Search Results

Basal-IQ technology is intended for use with compatible integrated continuous glucose monitors (iCGM) and alternate controller enabled (ACE) pumps to automatically suspend delivery of insulin based on iCGM readings and predicted glucose values.

Basal-IQ technology is intended for the management of diabetes mellitus in persons six years of age and greater.

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

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

The bolus calculator is indicated for the management of diabetes by calculating an insulin dose or carbohydrate intake based on user entered data.

Basal-IQ technology is a Predictive Low Glucose Suspend (PLGS) algorithm for the management of diabetes mellitus and is compatible with an Alternate Controller Enabled Infusion Pump (cleared under 21 CFR 880.5730)(ACE pump). Basal-IQ technology is only compatible with the Tandem t:slim X2 insulin pump (DEN180058). The Basal-IQ software and algorithm can receive interstitial sensor glucose values from a compatible iCGM system (cleared under 21 CFR 862.1355), via Bluetooth Low Energy (BLE) communication. Compatible iCGM systems are cleared and marketed separately from the Basal-IQ algorithm and are identified in device labeling.

Basal-IQ assesses glucose information provided by a paired iCGM and sends commands to a compatible ACE pump to temporarily suspend insulin delivery in cases of impending or detected low blood glucose. Every 5 minutes, the Basal-IQ feature assesses glucose information provided by the iCGM to predict whether glucose values will fall below 80 mg/dL in the next 30 minutes or detect if glucose levels are currently below 70 mg/dL. Under these conditions it will command the compatible pump to suspend insulin delivery; otherwise insulin delivery continues as normal. After insulin delivery is suspended, insulin delivery resumption is commended when the system detects glucose values begin to rise. A sustained suspension period when blood glucose is above the sensor suspend threshold is mitigated by a maximum suspend time where Basal-IQ will command resume insulin delivery after 120 minutes of suspension within a 150-minute window. The Basal-IQ technology uses CGM sensor readings to send commands to a compatible insulin pump to stop and resume insulin based on the current sensor value and a 30-minute future predicted value along with the following rules:

• Insulin delivery is suspended if the current CGM sensor reading is less than 70 mg/dL 1.
• 2. Insulin delivery is suspended if the glucose value is predicted to be less than 80 mg/dL in 30 minutes.
• 3. Basal insulin delivery is resumed once the current CGM sensor reading increases compared to the previous reading.
• 4. Basal insulin delivery will also be resumed if the 30-minute predicted CGM reading is above 80 mg/dL, even if the CGM reading has not increased compared to the previous reading.
• 5. Basal insulin delivery is resumed if insulin delivery has been suspended for 2 hours in a 2.5 hour window.

The software comprising the Basal-IQ algorithm also includes an insulin bolus dose calculator. This calculator is for assisting patients with Type 1 diabetes who use insulin pumps as their insulin delivery therapy. It is used to calculate insulin bolus doses of rapid acting U-100 insulin analogs (Humalog and Novolog).The bolus calculator is used with manually-inputted glucose values and pump insulin delivery data to generate bolus size recommendations.

Here's an analysis of the acceptance criteria and study information for the Basal-IQ Technology, based on the provided document:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly present a table of acceptance criteria with numerical targets and the corresponding reported performance. Instead, it describes the functions of the Basal-IQ technology, which inherently define its intended performance. The clinical study then validated that the device functions as intended.

Here's an interpretation of the implicit operational criteria and the general statement of performance:

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

• Sample Size for Clinical Study (Test Set):
  • Total enrolled: 107 subjects with Type 1 Diabetes
  • Completed the study: 102 subjects
• Data Provenance:
  • Country of Origin: United States (4 sites)
  • Retrospective or Prospective: Prospective (clinical study with enrolled subjects)

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

The document does not mention the use of experts to establish a "ground truth" for the clinical study data in the traditional sense of independent readers reviewing output. The "ground truth" for this type of system is typically defined by real-time physiological data (CGM readings, blood glucose measurements) and the predefined algorithm logic. The clinical study assessed the device's performance against these physiological realities and its own built-in rules, observed by clinical investigators.

4. Adjudication Method for the Test Set

Not applicable in the context of this device and study design. The study evaluated the automated system's performance and user interaction, rather than requiring expert adjudication of outputs against a reference standard in the way an imaging AI algorithm might.

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

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not explicitly described or performed. The study was a crossover design comparing the Basal-IQ enabled pump to a Sensor Augmented Pump (SAP) in individuals with Type 1 Diabetes. This is a comparative effectiveness study but not in the MRMC format typically used for diagnostic or imaging AI.

• Effect size of how much human readers improve with AI vs without AI assistance: Not applicable, as this was not an MRMC study and the primary focus was on the direct performance and safety of the automated insulin suspension, not human reader improvement.

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

Yes, the core of the clinical study evaluates the standalone performance of the Basal-IQ algorithm (when enabled) in managing glucose levels by automatically suspending and resuming insulin delivery. While the device is used by a human, the algorithm's action (suspending/resuming basal insulin) is automatic based on CGM data. The "human factors study" later confirmed user interaction with the system, but the clinical efficacy portion inherently assesses the algorithm's automated functions.

7. The Type of Ground Truth Used

The ground truth used for evaluating the Basal-IQ algorithm's performance is based on:

• Physiological Data: Continuous Glucose Monitoring (CGM) readings and, presumably, corroborating blood glucose measurements (though not explicitly detailed as the "ground truth" method for the algorithm itself, these are the inputs).
• Pre-defined Algorithm Rules: The "ground truth" for whether the algorithm should have acted in a certain way is its own pre-programmed logic (e.g., predicted glucose below 80 mg/dL, current glucose below 70 mg/dL). Efficacy is then measured by how well applying these rules impacts patient outcomes (e.g., reduction in hypoglycemia).

8. The Sample Size for the Training Set

The document explicitly states: "No new non-clinical laboratory studies were needed for the separation from that system... The Basal-IQ Technology is the same algorithm, and the t:slim X2 Bolus Calculator is the same calculator, as reviewed in P180008, therefore no additional testing was conducted."

This implies that the algorithm was previously developed and validated. The current submission (K193483) is for a modification/reclassification rather than a new algorithm requiring a new training set. Therefore, information about the original training set sample size is not provided in this document.

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

Since the document states it's the "same algorithm" as previously reviewed (under P180008), the process for establishing ground truth for the original training set is not detailed in this K193483 submission. It would have been part of the P180008 documentation.

Ask a specific question about this device