor-based electromechanical pump drive system designed to operate on standard AC current (100/240 VAC, 50/60 Hz) or on internal, rechargeable batteries for intra-hospital transport.

The provided text describes the LifeSPARC System, an extracorporeal blood pump system, and its substantial equivalence determination by the FDA. The submission focuses heavily on non-clinical testing and an animal study, with a clinical evaluation based on retrospective registry data.

Here's an analysis of the acceptance criteria and the study that proves the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria are derived from the "Special Controls" outlined by the FDA. The "reported device performance" is summarized from the "Summary of Non-clinical Testing," "Animal Study," and "Clinical Evaluation" sections.

2. Sample size used for the test set and the data provenance

Animal Study (In vivo evaluation):

• Test Set Sample Size: 9 calves were supported with the LifeSPARC System. A control group of 5 calves were supported with the TandemHeart Pump.
• Data Provenance: Prospective animal study (bovine model, presumably conducted in a controlled laboratory setting). The specific country of origin is not stated but is implied by the US FDA submission process.

Clinical Evaluation:

• Test Set Sample Size: The text states, "The study is an enumerative study where all available ECMO data collected in the ELSO registry database that met the study criteria were analyzed." The exact number of patients for the LifeSPARC Pump group and the comparator group(s) is not explicitly stated in the provided text. However, it's inferred to be a large dataset given it's from the ELSO Registry.
• Data Provenance: Retrospective observational data from the Extracorporeal Life Support Organization (ELSO) Registry. The ELSO Registry is an international registry, so the data provenance would likely be multi-national, although this is not explicitly stated. The data was "observational data in a real-world setting."

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

Animal Study:

• The text describes "close monitoring of the systems identified early problems allowing repairs or pump exchange before becoming clinically significant." This implies veterinary or medical expertise for monitoring and assessing animal health.
• The assessment of "clinically significant thromboembolic or bleeding complications, vascular or myocardial injuries, adherent thrombus, or end organ damage" would typically fall under the purview of veterinary pathologists and clinicians.
• Number of experts and qualifications: Not specified.

Clinical Evaluation:

• The ground truth for the clinical outcomes (e.g., pump failure, hemolysis, thrombosis, death) would have been established by the clinicians and medical professionals managing the patients and inputting data into the ELSO Registry. The ELSO statistical group performed the analysis.
• Number of experts and qualifications: Not specified. The ELSO statistical group is mentioned for analysis ("independent statistical analysis," "propensity score analysis was performed by the ELSO group"), but not for establishing the initial ground truth.

4. Adjudication method for the test set

Animal Study:

• The text indicates "close monitoring" and "identified early problems allowing repairs or pump exchange." This suggests continuous assessment by the study team. However, a formal adjudication method for adverse events or outcomes (e.g., blinded review by a panel) is not described.

Clinical Evaluation:

• The clinical evaluation is based on existing data from the ELSO Registry. The ground truth (patient outcomes) would have already been established by the reporting institutions.
• A formal adjudication method for the test set outcomes, beyond the standard data collection and reporting mechanisms of the ELSO Registry, is not described. The analysis by the ELSO statistical group is a statistical comparison, not an adjudication 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

• No, an MRMC comparative effectiveness study was not done.
• The device is an extracorporeal blood pump system, not an AI-assisted diagnostic or interpretative device that involves "human readers" or "AI assistance" in the sense of image interpretation or similar tasks.
• The clinical evaluation was a comparison of the device's performance against other similar devices (other centrifugal pumps, including a predicate) using retrospective registry data, not a human-in-the-loop study.

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

• No, this concept is not directly applicable to the LifeSPARC System. This device is a mechanical pump, not an algorithm. Its performance is inherent to its physical operation and interaction with the patient's blood, not an independent algorithmic output. The "standalone" performance is its performance in the animal study and its outcomes as observed in the clinical registry data.

7. The type of ground truth used

Animal Study:

• Clinical observation and physiological measurements from the calves during the 14-day support period.
• Post-mortem analysis for "adherent thrombus, or end organ damage."

Clinical Evaluation:

• Outcomes data / Real-world patient data from the ELSO Registry. This includes clinically recorded events like "pump failure, hemolysis, thrombosis or clots in circuit component, patient death within 24 hours of stopping ECMO, and patient death at discharge," as well as "use of renal replacement therapy (RRT during ECMO, pulmonary hemorrhage, circuit change, central nervous system (CNS) hemorrhage, and CNS infarction." This data is based on reporting by treating clinicians.

8. The sample size for the training set

• The provided text relates to regulatory submission for a medical device (an extracorporeal blood pump system), not a machine learning model. Therefore, the concept of a "training set" in the context of AI/ML is not applicable.
• If interpreted as "data used to inform the device's design or prior evaluations," the text references a "currently-cleared devices" and "reference devices" (TandemHeart System, LifeSPARC System K183623), implying previous testing and clinical experience would have contributed to the device's development, but no specific "training set" size for this purpose is given.

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

• As the concept of a "training set" is not applicable in this context (not an AI/ML device), this question is moot.

Ask a specific question about this device

The TandemHeart System is a centrifugal blood pump system intended to assist in circulation of the patient's blood when part of an extracorporeal circuit including physiologic gas exchange of the patient's blood in adult patients with acute respiratory failure or acute cardiopulmonary failure, where other available treatment options have failed, and continued clinical deterioration is expected or the risk of death is imminent.

The TandemHeart Pump is a low priming volume centrifugal blood pump with a dual chamber design. The upper chamber provides a conduit for the flow of blood; the lower chamber provides communication with the controller, a hydrodynamic bearing, cooling of the motor and local anticoagulation.
The TandemHeart Escort Controller provides the interface between pump and user, the power and electrical signals to drive the pump, and the infusate fluid to the lower chamber of the pump. It is a microprocessor-based electromechanical pump drive and infusion system designed to operate on standard AC current (100/240 VAC, 50/60 Hz) or on internal, rechargeable batteries for intra-hospital transport. The controller contains a backup motor control unit, and backup batteries.

This document describes the acceptance criteria and the study that proves the TandemHeart Pump and Escort Controller, a centrifugal blood pump system, meets these criteria. The device is intended to assist in the circulation of a patient's blood as part of an extracorporeal circuit for adult patients with acute respiratory or cardiopulmonary failure.

1. Acceptance Criteria and Reported Device Performance

The acceptance criteria are derived from the "Special Controls" outlined in the FDA's Final Order, 81 FR 7451, Feb. 12, 2016. The study compared the TandemHeart pump against "comparator pumps" (non-TandemHeart centrifugal pumps) using data from the ELSO Registry.

• Pump Failure: -0.7% (95% CI -1.4%, 0.1%) (TandemHeart had 0.2% vs. 0.8% for comparator)
• Hemolysis: -3.6% (95% CI -14.2%, 7.2%) (TandemHeart had 8.4% vs. 11.9% for comparator, but not statistically significant)
• Thrombosis/Clots in Circuit Component: -2.2% (95% CI -4.8%, 0.4%) (TandemHeart had 1.5% vs. 3.8% for comparator)

The confidence intervals for these treatment effects generally covered 0%, indicating no statistically significant difference in these outcomes between the TandemHeart pump and comparator pumps in the primary analysis. The analysis concluded that the TandemHeart Pump is substantially equivalent to other pumps. |

2. Sample Size Used for the Test Set and Data Provenance

• Sample Size:
  • Weighted Propensity Score Analysis (January 2016-October 2020):
    • TandemHeart pump: 165 adult ECMO runs
    • Comparator pumps: 3525 adult ECMO runs
  • TandemHeart Pump and Comparator Pumps Analysis (January 2016-January 2019):
    • TandemHeart pump: 79 runs (38 VA mode, 32 VV mode, 79 All modes)
    • Other Centrifugal Pumps: 9801 runs (5214 VA mode, 3905 VV mode, 9801 All modes)
• Data Provenance: The data was extracted from the Extracorporeal Life Support Organization (ELSO) Registry. This registry collects data on ECMO runs globally. While the document mentions "All United States extracorporeal membrane oxygenation (ECMO) runs" for the second analysis, the first analysis does not explicitly state a geographical restriction, implying it might be broader given the ELSO's global nature. The data is retrospective, as it was extracted from an existing database of past clinical events.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

The propensity score analysis was conducted by the statistical group at ELSO in consultation with clinical experts in the field of ECLS. The exact number of experts is not specified. Their qualifications are described as "clinical experts in the field of ECLS."

4. Adjudication Method for the Test Set

The document does not describe an explicit adjudication method for the data within the ELSO registry. The data is reported directly from the registry. It's implied that the ELSO registry data itself serves as the "ground truth," which is collected and coded by the reporting institutions. The "clinical experts in the field of ECLS" were involved in guiding the analysis based on specific health risks, rather than adjudicating individual cases.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, What Was the Effect Size of How Much Human Readers Improve with AI vs Without AI Assistance?

This is not an MRMC comparative effectiveness study in the context of human readers being assisted by AI. This study compares the performance of two different medical devices (TandemHeart pump vs. other centrifugal pumps) directly, rather than evaluating human performance with or without AI assistance. Therefore, there is no information on the effect size of human reader improvement with AI assistance.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done?

This is not a study of a standalone algorithm's performance. The study evaluates the performance of the TandemHeart pump, a physical medical device, in a real-world clinical setting using retrospective registry data.

7. The Type of Ground Truth Used (Expert Consensus, Pathology, Outcomes Data, etc.)

The ground truth used is primarily outcomes data recorded in the ELSO Registry. This includes clinical outcomes such as pump failure, hemolysis, thrombosis/clots in circuit component, patient death, CNS hemorrhage, renal replacement therapy, pulmonary hemorrhage, and circuit change. These are documented complications and events observed during ECMO runs.

8. The Sample Size for the Training Set

No specific training set is mentioned as this is a device comparison study utilizing retrospective clinical registry data, not an AI model development study that typically involves distinct training and test sets. The entire ELSO dataset used in the analysis serves as the "test set" for the device's performance comparison.

9. How the Ground Truth for the Training Set Was Established

As no specific AI training set is described in the document, there is no information on how its "ground truth" would have been established.

Ask a specific question about this device