Micro models and layered prediction models for estimating sensor glucose values and reducing sensor glucose signal blanking

Patent metadata

Field	Value
Publication number	US-12138047-B2
Application number	US-202117156490-A
Country	US
Kind code	B2
Filing date	Jan 22, 2021
Priority date	Jan 22, 2021
Publication date	Nov 12, 2024
Grant date	Nov 12, 2024

Abstract

Methods, systems, and devices for improving continuous glucose monitoring (“CGM”) are described herein. More particularly, the methods, systems, and devices describe applying layered machine learning models to generate predicted sensor glucose values. The system may use the predicted sensor glucose values to display a sensor glucose value to a user. The layered models may generate more reliable sensor glucose predictions across many scenarios, leading to a reduction of sensor glucose signal blanking. The methods, systems, and devices described herein further comprise applying a plurality of micro model to estimate sensor glucose values under outlier conditions. The system may prioritize the models that are trained for certain outlier conditions when the system detects those outlier condition based on the sensor data.

First claim

What is claimed is:

1. A sensor device for applying layered machine learning models to reduce sensor glucose signal blanking, the sensor device comprising:
memory configured to store a plurality of machine learning models, wherein each machine learning model of the plurality of machine learning models differs with respect to one or more data characteristics comprising probabilistic reliance and is trained using training data comprising clinical data on sensor glucose behavior; and
a processor configured to:
receive continuous glucose monitoring (CGM) sensor data comprising:
an interstitial current signal;
an electrochemical impedance spectroscopy signal;
a counter voltage; or
any combination thereof;

input the CGM sensor data into the plurality of machine learning models;
receive outputs from the plurality of machine learning models indicating a plurality of predicted sensor glucose values; and
generate for display, on a display interface, a sensor glucose value based on weighting the plurality of predicted sensor glucose values according to probabilistic reliance of each of the plurality of machine learning models.

2. The sensor device of claim 1 , wherein the one or more data characteristics comprise CGM sensor data availability or CGM sensor data accuracy.

3. The sensor device of claim 1 , wherein the probabilistic reliance is based on past CGM sensor data trends.

4. The sensor device of claim 1 , wherein the processor is further configured to:
determine, from the plurality of machine learning models, a subset of machine learning models that generate results that are compliant with integrated continuous glucose monitoring (iCGM) criteria;
rank the subset of the plurality of machine learning models according to the probabilistic reliance of each machine learning model of the subset; and
select a sensor glucose value from the plurality of predicted sensor glucose values that is associated with a machine learning model of the subset having a lowest probabilistic reliance.

5. The sensor device of claim 1 , wherein the sensor glucose value is based on a weighted average of the plurality of predicted sensor glucose values.

6. The sensor device of claim 1 , wherein the processor is further configured to generate for display, on the display interface, a confidence value associated with the sensor glucose value.

7. A method for applying layered machine learning models to reduce sensor glucose signal blanking, the method comprising:
receiving, at a sensor device, CGM sensor data comprising:
an interstitial current signal;
an electrochemical impedance spectroscopy signal;
a counter voltage; or
any combination thereof;

inputting the CGM sensor data into a plurality of machine learning models, wherein each machine learning model of the plurality of machine learning models differs with respect to one or more data characteristics comprising probabilistic reliance and is trained using training data comprising clinical data on sensor glucose behavior;
receiving, at the sensor device, outputs from the plurality of machine learning models indicating a plurality of predicted sensor glucose values; and
generating for display, on a display interface, a sensor glucose value based on weighting the plurality of predicted sensor glucose values according to probabilistic reliance of each of the plurality of machine learning models.

8. The method of claim 7 , wherein the one or more data characteristics comprise CGM sensor data availability or CGM sensor data accuracy.

9. The method of claim 7 , wherein the probabilistic reliance is based on past CGM sensor data trends.

10. The method of claim 7 , further comprising:
determining, from the plurality of machine learning models, a subset of machine learning models that generate results that are compliant with iCGM criteria;
ranking the subset of the plurality of machine learning models according to the probabilistic reliance of each machine learning model of the subset; and
selecting a sensor glucose value from the plurality of predicted sensor glucose values that is associated with a machine learning model of the subset having a lowest probabilistic reliance.

11. The method of claim 7 , wherein the sensor glucose value is based on a weighted average of the plurality of predicted sensor glucose values.

12. The method of claim 7 , further comprising generating for display, on the display interface, a confidence value associated with the sensor glucose value.

13. A non-transitory computer-readable storage medium comprising instructions for applying layered machine learning models to reduce sensor glucose signal blanking, the instructions, when executed by one or more processors, causing operations comprising:
receiving, at a sensor device, CGM sensor data comprising:
an interstitial current signal;
an electrochemical impedance spectroscopy signal;
a counter voltage; or
any combination thereof;

inputting the CGM sensor data into a plurality of machine learning models, wherein each machine learning model of the plurality of machine learning models differs with respect to one or more data characteristics comprising probabilistic reliance and is trained using training data comprising clinical data on sensor glucose behavior;
receiving, at the sensor device, outputs from the plurality of machine learning models indicating a plurality of predicted sensor glucose values; and
generating for display, on a display interface, a sensor glucose value based on weighting the plurality of predicted sensor glucose values according to probabilistic reliance of each of the plurality of machine learning models.

14. The non-transitory computer-readable storage medium claim 13 , wherein the one or more data characteristics comprise CGM sensor data availability or CGM sensor data accuracy.

15. The non-transitory computer-readable storage medium of claim 13 , wherein the operations further comprise:
determining, from the plurality of machine learning models, a subset of machine learning models that generate results that are compliant with iCGM criteria;
ranking the subset of the plurality of machine learning models according to the probabilistic reliance of each machine learning model of the subset; and
selecting a sensor glucose value from the plurality of predicted sensor glucose values that is associated with a machine learning model of the subset having a lowest probabilistic reliance.

16. The non-transitory computer-readable storage medium of claim 13 , wherein the sensor glucose value is based on a weighted average of the plurality of predicted sensor glucose values.

17. The non-transitory computer-readable storage medium of claim 13 , wherein the operations further comprise generating for display, on the display interface, a confidence value associated with the sensor glucose value.

Assignees

• Medtronic Minimed Inc

Inventors

• Ajemba Peter
• Nogueira Keith

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0895

  Weakly supervised learning, e.g. semi-supervised or self-supervised learning · CPC title

• A61B5/7475

  User input or interface means, e.g. keyboard, pointing device, joystick · CPC title

• G06N20/20 Primary

  Ensemble learning · CPC title

• G16H50/30

  for calculating health indices; for individual health risk assessment · CPC title