Search Results

The DEKA ACE Pump System is intended for the subcutaneous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin, ages six and above. The pump is able to reliably and securely communicate with compatible, digitally connected devices, including automated insulin dosing software, to receive, and confirm commands from these devices. The pump is intended for single patient, home use and requires a prescription.

The bolus calculator is indicated for use for aiding the user in determining the bolus insulin dosage for management of diabetes mellitus based on consumed carbohydrates, operator entered blood glucose, insulin sensitivity, insulin to carbohydrate ratio, target glucose values, and current insulin on board.

The DEKA ACE Pump System described herein contains a modification of the labeling of the previously cleared DEKA ACE Pump System (K233952) to add Novolog U-100 (insulin aspart) as a compatible insulin. The device is a wearable alternate controller enabled (ACE) insulin infusion pump intended to subcutaneously deliver insulin. at set and variable rates, for the management of diabetes mellitus in persons requiring insulin. The pump is able to reliably and securely communicate with compatible, digitally connected devices to receive, confirm, and execute commands. The pump is intended for single patient, ambulatory use and requires a prescription. The DEKA ACE Pump System is intended for the management of diabetes mellitus in persons six years of age and greater.

The pump was previously cleared and is indicated for use with Humalog U-100 insulin. This Special 510(k) utilizes the same methodology and acceptance criteria used to obtain the clearance and indication for Humalog U-100 to add the indication for Novolog U-100.

The DEKA ACE Pump System, consistent with the predicate K233952, consists of the following durable and disposable components:

1. Pump: A durable pump that incorporates fluid delivery algorithms and interfaces to a cassette, external wireless user interface, and iCGM. The pump is powered by a rechargeable lithium ion battery.
2. Cassette: A single-use pumping cassette that combines microfluidic valves, a pump chamber, insulin reservoir, and Acoustic Volume Sensing (AVS) measurement chamber. The cassette interfaces to a pump and off-the-shelf infusion sets.
3. DEKA Loop App: An iOS mobile application that serves as the primary user interface for the system. The DEKA Loop app can be downloaded onto the user's personal iPhone.

Also consistent with the predicate, the DEKA ACE Pump system includes the following electronic interfaces:

• Dexcom G6 iCGM: The DEKA ACE Pump System is compatible with Dexcom G6 iCGMs. Communication between the Dexcom G6 and DEKA ACE Pump is unchanged from the predicate device. The BLE central role on the ACE Pump radio processor connects directly to the Dexcom iCGM transmitter using the sensor's medical slot. The ACE Pump communicates with the iCGM transmitter at a regular interval to provide iCGM sensor data to the iAGC.
• Halo Cloud: Halo Cloud is a digital platform that connects the DEKA ACE Pump System to a variety of cloud-related services. These services include:
• Patient onboarding and device pairing via secure key transfer
• Prescription setting downloads to the ACE Pump
• Event log uploads from the ACE pump to the cloud
• Remote (OTA) software updates

No changes to the hardware or software of the system from that of the predicate are necessary to add the Novolog U-100 compatibility claim.

The provided text describes a 510(k) submission for the DEKA ACE Pump System, specifically seeking to add Novolog U-100 (insulin aspart) as a compatible insulin. The core of this submission is to demonstrate substantial equivalence to a previously cleared version of the same pump (K233952), which was cleared for use with Humalog U-100 insulin.

Based on the provided text, the device in question is an insulin pump, not an AI/Software as a Medical Device (SaMD) for image analysis or diagnosis. Therefore, many of the typical acceptance criteria and study aspects related to AI/SaMD (like MRMC studies, expert ground truth establishment for diagnostic images, sample sizes for training/test sets for AI models, etc.) are not applicable to this type of medical device submission.

The "acceptance criteria" in this context refer to the demonstration that the modified device (with Novolog U-100 compatibility) is substantially equivalent to the predicate device (with Humalog U-100 compatibility) in terms of safety and effectiveness, and that it meets all applicable Special Controls for Alternate Controller Enabled Infusion Pumps (21 CFR 880.5730, Product Code QFG).

Here's an attempt to answer the questions based on the provided document, highlighting what is and isn't applicable:

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

   The document doesn't present a table of quantitative acceptance criteria and reported numerical performance for a new study, but rather asserts that the same methodology and acceptance criteria used for Humalog U-100 were applied to Novolog U-100, implying the new performance matches the previously accepted thresholds. The comparison focuses on demonstrating equivalence across various characteristics.

   Implicit Acceptance Criteria (based on equivalence to predicate and Special Controls):

   • Indications for Use: No change.
   • Prescription Use: Yes.
   • Intended Population: Persons with Diabetes Mellitus ages 6 and above.
   • Environment of Use: Professional healthcare facilities and home healthcare environments.
   • Patient Environment: On-body wearable ambulatory pump.
   • Delivery Method: Microprocessor controlled Micro-dosing pump mechanism supplemented with acoustic volume sensor (AVS) feedback for monitoring delivery accuracy.
   • Insulin Basal Rate Delivery Range: 0 units/hour - 30 units/hour.
   • Insulin Bolus Delivery Range: Programmable from 0.05 - 25.00 Units in 0.01 Unit increments.
   • Basal Accuracy: Unchanged from K213536 (predicate's reference).
   • Bolus Accuracy: Unchanged from K213536 (predicate's reference).
   • Bolus Volume after Occlusion Release: No more than 0.74 units.
   • Time to occlusion alarm: 10 min (Bolus); 3 hours (Basal, 1 U/h); 6 hours (Basal, 0.1 U/hr).
   • Material Biocompatibility: Compliant with ISO-10993.
   • Cartridge/Cassette Shelf Life: 1 year.
   • Ingress Protection: IP28.
   • Applicable Safety Standards: Compliance with listed IEC and ISO standards (e.g., IEC 60601-1, ISO 14971).
   • Power Source: Rechargeable Lithium Ion Battery.
   • Pump Storage Conditions: Temperatures of -25 °C to 70 °C, Non-condensing humidity 15% to 90%.
   • Operating Conditions: Temperatures of 5 °C to 40 °C, Non-condensing humidity of 15% to 90%.
   • System User Feedback: Visual, audible, and vibratory.
   • Battery Operating Time: 72 hours.
   • New "Acceptance": The addition of Novolog U-100 compatibility does not impact the safety and effectiveness, and testing with Novolog U-100 is equivalent to that performed with Humalog U-100.

   Reported Device Performance (based on assertion of equivalence):
   For all the criteria listed above, the "Subject Device" (DEKA ACE Pump with Novolog U-100 compatibility) is stated to have the exact same characteristics and performance as the "Predicate Device" (DEKA ACE Pump with Humalog U-100 compatibility), or that its performance is equivalent to the predicate's established performance for the relevant aspects (e.g., Novolog U-100 testing being equivalent to Humalog U-100 testing). The key is that "No new or modified risks" are introduced.

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

   The document states: "No clinical data was obtained in support of this premarket submission."
   It also states that "Performance testing was performed in order to establish substantial equivalence for the compatibility claim of Novolog U-100 in comparison to the previously cleared compatibility claim for Humalog U-100 in terms of both safety and effectiveness..."

   This heavily implies in vitro or bench testing rather than human subject testing. The "test set" in this context would likely refer to the number of tests/replicates performed with Novolog U-100 to demonstrate equivalent performance to Humalog U-100, not a dataset for an AI model. This information (specific sample sizes for bench testing, e.g., number of pump cycles, number of insulin lots tested) is not provided in this summary document. Data provenance for such non-clinical testing is typically internal lab data, not human patient data with country of origin.

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. This is not an AI/SaMD for diagnostic image analysis. Ground truth would be established through engineering specifications, validated test methods, and measurement standards for pump performance, not expert consensus on medical images.

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

   Not applicable. This is not an AI/SaMD for diagnostic image analysis.

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 is not an AI/SaMD for diagnostic image analysis. The device is an insulin pump.

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

   Not applicable. The device has algorithms (fluid delivery, bolus calculator) but it's a physical pump with user interaction, not a standalone diagnostic algorithm being evaluated. Its performance is inherent to the pump's mechanical and electronic design working with insulin.

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

   The ground truth for the performance of the insulin pump (e.g., basal/bolus accuracy, occlusion detection, battery life) would be established through engineering specifications, standardized test methods, and quantifiable measurements (e.g., precision volumetric measurements, pressure sensor readings, time-to-event logging). For this submission, the "ground truth" for Novolog U-100 compatibility is that its chemical and physical properties allow the pump to perform identically or equivalently to how it performs with Humalog U-100, which has already been established as safe and effective.

8. The sample size for the training set

   Not applicable. This device does not use machine learning that requires a "training set" in the conventional AI sense for diagnostic tasks. The "training" of the pump would be its design, development, and validation testing based on engineering principles and regulatory standards.

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

   Not applicable. As above, there isn't a "training set" in the AI sense. The "ground truth" for the pump's design and function comes from established medical device standards, diabetes management requirements, and rigorous engineering testing.

Ask a specific question about this device

The DEKA ACE Pump System is intended for the subcutaneous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin, ages six and above. The pump is able to reliably and securely communicate with compatible, digitally connected devices, including automated insulin dosing software, to receive, execute, and confirm commands from these devices. The pump is intended for single patient, home use and requires a prescription.

The bolus calculator is indicated for use for aiding the bolus insulin dosage for management of diabetes mellitus based on consumed carbohydrates, operator entered blood glucose, insulin to carbohydrate ratio, target glucose values, and current insulin on board.

The DEKA ACE Pump system is a modification of the previously cleared DEKA ACE Pump System (K213536). The modified device is a wearable alternate controller enabled (ACE) insulin infusion pump (DEKA ACE Pump System) with the addition of an embedded iAGC (DEKA Loop). The user interface is an iOS app that can be downloaded to a user's iPhone.

The updated system is still intended to subcutaneously deliver insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin. The pump is able to reliably and securely communicate with compatible, digitally connected devices to receive, confirm, and execute commands. The Pump is intended for single patient, ambulatory use and requires a prescription. The Pump is indicated for use with Humalog U-100 insulin.

The DEKA ACE Pump System is indicated for the management of mellitus in persons six years of age and greater.

The system as described in this submission is able to be integrated with an integrated Continuous Glucose Monitor (iCGM). This submission also details the integration process that was used to incorporate the DEKA Loop iAGC Algorithm onto the DEKA ACE Pump.

The DEKA ACE Pump System consists of the following durable and disposable components:

• Pump: A durable pump that incorporates fluid delivery algorithms and interfaces to a cassette, external wireless user interface, and iCGM. The pump is powered by a rechargeable lithium ion battery.
• Cassette: A single-use pumping cassette that combines microfluidic valves, a pump chamber, insulin reservoir, and Acoustic Volume Sensing (AVS) measurement chamber. The cassette interfaces to a pump and off-the-shelf infusion sets.
• DEKA Loop App: An iOS mobile application that serves as the primary user face for the system. The DEKA Loop app can be downloaded onto the user's personal iPhone.

The provided text is a 510(k) summary for the DEKA ACE Pump System, focusing on demonstrating substantial equivalence to a previously cleared predicate device. It addresses modifications, primarily the addition of an embedded iAGC (DEKA Loop) and a change in the user interface to an iOS app, as well as a broadened age indication (from 13+ to 6+).

However, the document does not contain the specific information required to answer your request about acceptance criteria and study proving the device meets those criteria for aspects like deep learning model performance (e.g., accuracy, sensitivity, specificity, or AUC). The request is structured as if the document describes an AI/ML-based diagnostic device where performance metrics against a ground truth dataset would be evaluated.

Instead, this document describes a medical device (insulin pump) where the "performance testing" refers to engineering and quality assurance testing against technical standards and safety requirements (e.g., electrical safety, electromagnetic compatibility, biocompatibility, delivery accuracy of insulin). The "Acceptance Criteria" implicitly refer to meeting these established engineering and regulatory standards rather than statistical performance metrics of a diagnostic algorithm against a labeled test set.

Specifically:

• There is no mention of a deep learning model's performance metrics (accuracy, sensitivity, specificity, AUC) or related test set details (sample size, data provenance, ground truth establishment, expert adjudication, MRMC studies).
• The "DEKA Loop iAGC Algorithm" is mentioned as being integrated, but there are no details on how its performance was evaluated, other than "Human Factors testing demonstrates equivalent safety and effectiveness for the indicated population" (page 7). This suggests focus on usability and safety in human interaction, not algorithmic diagnostic performance.
• The document explicitly states "No clinical data was obtained in support of this premarket submission" (page 13), reinforcing that the evaluation was primarily non-clinical and focused on substantial equivalence based on existing data and engineering tests.

Given the content of the provided document, I cannot fulfill your request for the specific details outlined for an AI/ML-based diagnostic device. The available information relates to the regulatory submission for an insulin pump, which is evaluated against different types of performance criteria.

Ask a specific question about this device

The DEKA ACE Pump System is intended for the subcutaneous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin, ages 13 and above. The pump is able to reliably and securely communicate with compatible, digitally connected devices, including automated insulin dosing software, to receive, execute, and confirm commands from these devices. The pump is intended for single patient, home use and requires a prescription.

The bolus calculator is indicated for use for aiding the user in determining the bolus insulin dosage for management of diabetes mellitus based on consumed carbohydrates, operator-entered blood glucose, insulin to carbohydrate ratio, target glucose values, and current insulin on board.

The DEKA ACE Pump System is intended for the subcutaneous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin, ages 13 and above. The Pump is able to reliably and securely communicate with compatible, digitally connected devices, including automated insulin dosing software, to receive, and confirm commands from these devices. The Pump is intended for single patient, home use and requires a prescription.

The system as described in this submission is able to be integrated with a Dexcom G6 interoperable Continuous Glycemic Controller (iCGM). This submission also details the integration process that can be used to incorporate an iAGC.

The DEKA ACE Pump System consists of the following components:

1. Pump: A durable pump that incorporates fluid delivery algorithms and interfaces to an DEKA ACE Pump cassette, Remote Interface, iCGM, and iAGC. The pump is powered by a rechargeable lithium ion battery.
2. Cassette: A single-use pumping cassette that combines microfluidic valves, a pump chamber, insulin reservoir, and Acoustic Volume Sensing (AVS) measurement chamber. The cassette interfaces to an DEKA ACE Pump and off-the-shelf infusion set.
3. Remote Interface (Controller): A wireless controller that serves as the user interface to the DEKA ACE Pump system. This includes a large color touch display for ease of use.

The provided text describes the DEKA ACE Pump System, an insulin pump, and its comparison to a predicate device (Tandem t:Slim X2 insulin pump) for FDA 510(k) clearance. The document focuses on demonstrating substantial equivalence, primarily through non-clinical performance testing rather than clinical studies involving human patients.

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

1. Table of acceptance criteria and the reported device performance

The document presents basal and bolus accuracy as performance metrics, comparing the subject device (DEKA ACE Pump) to the predicate device (Tandem t:Slim X2). The acceptance criteria for these accuracies are implicitly set by matching or improving upon the predicate device's performance, and by meeting the "Special Controls" requirements for this device type.

Basal Accuracy Comparison (Example for 0.1 U/hr)

Bolus Accuracy Comparison (Example for 0.05U Bolus Accuracy - % of boluses within ranges)

Other performance criteria mentioned in the "Non-Clinical/Performance Testing" section include:

• Worst Case Accuracy
• Occlusions
• Fault Insertion
• Sound Testing
• Incidental Delivery
• Reliability
• Drug Compatibility and Particulate Testing
• System Level Functionality
• Battery Performance
• Environmental Conditions
  The acceptance criteria for these are generally that the device meets applicable standards (e.g., IEC 60601 series, ISO 10993, ISO 14971) and demonstrates performance "equivalent or better" or "meets all Special Controls requirements" compared to the predicate or established safety thresholds. Specific numerical acceptance criteria for these are not explicitly detailed in the provided text, but it states that the "subject device meets all Special Controls requirements."

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

The provided text only discusses non-clinical, in-vitro performance testing.

• Sample Size for Basal Accuracy: The tables show data collected over intervals of 1, 6, and 12 hours. The number of individual test runs or devices included in these measurements is not explicitly stated.
• Sample Size for Bolus Accuracy:
  • For 0.05U bolus accuracy: 800 boluses were analyzed for both predicate and subject device.
  • For 2.5U bolus accuracy (predicate) and 5U bolus accuracy (subject): 800 boluses were analyzed.
  • For 25U bolus accuracy: 256 boluses for predicate, 224 boluses for subject device.
• Data Provenance: The data is from non-clinical/performance testing (in vitro), not human patient data. Therefore, country of origin or retrospective/prospective classification in the context of human studies is not applicable. This data would have been generated in a lab setting by the manufacturer, DEKA Research & Development.

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 information is not applicable as there was no clinical study involving human data, and thus no expert consensus or ground truth established in the traditional sense of medical image analysis or diagnostic studies. The performance testing was based on direct physical measurements of insulin delivery accuracy against engineered specifications and regulatory standards.

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

This information is not applicable as there was no clinical study involving human data or expert review. Adjudication methods are relevant for resolving discrepancies in expert interpretations, which did not occur here.

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 an insulin pump, not an AI-powered diagnostic tool requiring human interpretation. No MRMC study was conducted. The document explicitly states: "No clinical data was obtained in support of this premarket submission."

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

Yes, in a sense, the non-clinical performance testing represents the standalone performance of the device (pump mechanism and associated components) in a controlled environment, without human intervention in the delivery process itself. The data presented for basal and bolus accuracy directly reflects the algorithmically controlled delivery mechanism's performance.

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

The "ground truth" for the performance testing of an insulin pump is the physical delivery of insulin and adherence to predefined engineering specifications and regulatory standards for accuracy, flow rates, and safety features (e.g., occlusion alarms). This is established through highly precise laboratory instrumentation and metrology, not derived from expert consensus, pathology, or outcomes data.

8. The sample size for the training set

This is not explicitly stated in the document. As an insulin pump, it's a hardware device with embedded software for control. While the software might have been developed using iterative testing and tuning, the concept of a "training set" as understood in machine learning (e.g., for an AI algorithm) is generally not directly applicable or documented in this type of submission for a physical medical device's core functionality.

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

This is not applicable in the context of this 510(k) submission for an insulin pump. The device's functionality is based on known physical principles and engineered control systems, not on learning from a "training set" with established ground truth labels in the AI sense. Development would involve extensive engineering verification and validation against performance requirements and regulatory standards rather than a machine learning training paradigm.

Ask a specific question about this device