Search Results

The Omnipod DASH™ Insulin Management System is intended for subcutaneous delivery of insulin at set and variable rates for the management of diabetes mellitus in persons requiring insulin.

Additionally. the Omnipod DASH System is interoperable with a compatible blood glucose meter to receive and display glucose measurements.

The Omnipod DASH™ Insulin Management System provides for the management of insulin therapy by patients with diabetes mellitus. It is comprised of two primary components: the disposable insulin infusion pump (Pod), and an associated Bluetooth Low Energy (BLE) enabled remote controller referred to as the Personal Diabetes Manager (PDM). The PDM can communicate with interoperable, compatible BLE enabled blood glucose meters. The PDM incorporates a suggested bolus calculator which aids the user in determining the insulin bolus dosage needed based on carbohydrates ingested, most recent blood glucose reading, programmable correction factor, insulin to carbohydrate ratio, target blood glucose value and Insulin on Board (IoB).

The Pod is a body-wearable insulin pump that affixes to the user on the back of the arm, the lower back or abdomen, the thigh area, or any site that has a layer of fatty tissue available. It is held in place by an adhesive pad and provides up to three days of insulin before it is removed and replaced with a new Pod. The PDM is a handheld device that controls the Pod. The user interfaces with the device system through the PDM using a touch screen, similar to a smartphone, where they control basal and bolus delivery and various insulin program settings and calculations. The PDM also has a food library to assist with carbohydrate calculations, and it maintains several variables in a history log for the viewer to track their diabetes therapy. The device system is for prescription use only.

The provided text describes the Omnipod DASH™ Insulin Management System and its 510(k) submission (K180045) to the FDA. The submission asserts substantial equivalence to a predicate device (K162296 Omnipod® Insulin Management System).

However, the document does not contain a typical acceptance criteria table with reported device performance in a numerical or statistical format that is commonly used for AI/ML-based diagnostic devices. Instead, it focuses on demonstrating safety and effectiveness through compliance with various standards and guidance documents. The "acceptance criteria" are implied by the successful completion of various tests and adherence to regulatory standards.

Here's a breakdown of the requested information based on the provided text, recognizing that it pertains to an insulin pump and not a diagnostic AI/ML device:

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

The document does not explicitly provide a table of acceptance criteria with corresponding performance metrics like sensitivity, specificity, accuracy, or other statistical measures, as would be common for AI/ML diagnostic tools. Instead, the "acceptance criteria" are implied by successful completion of various engineering, safety, and regulatory compliance tests. The "reported device performance" is that the device met these criteria.

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 lists various types of tests (bench testing, software V&V, human factors, biocompatibility, safety, EMC) but does not specify the sample sizes for these tests. For instance, for mechanical integrity or software testing, there's no mention of the number of devices or iterations tested. Similarly, no information is provided regarding data provenance (e.g., country of origin, retrospective/prospective). This type of detail is typically contained within the full test reports referenced by the 510(k) summary, but not elaborated here.

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 provided in a way that is relevant to AI/ML ground truthing. The "ground truth" for an insulin pump's performance is typically established through engineering specifications, regulatory standards, and clinical outcomes, rather than expert consensus on diagnostic interpretations. The document states that the safety assurance case and human factors validation involved identifying and mitigating risks (suggesting expert input in risk analysis), but it doesn't specify the number or qualifications of experts in the context of "ground truth" as it would for image-based diagnostic AI.

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

This detail is not mentioned. Adjudication methods like "2+1" or "3+1" are usually associated with establishing ground truth in clinical studies, particularly for diagnostic imaging. For an insulin pump, validation often relies on meeting predetermined engineering and safety specifications through direct measurement and testing, rather than a consensus-based adjudication process for interpretations.

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 MRMC study was performed or is referenced. This type of study is specifically designed for evaluating the impact of AI on human readers in diagnostic tasks, which is not applicable to the Omnipod DASH™ Insulin Management System as it is an insulin delivery device, not a diagnostic AI.

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

The device is an insulin pump, and its "algorithm" (the insulin dose calculator) is inherently designed for human-in-the-loop operation, where the user makes decisions based on the device's calculations and their own blood glucose readings. Therefore, a standalone "algorithm-only" performance study in the context of removing human interaction is not relevant or described. The document does mention "PDM software algorithm error results in errant insulin infusion program on the Pod," indicating that the software's performance is tested, but not as a replacement for human decision-making.

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

For this medical device, the "ground truth" is largely defined by:

• Engineering Specifications: The device must deliver insulin accurately, reliably, and safely according to its design parameters.
• Regulatory Standards: Compliance with international standards (e.g., ISO 14971 for risk management, ISO 10993 for biocompatibility, IEC 60601 series for electrical safety and performance, IEC 62304 for software).
• Safety Assurance Cases: Demonstrating that identified hazards are adequately mitigated and the system is acceptably safe for its intended use.
• Intended Use Validation: Human factors studies confirm that the device can be used safely by the target population.
• Clinical Efficacy (implied by predicate): The device's primary function (insulin delivery for diabetes management) is well-established therapy. The device demonstrates substantial equivalence to a predicate device, implying similar clinical effectiveness.

It is not based on expert consensus, pathology, or outcomes data in the typical sense of a diagnostic AI.

8. The sample size for the training set

Not applicable. The Omnipod DASH™ Insulin Management System is not an AI/ML device that undergoes "training" based on a dataset. It is a programmed medical device that operates based on predefined algorithms and user input.

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

Not applicable, as there is no "training set" for this type of medical device.

Ask a specific question about this device