(278 days)
The Volumetric Infusion Controller is intended for the delivery of general maintenance fluids and non-critical antibiotics to adult patients using gravity infusion in a clinical setting by a trained medical professional. The device is not intended to administer critical fluids, including high-risk medications.
The Volumetric Infusion Controller (VIC) is a gravity-based electronic infusion controller relying on head height to provide the delivery pressure necessary to meet the target infusion delivery rates. The drip chamber of an administration set is monitored by a vision system for drop growth. This information is used to provide feedback to the flow control valve to establish and maintain the target flow-rate without user intervention.
Here's an analysis of the acceptance criteria and study information for the Volumetric Infusion Controller, based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance:
| Condition | Acceptance Criteria | Reported Device Performance | Conclusion |
|---|---|---|---|
| Flow Rate Accuracy | |||
| Minimum 72 hour infusion at 10 mL/h and head height within operating range | The second hour is within ±20% of the programmed flow rate | -1.00 to -2.00% | Pass |
| The last hour of the infusions are within ±20% of the programmed flow rate | -0.80 to -1.80% | Pass | |
| Minimum 2 hour infusion at 25 mL/h and a head height simulated negative back pressure | The second hour is within ±20% of the programmed flow rate | -0.40 to -2.93% | Pass |
| 1L infusion at 300 mL/h and a head height within operating range | The second hour is within ±20% of the programmed flow rate | 5.56 to 6.42% | Pass |
| The last hour of the infusions are within ±20% of the programmed flow rate | 5.33 to 6.64% | Pass | |
| Maintenance of Set Flow Rate Despite Changes in Head Height (Increasing) | |||
| Increasing head height (50 to 200cm) at 10mL/h | • The device continues infusing and the accuracy does not exceed ±25%; or | ||
| • The device alarms | -7.03 to -9.87% | Pass | |
| Increasing head height (50 to 200cm) at 300mL/h | • The device continues infusing and the accuracy does not exceed ±25%; or | ||
| • The device alarms | 6.23 to 6.27% (1 device alarmed) | Pass | |
| Maintenance of Set Flow Rate Despite Changes in Head Height (Decreasing) | |||
| Decreasing head height (200 to 50cm) at 10mL/h | • The device continues infusing and the accuracy does not exceed ±25%; or | ||
| • The device alarms | -10.57 to -13.17% | Pass | |
| Decreasing head height (200 to 50cm) at 300mL/h | • The device continues infusing and the accuracy does not exceed ±25%; or | ||
| • The device alarms | 4.01% (2 devices alarmed) | Pass | |
| Reliability Analysis (Mean Time Between Failures - MTBF) | |||
| The devices performed the minimum number of infusion hours for the associated number of failures to establish MTBF $\ge$ 6 months with 90% confidence ($\ge$ 7596 hours for zero failures). | In order to pass with zero failures the number of infusion hours... would need to be $\ge$ 7596 hours. The actual number of infusion hours before discontinuing testing, with zero failures, was 7951 hours. | Pass | |
| Minimum number of infusion hours for associated number of failures (examples provided from table in text for context): | |||
| 0 failures: 7596 hours | |||
| 1 failure: 12816 hours | |||
| 2 failures: 17532 hours | 7951 hours (with zero failures) | Pass |
2. Sample size(s) used for the test set and the data provenance:
- Flow Rate Accuracy & Maintenance of Set Flow Rate despite changes in Head Height: The text does not explicitly state the number of devices or test runs/infusions used for each specific condition within these tests. However, the data provided shows a range of percentages for the "Test Results," implying multiple measurements were taken. The study was a bench test, conducted in accordance with IEC 60601-2-24:2012. The provenance is not specified, but it can be inferred to be from the manufacturer's testing (DEKA Research & Development). This would be prospective data collection.
- Reliability Analysis (MTBF): The test involved multiple devices ("devices are setup...continuously infusing solutions"). The actual number of infusion hours before discontinuing testing, with zero failures, was 7951 hours. The number of individual devices is not specified, nor is the cumulative number of "uses" for the 7951 hours. This is also prospective bench testing data.
- Other Testing (Electrical, EMC, Safety Control Mechanisms, Software, Human Factors): No specific sample sizes for these tests are provided in the summary. These are generally conducted as part of product development and validation, likely under controlled laboratory conditions, making them prospective bench testing.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- The studies described are bench tests and human factors evaluations, not clinical studies involving expert medical interpretation of data for ground truth establishment in a diagnostic context.
- For the performance testing (flow rate, reliability), the "ground truth" is defined by the objective physical measurements and adherence to specified standards (e.g., IEC 60601-2-24). No human experts are used to establish ground truth in this context; rather, calibrated measurement equipment and predefined standards serve this purpose.
- For Human Factors testing, usability evaluation and human factors testing were conducted, which would typically involve end-users (e.g., medical professionals), but the number and qualifications of these individuals, or how "ground truth" (e.g., successful task completion) was established, are not detailed in this summary.
4. Adjudication method for the test set:
- Not applicable. The reported studies are bench tests and engineering validations, not studies requiring adjudication of human-interpreted data. The results are based on objective measurements against pre-defined engineering and performance criteria.
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. This device is a Volumetric Infusion Controller, not an AI-powered diagnostic tool that assists human readers/interpreters. The FDA letter explicitly states, "A clinical investigation was not conducted, as the bench testing and human factors testing is sufficient to show the product is substantially equivalent in performance for its intended use."
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:
- Yes, the performance tests described (Flow rate accuracy, Maintenance of set flow rate, Reliability analysis, Electrical/hardware/mechanical safety, EMC, Alarms, Software V&V) were all standalone (algorithm/device only) performance evaluations. The device's ability to accurately control fluid delivery and operate reliably was tested independent of human interaction for the core performance metrics. Human factors testing assessed the interaction of users with the device, but the device's technical performance itself was standalone.
- The "vision system" that monitors drop growth and controls the flow valve functions as an algorithm/device-only system once initialized, without continuous human intervention.
7. The type of ground truth used:
- Objective physical measurements against predefined engineering standards and specifications (e.g., mL/hr, percentage accuracy, hours of operation). For example, flow rate accuracy is measured against the programmed flow rate. The reliability is measured against a statistical model for MTBF.
8. The sample size for the training set:
- This device does not appear to involve a "training set" in the context of machine learning, which is typically associated with AI/ML algorithms that learn from data. While the device uses a "vision system" and "similar algorithms" for drop detection, the text does not describe a machine learning model that was "trained" on a dataset. It refers to the vision system as using "similar algorithms" to make distinctions, implying a rules-based or traditional image processing approach rather than a modern deep-learning "training" paradigm. Therefore, a training set as typically understood for AI/ML is not applicable here based on the provided information.
9. How the ground truth for the training set was established:
- As a training set for an AI/ML model is not indicated from the text, the method for establishing its "ground truth" is not applicable. The device's "vision system" relies on specific detection algorithms that are likely designed and validated by engineers based on physical principles of drop formation and optical detection, rather than being "trained" on labeled data.
§ 880.5725 Infusion pump.
(a)
Identification. An infusion pump is a device used in a health care facility to pump fluids into a patient in a controlled manner. The device may use a piston pump, a roller pump, or a peristaltic pump and may be powered electrically or mechanically. The device may also operate using a constant force to propel the fluid through a narrow tube which determines the flow rate. The device may include means to detect a fault condition, such as air in, or blockage of, the infusion line and to activate an alarm.(b)
Classification. Class II (performance standards).