Search Results

The Tandem Mobi insulin pump with interoperable technology (the pump) is intended for the subcutaneous delivery of 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, 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 pump is indicated for use in individuals six years of age and greater.

The Subject Device, Tandem Mobi insulin pump with interoperable technology (" Mobi pump", "the pump"), is an Alternate Controller Enabled (ACE) Infusion Pump intended for the infusion of insulin into a patient requiring insulin therapy. The Tandem Mobi insulin pump with interoperable technology ( "pump") is screenless and includes visual LED, sound and vibratory indicators to alert the user of the pump status. The Tandem Mobi insulin pump with interoperable technology system also includes: the t:connect mobile app (K203234) and a 2mL (200 insulin unit) Tandem Mobi cartridge and a compatible FDA cleared infusion set. The t:connect mobile app (" Mobile app") displays all information from, and is the primary controller of, the pump. Through the Mobile app, users will program all aspects of basal and bolus insulin delivery therapy including managing personal profiles, viewing pump and CGM data, and actively acknowledging all pump and mobile app alerts, alarms, reminders, notifications and messages. The t:connect mobile app will also be used to transmit historical pump and mobile app therapy data to the Tandem Cloud. The t:connect mobile app will be made available via the Apple® App Store for iOS compatible smartphones based on completed device verification and validation. The Tandem Mobi cartridge is a disposable insulin cartridge compatible only with the Tandem Mobi pump.

The Tandem Mobi ACE pump can be used for basal and bolus insulin delivery with or without a CGM or with any compatible interoperable automated dosing algorithm.

The pump may be used in combination with a compatible continuous glucose monitor (CGM) system, such as the Dexcom G6 Continuous Glucose Monitoring System (DEN170088). Use of CGM is optional.

The provided document is a 510(k) summary for the Tandem Mobi insulin pump. It primarily focuses on demonstrating substantial equivalence to a predicate device rather than presenting a detailed clinical study with specific acceptance criteria and performance data in the format requested.

Here's an analysis of the provided text in relation to your questions:

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

The document does not provide a specific table of acceptance criteria with corresponding device performance metrics in the format you requested, such as sensitivity, specificity, or similar quantitative measures for an AI-driven diagnostic. Instead, it describes general non-clinical performance tests and states that the device "met specified requirements" or "performed as intended."

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

The document explicitly states: "No new clinical testing was performed to support this Traditional 510(k) Notification." Therefore, there is no test set of patient data from a clinical study to report on for the Tandem Mobi insulin pump itself. The evaluation relies on non-clinical performance tests and comparison to a predicate device.

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

Since no new clinical testing was performed, the concept of "ground truth" derived from expert consensus on a test set is not applicable here.

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

Not applicable, as no new clinical testing was performed.

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. The Tandem Mobi insulin pump is an Alternate Controller Enabled (ACE) Infusion Pump intended for insulin delivery, not a diagnostic device requiring human interpretation of medical images or data. Therefore, an MRMC study and the concept of "human readers improve with AI assistance" are not relevant to this device's evaluation.

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

The device itself is a pump for insulin delivery. While it has "interoperable technology" connecting to a mobile app and potentially automated insulin dosing software, the document doesn't describe a standalone algorithm performance in the context of a diagnostic AI system. The "algorithm only" would likely pertain to the internal control logic of the pump or any integrated automated dosing algorithms (which are referenced as separate devices like Control-IQ technology). The evaluation here focuses on the pump as a system, including its communication capabilities.

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

Given the nature of the device (an insulin pump), the "ground truth" for its performance would typically involve:

• Physiological measurements: Accurate insulin delivery, blood glucose control (as observed in predicate/reference device studies, not new studies for this 510k).
• Engineering specifications: Meeting defined tolerances for flow rate, pressure, safety, etc. (evaluated in non-clinical tests).
• System Integrity: Reliability of communication, software function, alarm systems, etc. (evaluated in non-clinical tests).

The summary indicates that Usability/Human Factors, Software Verification and Validation, Electrical Safety/EMC, Insulin Compatibility and Biocompatibility, and Sterilization and Shipping tests were performed. These tests use their own respective "ground truths" in the form of pre-defined requirements, standards (e.g., IEC 60601), or established safety profiles. For instance, in "Insulin Compatibility," the ground truth is that the insulin performs "as intended" in the device, presumably based on its known pharmacological properties and stability.

8. The sample size for the training set

Not applicable, as no new clinical testing or development of a new AI algorithm (in the diagnostic sense) is described. The device is a medical device leveraging established technology for insulin delivery.

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

Not applicable.

Summary Regarding Acceptance Criteria and Study:

The document effectively communicates that the Tandem Mobi insulin pump's acceptance criteria are primarily based on non-clinical performance tests demonstrating compliance with relevant standards and guidelines, and its substantial equivalence to an existing predicate device (K203234, the t:slim X2 Insulin Pump with interoperable technology).

The "study that proves the device meets the acceptance criteria" consists of:

• Usability/Human Factors validation testing: To ensure users can safely and effectively operate the device.
• Software Verification and Validation: To confirm the software meets requirements and performs as intended, adhering to standards like IEC 62304 and FDA guidance. This includes cybersecurity evaluations.
• Electrical Safety/EMC Testing: To ensure compliance with electrical safety and electromagnetic compatibility standards (e.g., IEC 60601).
• Insulin Compatibility and Biocompatibility Testing: To confirm the device functions correctly with approved insulins and that materials are biocompatible.
• Sterilization and Shipping Testing: To ensure the integrity of the device and its sterilization.
• Special Controls Adherence: Demonstrating compliance with special controls established for the predicate and reference devices.

The key takeaway is that for this 510(k) submission, the safety and effectiveness of the Tandem Mobi insulin pump are established through bench testing, engineering validation, and comparison to a legally marketed predicate device, rather than new clinical trials with patient-level data and AI performance metrics.

Ask a specific question about this device

Control-IQ technology is intended for use with compatible integrated continuous glucose monitors (iCGM) and alternate controller enabled (ACE) pumps to automatically increase, decrease, and suspend delivery of basal insulin based on iCGM readings and predicted glucose values. It can also deliver correction boluses when the glucose value is predicted to exceed a predefined threshold.

Control-IQ technology is intended for the management of Type 1 diabetes mellitus in persons 14 years of age and greater.

Control-IQ technology is intended for single patient use and requires a prescription.

Control-IQ technology is indicated for use with NovoLog or Humalog U-100 insulin.

Control-IQ technology (Control-IQ, the device) is a software-only device intended for use by people with diabetes. The device controls insulin delivery from a compatible alternate controller enabled insulin pump (ACE pump) based on inputs provided by a compatible integrated continuous glucose monitor (iCGM) and inputs provided the user (e.g., carbohydrate intake, exercise, and sleep schedule). Control-IQ technology is meant to be installed on a compatible ACE pump.

Control-IQ technology works to control glucose towards a glucose target range of 112.5-160 mg/dL during normal use. Glucose targets are not customizable but can be changed by a user if sleep or exercise modes are set or announced. During sleep mode, this range is changed to 112.5-120 mg/dL, and it is changed to 140-160 mg/dL during exercise mode.

Control-IQ technology includes an integrated feature whereby iCGM values are automatically populated into the glucose field of the integrated bolus calculator when the Control-IO technology is active (i.e., the device is operating in closed-loop mode). This feature is disabled when Control-IQ is turned off.

Using Control-IQ technology requires that users input their weight and their total daily insulin requirement, which should be established with the help of a health care provider before using the device.

Acceptance Criteria and Device Performance for Control-IQ Technology

1. Table of Acceptance Criteria and Reported Device Performance

The provided document does not explicitly state pre-defined acceptance criteria in terms of numerical thresholds for outcomes like HbA1c, Time in Range, or adverse event rates. Instead, the "Performance Characteristics" section reports the observed results of the pivotal study, which are then analyzed in the "Benefit/Risk Analysis" to generally conclude that the device's benefits outweigh its risks in light of the special controls.

However, based on the implicit goals of the study and the reported findings, we can infer some criteria that the device's performance needed to satisfy to be considered acceptable for De Novo authorization. These inferred criteria are focused on demonstrating improvements in glycemic control without an unacceptable increase in adverse events compared to the control arm (Sensor-augmented pump - SAP).

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

• Test Set Sample Size: The pivotal clinical study (randomization phase) included 168 participants, with 112 in the intervention arm (Control-IQ) and 56 in the control arm (Sensor-augmented pump - SAP). The study involved 6 months of follow-up for the primary study.
• Data Provenance: The study was a prospective, multicenter clinical trial conducted at Seven US clinical sites. The participants were diagnosed with Type 1 Diabetes and were aged 14 years and older.

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

The document does not describe the use of experts to establish a "ground truth" for the test set in the traditional sense of diagnostic accuracy studies (e.g., radiologists interpreting images). Instead, the clinical study directly measured physiological outcomes (CGM readings, HbA1c, adverse events) from the participants. These are considered the objective "ground truth" data for assessing the device's performance in managing diabetes.

Healthcare providers were involved in training participants and optimizing pump settings, and monitoring participants for adverse events, but not for establishing a separate "ground truth" label for individual data points that the algorithm would then predict against.

4. Adjudication Method for the Test Set

As the test set involved direct physiological measurements and reported adverse events, a formal adjudication method like "2+1" or "3+1" (common in image-based diagnostic studies) is not applicable in the same way.

Adverse events were reported and summarized. The document mentions an evaluation of hyperglycemia/ketosis events:

• "Hyperglycemia / ketosis events not meeting the definition of DKA were reportable if they met one of the following criteria: evaluation or treatment was obtained at a health care provider facility... blood ketone level ≥1.0 mmol/L and communication occurred with a health care provider... blood ketone level ≥3.0 mmol/L, even if there was no communication with a health care provider."
• This indicates predefined criteria for event reporting rather than an external expert adjudication process of subjective assessments. The DKA event was noted as having an "infusion set failure" as its cause, implying a clinical assessment of the event's origin.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No, an MRMC comparative effectiveness study was not explicitly conducted or mentioned. This type of study is more relevant for diagnostic devices where human readers interpret patient cases (e.g., images) with and without AI assistance. The Control-IQ device is an automated insulin delivery system, directly affecting physiological outcomes rather than assisting a human in making a diagnostic interpretation.

6. Standalone (Algorithm Only) Performance

Yes, the pivotal clinical study assessed the standalone performance of the Control-IQ technology. The intervention arm used "t:slim X2 with Control-IQ Technology and Dexcom G6 iCGM" (closed-loop control), representing the algorithm's performance without direct continuous human intervention in real-time insulin dosing decisions (though users still manually entered meal boluses and could adjust settings). The control arm (Sensor-augmented pump - SAP, with no automated insulin delivery) provided a baseline for comparison for human-managed delivery using similar sensor data.

The "Safety of CGM Auto-populating Bolus Calculator Feature" section also analyzes the performance when the iCGM values are "Automatic" (meaning auto-populated by the device) versus "Manual" (meaning patients manually adjusted the bolus calculation), providing insights into a more granular standalone feature compared to human override.

7. Type of Ground Truth Used

The ground truth used in the pivotal study was outcomes data directly measured from participants, including:

• Continuous Glucose Monitoring (CGM) readings (e.g., Time in Range, Time Below Range, Mean Glucose).
• HbA1c levels (a measure of average blood glucose over 2-3 months).
• Reported adverse events (severe hypoglycemia, DKA, hyperglycemia/ketosis).
• Surveys on quality of life aspects (diabetes-related distress and fear of hypoglycemia).

8. Sample Size for the Training Set

The document does not provide information on a separate "training set" sample size for the Control-IQ algorithm itself. Control-IQ is a software-only device containing a proprietary algorithm. Generally, such algorithms are developed and refined using a combination of preclinical modeling, simulations, and smaller-scale human studies, which would constitute the "training" or development sets. However, the exact size and nature of this development data are not detailed in this regulatory document, which focuses on the pivotal clinical validation study.

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

As with the training set sample size, the document does not describe how ground truth was established for any internal algorithm training set. For an automated insulin delivery algorithm, the ground truth for training would typically involve physiological responses to insulin in people with diabetes, potentially derived from historical patient data, controlled meal challenges, exercise protocols, and various other scenarios, often simulated or collected under tightly controlled observational studies. The goal is to build a model that accurately predicts glucose responses to insulin and other physiological factors.

Ask a specific question about this device