Systems and methods for creating and selecting models for predicting medical conditions

• US 10,672,509 B2
• Filed: 10/13/2017
• Issued: 06/02/2020
• Est. Priority Date: 06/26/2013
• Status: Active Grant

• Current Assignee: WellDoc, Inc
• Original Assignee: WellDoc, Inc
• Inventors: Sudharsan, Bharath
• Primary Examiner(s): Chen, Alan
• Application Number:
  US15/783,674
• Publication Number:
  US 20180039757A1
• Time in Patent Office: 963 Days
• Field of Search: None
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G16H 20/10 relating to drugs or medications, e.g. for ensuring correct administration to patients
• CPC Class Codes: G16H 40/20 for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 40/67 for remote operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

Systems and methods are provided for selecting one or more models for predicting medical conditions. Art exemplary method may include receiving data related to a patient and extracting metadata from the received data. The method may further include selecting the one or more models from a library of models based on the extracted metadata and applying the selected one or more models. The method also may include generating a notification when the application of the selected one or more model indicates an intervention is necessary.

Claims

1. A computer-implemented method for selecting one or more models for predicting medical conditions, the method comprising:

receiving, using a processor, initial data related to a user, the initial data including blood glucose levels of the user, wherein receiving the initial data including blood glucose levels of the user includes receiving at least some initial data, wirelessly over a server, from a continuous glucose monitoring (CGM) device, and includes receiving at least some initial data that is self-monitored blood glucose data;

processing, using the processor and one or more machine learning algorithms, the initial data to identify patterns between the blood glucose levels of the user, and other data related to the user'"'"'s medical condition;

assigning values to an unmeasured data category for which no measured data is received by the processor, and processing the unmeasured data category, using the processor and the one or more machine algorithms, to identify one or more patterns between the values of the unmeasured data category, and blood glucose levels of the user and the other data,

creating a library of models containing a plurality of models based on the identified patterns;

after creating the library of models, receiving additional data related to the user, the additional data including blood glucose levels of the user, wherein receiving additional data including blood glucose levels of the user includes receiving at least some additional data, wirelessly over a server, from a continuous glucose monitoring (CGM) device, and includes receiving at least some additional data that is self-monitored blood glucose data;

extracting metadata from at least some of the received additional data, wherein extracting metadata from at least some of the received additional data comprises extracting metadata indicating whether a given stream of data relating to blood glucose levels is continuous glucose monitor data or self-monitored blood glucose data;

selecting the one or more models from the library of models based on the extracted metadata, wherein selecting the one or more models from the library of models based on the extracted metadata comprises selecting a first type of model when the extracted metadata indicates that the blood glucose levels are continuous glucose monitor data, selecting a second type of model when the extracted metadata indicates that the blood glucose levels are self-monitored blood glucose data, and selecting the second type of model when the extracted metadata indicates that the blood glucose levels include both continuous glucose monitor data and self-monitored blood glucose data;

applying the selected one or models; and

generating a notification when the application of the selected one or more model indicates an intervention is necessary.

2. The computer-implemented method of claim 1, further comprising:

transmitting the generated notification to one or more user devices.

3. The computer-implemented method of claim 1, wherein the notification includes corrective actions to be taken by the user to change a diet of the user.

4. The computer-implemented method of claim 1, wherein the one or more machine learning algorithms include one or more of support vector machines, k-nearest neighbor, Bayesian statistics, multi-layer perceptrons, and multivariate regressions.

5. The computer-implemented method of claim 1, further including predicting an occurrence of a hypoglycemic event, and a time at which the hypoglycemic event will occur, wherein the intervention includes one or more actions for the user to take to prevent occurrence of the hypoglycemic event.

6. The computer-implemented method of claim 1, wherein the intervention includes sending a control signal to an insulin pump of the user to regulate an amount of insulin administered to the user.

7. The computer implemented method of claim 1, wherein assigning values to the unmeasured data category is based on data received by the processor.

8. The computer-implemented method of claim 1, further including receiving a weight of the user, wherein the unmeasured data category is blood pressure of the user, the processor does not receive any blood pressure data, and values are assigned to the blood pressure of the user based on the weight of the user.

9. The computer-implemented method of claim 1, wherein the initial data also includes one or more of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user, and the method further includes processing, using one or more machine learning algorithms, the initial data to identify patterns between the blood glucose levels of the user, and the one or more of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user.

10. The computer-implemented method of claim 1, wherein the initial data also includes each of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user, and the method further includes processing, using one or more machine learning algorithms, the initial data to identify patterns between each of the blood glucose levels of the user, and the one or more of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user.

11. A system for selecting one or more models for predicting medical conditions, the system comprising:

a memory having processor-readable instructions stored therein; and

a processor configured to access the memory and execute the processor-readable instructions, which, when executed by the processor configures the processor to perform a method, the method comprising;

receiving initial data related to a user, the initial data including blood glucose levels of the user, wherein receiving the initial data including blood glucose levels of the user includes receiving at least some initial data, wirelessly over a server, from a continuous glucose monitoring (CGM) device, and includes receiving at least some initial data that is self-monitored blood glucose data;

assigning values to a first data category for which no measured data is received by the processor;

processing, using one or more machine learning algorithms, the initial data to identify patterns between the blood glucose levels of the user, and the values of the first data category;

creating a library of models containing a plurality of models based on the identified patterns;

after creating the library of models, receiving additional data related to the user, the additional data including blood glucose levels of the user, wherein receiving additional data including blood glucose levels of the user includes receiving at least some additional data, wirelessly over a server, from a continuous glucose monitoring (CGM) device, and includes receiving at least some additional data that is self-monitored blood glucose data;

extracting metadata from at least some of the received additional data, wherein extracting metadata from at least some of the received additional data comprises extracting metadata indicating whether a given stream of data relating to blood glucose levels is continuous glucose monitor data or self-monitored blood glucose data;

selecting the one or more models from the library of models based on the extracted metadata, wherein selecting the one or more models from the library of models based on the extracted metadata comprises selecting a first type of model when the extracted metadata indicates that the blood glucose levels are continuous glucose monitor data, and selecting a second type of model when the extracted metadata indicates that the blood glucose levels are self-monitored blood glucose data;

applying the selected one or models; and

generating a notification when the application of the selected one or more model indicates an intervention is necessary.

12. The system of claim 11, wherein the method further comprises:

transmitting the generated notification to one or more user devices.

13. The system of claim 12, wherein the generated notification contains one or more intervention options.

14. The system of claim 11, wherein assigning values to the first data category is based on data received by the processor.

15. The system of claim 11, further including receiving a weight of the user, wherein the first data category is blood pressure of the user, the processor does not receive any blood pressure data, and values are assigned to the blood pressure of the user based on the weight of the user.

16. A non-transitory computer-readable medium storing instructions, the instructions, when executed by a computer system cause the computer system to perform a method, the method comprising:

receiving initial data related to a user, the initial data including blood glucose levels of the user, wherein receiving the initial data including blood glucose levels of the user includes receiving at least some initial data, wirelessly over a server, measured from two different types of devices;

extracting metadata from the received initial data to determine which of the two different types of devices was used to measure the blood glucose data;

assigning values to a first data category for which no measured data is received by the processor;

processing, using one or more machine learning algorithms, the initial data to identify patterns between the blood glucose levels of the user, and the values of the first data category;

creating a library of models containing a plurality of models based on the identified patterns;

after creating the library of models, receiving additional data related to the user, the additional data including blood glucose levels of the user, wherein receiving additional data including blood glucose levels of the user includes receiving at least some additional data, wirelessly over a server, from each of the two different types of devices used to measure the blood glucose data;

selecting the one or more models from the library of models based on the extracted metadata, wherein selecting the one or more models from the library of models based on the extracted metadata comprises selecting a first type of model when the extracted metadata indicates that the blood glucose levels were measured from a first type of the two different types of devices, and selecting a second type of model when the extracted metadata indicates that the blood glucose levels were measured from a second type of the two different types of devices;

applying the selected one or models; and

generating a notification when the application of the selected one or more model indicates an intervention is necessary.

17. The non-transitory computer readable medium of claim 16, wherein the two types of devices includes a continuous glucose monitoring device, and a device for collecting self-monitored blood glucose data.

18. The non-transitory computer readable medium of claim 16, wherein assigning values to the first data category is based on data received by the processor.

19. The non-transitory computer readable medium of claim 16, further including receiving a weight of the user, wherein the first data category is blood pressure of the user, the processor does not receive any blood pressure data, and values are assigned to the blood pressure of the user based on the weight of the user.

20. The non-transitory computer readable medium of claim 16, wherein the initial data also includes one or more of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user, and the method further includes processing, using one or more machine learning algorithms, the initial data to identify patterns between the blood glucose levels of the user, and the one or more of nutrition consumed by the user, times of day associated with checking the blood glucose levels of the user, times of day associated with the nutrition consumed by the user, sleep of the user, exercise performed by the user, medication consumed by the user, and mood of the user.