Granted Patent B2

US 12,731,684 · App. 18/057,060 · Granted Sep 8, 2026

Inventors: Larissa Becka (Munich, DE); Alexander Buesser (Zürich, CH); Tony Huschto (Speyer, DE); Yannick Klopfenstein (Neuchatel, CH); David Mesterhazy (Nuremberg, DE); Mike Rinderknecht (Zürich, CH); Christian Ringemann (Mannheim, DE)

Assignee: Roche Diabetes Care, Inc.

G16H40/63G06N20/00

Abstract

Claims (39)

1 . A method for generating a software-implemented module for use in a glucose monitoring system to determine a glucose value in a body fluid, the method comprising:

(A) providing a first set of input data indicative of first values measured for a first input parameter and a second input parameter, wherein the first input parameter and/or the second input parameter is selected from the group of input parameters consisting of: glucose level measured by a continuous glucose sensor, insulin bolus, carbohydrate intake, active bolus insulin, active basal insulin, activity level, insulin sensitivity factor, carbohydrate ratio, stress level, and glycemic index of carbohydrates;

(B) providing a second set of input data indicative of second values for the first and second input parameters, the second values comprising:

(i) an augmented value for the first parameter, the augmented value being different from the first value measured for the first input parameter, and

(ii) the first value for the second input parameter;

(C) determining first analyte data indicative of a first set of glucose values in a body fluid by processing the first set of input data from Step (A) by a physiological model;

(D) determining second analyte data indicative of a second set of glucose values by processing the second set of input data from Step (B) by the physiological model;

(E) determining a set of training data from both the first analyte data and the second analyte data;

(F) determining a set of test data different from the set of training data;

(G) providing a software-implemented machine learning model configured to determine a glucose value in a body fluid of a patient;

(H) training the software-implemented machine learning model by the set of training data;

(I) testing the software-implemented machine learning model by the set of test data;

(J) deploying the generated software implemented module for use in the glucose monitoring system to analyze a present input data and determine an analyte value, wherein the analyte value is a present analyte value and/or a future analyte value.

2 . The method of claim 1 , wherein: determining of the first analyte data comprises determining first predictive analyte data indicative of a first time dependent course of the analyte values for the analyte over a prediction time period by processing the first set of input data by the physiological model; and determining of the second analyte data comprises determining second predictive analyte data indicative of a second time dependent course of the analyte values for the analyte over the prediction time period by processing the second set of input data by the physiological model.

3 . The method of claim 2 , wherein the prediction time period is provided as a continuation of the measurement time period.

4 . The method of claim 1 , wherein: providing of the first set of input data comprises providing a first set of input data indicative of first values measured for the first and second input parameters over a measurement time period; and providing of the second set of input data comprises providing a second set of input data indicative of second values for the first and second the input parameters over the measurement time period.

5 . The method of claim 1 , further comprising: receiving limit data indicative of a parameter limit for the first input parameter; and limiting augmenting of the first value for the first input parameter.

6 . The method of claim 1 , wherein determining of the set of test data comprises determining a set of test data from the first analyte data only.

7 . The method of claim 1 , wherein determining of the set of training data comprises: determining residual analyte data; determining augmented analyte data from the second analyte data and the residual analyte data; and determining the set of training data at least from the augmented analyte data.

8 . The method of claim 7 , wherein the determining of the residual analyte data comprises determining residual analyte data from the first analyte data and measured analyte data.

9 . The method of claim 1 , wherein providing of the first set of input data comprises providing a first set of continuous input data indicative of first values continuously measured for a plurality of input parameters.

10 . The method of claim 1 , further comprising: providing a third set of input data indicative of third values of input parameters, the third values comprising (i) an augmented value for at least one other input parameter from the first and second of input parameters, the augmented value being different from the first value measured for the at least one other input parameter and determined by augmenting the first value, and (ii) the first value for at least one remaining input parameter; determining third analyte data indicative of a third plurality of analyte values for the analyte by processing the third set of input data by the physiological model; and determining a set of training data from the first analyte data, the second analyte data, and the third analyte data.

11 . A non-transitory computer readable medium having stored thereon computer-executable instructions for performing the method according to claim 1 .

12 . A method for determining a glucose value in a body fluid, comprising, in an arrangement of one or more data processors: providing a software-implemented module generated by the method according to claim 1 ; providing present input data indicative of present values measured for a plurality of input parameters for a fluid containing an analyte being glucose in a body fluid; and determining a glucose value in a body fluid of a patient comprising analyzing the present input data by the software-implemented module.

13 . The method of claim 12 , further comprising at least one of: outputting the analyte value to the patient through an output device; and if the glucose value is below a minimum threshold or above a maximum threshold, outputting an alarm to the patient.

14 . A system for determining a glucose value in a body fluid, the system having an arrangement of one or more data processors and a software-implemented module generated by the method according to claim 1 , wherein the one or more data processors are configured to: provide present input data indicative of present values measured for a plurality of input parameters for a body fluid containing glucose; and determine a glucose value by analyzing the present input data by the software-implemented module.

15 . A system for generating a software-implemented module for use in a glucose monitoring system to determine a glucose value in a body fluid, the system comprising an arrangement of one or more data processors, wherein the one or more processors are configured to:

(A) provide a first set of input data indicative of first values measured for a first input parameter and a second input parameter, wherein the first input parameter and/or the second input parameter is selected from the group of input parameters consisting of: glucose level measured by a continuous glucose sensor, insulin bolus, carbohydrate intake, active bolus insulin, active basal insulin, activity level, insulin sensitivity factor, carbohydrate ratio, stress level, and glycemic index of carbohydrates;

(B) provide a second set of input data indicative of second values for the first and second input parameters, the second values comprising

(i) an augmented value for the first parameter, the augmented value being different from the first value measured for the first input parameter, and

(ii) the first value for the second input parameter;

(C) determine first analyte data indicative of a first set of glucose values in a body fluid by processing the first set of input data from (A) by a physiological model;

(D) determine second analyte data indicative of a second set of glucose values by processing the second set of input data from (B) by the physiological model;

(E) determine a set of training data from both the first analyte data and the second analyte data;

(F) determine a set of test data different from the set of training data;

(G) provide a software-implemented machine learning model configured to determine a glucose value in a body fluid of a patient;

(H) train the software-implemented machine learning model by the set of training data;

(I) test the software-implemented machine learning model by the set of test data; and

(J) deploy the generated software implemented module for use in the glucose monitoring system to analyze a present input data and determine an analyte value, wherein the analyte value is a present analyte value and/or a future analyte value.

Assignments (4)

Priority Claims (1)

Continuity (2)

References Cited (10)