Granted Patent B2

US 12,246,188 · App. 18/155,803 · Granted Mar 11, 2025

Inventors: Siddharth Dani (Minneapolis, MN); Tarek D. Haddad (Minneapolis, MN); Donald R. Musgrove (Minneapolis, MN); Andrew Radtke (Minneapolis, MN); Niranjan Chakravarthy (Singapore, SG); Rodolphe Katra (Blaine, MN); Lindsay A. Pedalty (Minneapolis, MN)

Assignee: Medtronic, Inc.

A61N1/3956A61N1/36592G16H10/60G16H40/63G16H50/20G16H50/50

Abstract

Claims (104)

1. A method comprising:

generating, by a computing system that comprises processing circuitry and one or more storage media, a set of sample probability values by applying a machine learning model to a sample set of patient data, wherein:

the machine learning model is trained using patient data for a plurality of patients,

the sample set comprises a plurality of temporal windows, and

for each respective temporal window of the plurality of temporal windows, the machine learning model is configured to determine a respective probability value in the set of sample probability values that indicates a probability that a cardiac arrhythmia occurred during the respective temporal window;

generating, by the computing system, graphical data based on the sample probability values;

outputting, by the computing system, a user interface for display on a display device, the user interface comprising the graphical data;

receiving, by the computing system, via the user interface, an indication of user input to select a probability threshold for a patient;

receiving, by the computing system, patient data for the patient, wherein the patient data for the patient is collected by one or more medical devices;

applying, by the computing system, the machine learning model to the patient data for the patient to determine a current probability value that indicates a probability that the patient has experienced an occurrence of the cardiac arrhythmia;

determining, by the computing system, that the current probability value exceeds the probability threshold for the patient;

in response to determining that the current probability value is greater than or equal to the probability threshold for the patient, generating, by the computing system, a notification indicating that the patient has likely experienced the occurrence of the cardiac arrhythmia; and

presenting, by the computing system, data indicating an anticipated review burden versus an anticipated diagnostic yield for the probability threshold for the patient.

2. The method of claim 1 , wherein:

generating the graphical data comprises generating, by the computing system, a receiver operating curve (ROC), wherein generating the ROC comprises:

for each evaluation probability threshold of a plurality of evaluation probability thresholds:

determining, by the computing system, a sensitivity value for the respective evaluation probability threshold as a ratio of: (i) a total number of sample probability values in the set of sample probability values that are greater than or equal to the respective evaluation probability threshold to (ii) a total number of the temporal windows in the sample set that actually contain occurrences of the cardiac arrhythmia that actually occurred in the sample set;

determining, by the computing system, a specificity value for the respective probability value as a ratio of: (i) a total number of the sample probability values that are not greater than or equal to the respective evaluation probability threshold to (ii) a total number of the temporal windows in the sample set that do not actually contain occurrences of the cardiac arrhythmia; and

determining, by the computing system, a point on the ROC that corresponds to the respective probability value, wherein the point on the ROC that corresponds to the respective probability value is based on the sensitivity value for the respective evaluation probability threshold and the specificity value for the respective evaluation probability threshold; and

receiving the indication of user input to select the probability threshold for the patient comprises: receiving, by the computing system, an indication of user input to select a point on the ROC that corresponds to the probability threshold for the patient.

3. The method of claim 1 , wherein:

generating the graphical data comprises:

generating, by the computing system, a graph that maps the sample probability values against time; and

generating, by the computing system, a threshold indicator; and

receiving the indication of user input comprises receiving, by the computing system, an indication of user input to position the threshold indicator at a location in the graph corresponding to the probability threshold for the patient.

4. The method of claim 3 , wherein the threshold indicator comprises a threshold bar superimposed on the graph and oriented parallel to a time axis of the graph.

5. The method of claim 1 , wherein receiving the patient data for the patient comprises receiving, by the computing system, cardiac electrical waveform data for the patient.

6. The method of claim 1 , further comprising receiving, by the computing system, an indication of user input to update the probability threshold for the patient.

7. The method of claim 1 , wherein the medical devices include a wearable medical device or an implantable medical device (IMD).

8. The method of claim 1 , wherein:

the cardiac arrhythmia is a first cardiac arrhythmia in a plurality of cardiac arrhythmias, and

the method comprises, for each respective cardiac arrhythmia of the plurality of cardiac arrhythmias:

generating, by the computing system, a respective set of sample probability values by applying a respective machine learning model to a respective sample set of patient data, wherein:

the respective machine learning model is trained using patient data for the plurality of patients,

the respective sample set comprises a respective plurality of temporal windows, and

for each respective temporal window of the respective plurality of temporal windows, the respective machine learning model is configured to determine a respective probability value in the respective set of sample probability values that indicates a probability that the respective cardiac arrhythmia occurred during the respective temporal window;

generating, by the computing system, respective graphical data based on the respective set of sample probability values;

outputting, by the computing system, the user interface for display on the display device such that the user interface comprises the respective graphical data;

receiving, by the computing system, via the user interface, an indication of user input to select a respective probability threshold for the patient;

applying, by the computing system, the machine learning model to the patient data to determine a respective probability value that indicates a probability that the patient has experienced an occurrence of the respective cardiac arrhythmia;

determining, by the computing system, that the respective probability value exceeds the respective probability threshold; and

in response to determining that the respective probability value is greater than or equal to the respective probability threshold, generating, by the computing system, a notification indicating that the patient has likely experienced the occurrence of the respective cardiac arrhythmia.

9. The method of claim 1 , wherein the patient is a first patient, the patient data for the patient is first patient data, the current probability value is a first current probability value, and the one or more medical devices are one or more first medical devices, and the method further comprises:

receiving, by the computing system, second patient data for a second patient, wherein the second patient data is collected by one or more second medical devices;

applying, by the computing system, the machine learning model to the second patient data to determine a second current probability value that indicates a probability that the second patient has experienced an occurrence of the cardiac arrhythmia;

determining, by the computing system, that the second current probability value exceeds a default probability threshold, wherein the default probability threshold is set to maximize diagnostic yield; and

in response to determining that the second current probability value is greater than or equal to the default probability threshold, generating, by the computing system, a second notification indicating that the second patient has likely experienced the occurrence of the cardiac arrhythmia.

10. A computing system comprising:

one or more processing circuits; and

one or more storage media storing instructions that, when executed by the one or more processing circuits, cause the one or more processing circuits to:

generate a set of sample probability values by applying a machine learning model to a sample set of patient data, wherein:

the machine learning model is trained using patient data for a plurality of patients,

the sample set comprises a plurality of temporal windows, and

for each respective temporal window of the plurality of temporal windows, the machine learning model is configured to determine a respective probability value in the set of sample probability values that indicates a probability that a cardiac arrhythmia occurred during the respective temporal window;

generate graphical data based on the sample probability values;

output a user interface for display on a display device, the user interface comprising the graphical data;

receive, via the user interface, an indication of user input to select a probability threshold for a patient;

receive patient data for the patient, wherein the patient data for the patient is collected by one or more medical devices;

apply the machine learning model to the patient data for the patient to determine a current probability value that indicates a probability that the patient has experienced an occurrence of the cardiac arrhythmia;

determine that the current probability value exceeds the probability threshold for the patient;

in response to determining that the current probability value is greater than or equal to the probability threshold for the patient, generate a notification indicating that the patient has likely experienced the occurrence of the cardiac arrhythmia; and

present data indicating an anticipated review burden versus an anticipated diagnostic yield for the probability threshold for the patient.

11. The computing system of claim 10 , wherein:

the one or more processing circuits are configured such that, as part of generating the graphical data, the one or more processing circuits generate a receiver operating curve (ROC), wherein, the one or more processing circuits are configured such that, as part of generating the ROC, the one or more processing circuits:

for each evaluation probability threshold of a plurality of evaluation probability thresholds:

determine a sensitivity value for the respective evaluation probability threshold as a ratio of: (i) a total number of sample probability values in the set of sample probability values that are greater than or equal to the respective evaluation probability threshold to (ii) a total number of the temporal windows in the sample set that actually contain occurrences of the cardiac arrhythmia that actually occurred in the sample set;

determine a specificity value for the respective probability value as a ratio of: (i) a total number of the sample probability values that are not greater than or equal to the respective evaluation probability threshold to (ii) a total number of the temporal windows in the sample set that do not actually contain occurrences of the cardiac arrhythmia; and

determine a point on the ROC that corresponds to the respective probability value, wherein the point on the ROC that corresponds to the respective probability value is based on the sensitivity value for the respective evaluation probability threshold and the specificity value for the respective evaluation probability threshold; and

the one or more processing circuits are configured such that, as part of receiving the indication of user input to select the probability threshold for the patient, the one or more processing circuits receive an indication of user input to select a point on the ROC that corresponds to the probability threshold for the patient.

12. The computing system of claim 10 , wherein:

the one or more processing circuits are configured such that, as part of generating the graphical data, the one or more processing circuits:

generate a graph that maps the sample probability values against time; and

generate a threshold indicator; and

the one or more processing circuits are configured such that, as part of receiving the indication of user input, the one or more processing circuits receive an indication of user input to position the threshold indicator at a location in the graph corresponding to the probability threshold for the patient.

13. The computing system of claim 12 , wherein the threshold indicator comprises a threshold bar superimposed on the graph and oriented parallel to a time axis of the graph.

14. The computing system of claim 10 , wherein the one or more processing circuits are configured such that, as part of receiving the patient data for the patient, the one or more processing circuits are configured to receive cardiac electrical waveform data for the patient.

15. The computing system of claim 10 , wherein the one or more processing circuits are further configured to receive an indication of user input to update the probability threshold for the patient.

16. The computing system of claim 10 , wherein the one or more medical devices comprises a wearable medical device or an implantable medical device (IMD).

17. The computing system of claim 10 , wherein:

the cardiac arrhythmia is a first cardiac arrhythmia in a plurality of cardiac arrhythmias, and

the one or more processing circuits are configured to, for each respective cardiac arrhythmia of the plurality of cardiac arrhythmias:

generating, by the computing system, a respective set of sample probability values by applying a respective machine learning model to a respective sample set of patient data, wherein:

the respective machine learning model is trained using patient data for the plurality of patients,

the respective sample set comprises a respective plurality of temporal windows, and

for each respective temporal window of the respective plurality of temporal windows, the respective machine learning model is configured to determine a respective probability value in the respective set of sample probability values that indicates a probability that the respective cardiac arrhythmia occurred during the respective temporal window;

generating, by the computing system, respective graphical data based on the respective set of sample probability values;

outputting, by the computing system, the user interface for display on the display device such that the user interface comprises the respective graphical data;

receiving, by the computing system, via the user interface, an indication of user input to select a respective probability threshold for the patient;

applying, by the computing system, the machine learning model to the patient data for the patient to determine a respective probability value that indicates a probability that the patient has experienced an occurrence of the respective cardiac arrhythmia;

determining, by the computing system, that the respective probability value exceeds the respective probability threshold; and

in response to determining that the respective probability value is greater than or equal to the respective probability threshold, generating, by the computing system, a notification indicating that the patient has likely experienced the occurrence of the respective cardiac arrhythmia.

18. One or more non-transitory computer-readable data storage media having instructions stored thereon that, when executed by one or more processing circuit of a computing system, cause the one or more processing circuits to:

generate a set of sample probability values by applying a machine learning model to a sample set of patient data, wherein:

the machine learning model is trained using patient data for a plurality of patients,

the sample set comprises a plurality of temporal windows, and

for each respective temporal window of the plurality of temporal windows, the machine learning model is configured to determine a respective probability value in the set of sample probability values that indicates a probability that a cardiac arrhythmia occurred during the respective temporal window;

generate graphical data based on the sample probability values;

output a user interface for display on a display device, the user interface comprising the graphical data;

receive, via the user interface, an indication of user input to select a probability threshold for a patient;

receive patient data for the patient, wherein the patient data for the patient is collected by one or more medical devices;

apply the machine learning model to the patient data for the patient to determine a current probability value that indicates a probability that the patient has experienced an occurrence of the cardiac arrhythmia;

determine that the current probability value exceeds the probability threshold for the patient;

in response to determining that the current probability value is greater than or equal to the probability threshold for the patient, generate a notification indicating that the patient has likely experienced the occurrence of the cardiac arrhythmia; and

present data indicating an anticipated review burden versus an anticipated diagnostic yield for the probability threshold for the patient.

Assignments (1)

Continuity (3)

References Cited (171)