Systems and methods for creating and selecting models for predicting medical conditions
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-10672509-B2
Publication Date
2020-06-02
Expiration Date
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.
Core Innovation
A computer-implemented method for selecting one or more models for predicting medical conditions receives initial data related to a user, including blood glucose levels. The initial blood glucose data includes at least some continuous glucose monitoring (CGM) data received wirelessly over a server, and includes at least some self-monitored blood glucose data. The method processes the initial data using one or more machine learning algorithms to identify patterns between the blood glucose levels and other data related to the user's medical condition.
The method assigns values to an unmeasured data category for which no measured data is received by the processor, and processes 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, the blood glucose levels, and the other data. Based on the identified patterns, it creates a library of models containing a plurality of models.
After creating the library of models, the method receives additional data related to the user, including additional blood glucose levels from CGM and self-monitored sources. From the received additional data, metadata is extracted that indicates whether a given stream of blood glucose data is continuous glucose monitor data or self-monitored blood glucose data, and the selected one or more models are applied.
The method selects a first type of model when the blood glucose levels are CGM data, selects a second type of model when the blood glucose levels are self-monitored blood glucose data, and selects the second type of model when the blood glucose levels include both CGM data and self-monitored blood glucose data. When application of the selected one or more models indicates an intervention is necessary, a notification is generated.
Claims Coverage
The partial content includes three independent claims: a method for selecting predictive models, a system configured to perform the method, and a non-transitory computer-readable medium storing instructions to perform the method. Across these independent claims, the inventive features center on model-library creation from patterns including an unmeasured data category, metadata-driven model selection based on CGM versus self-monitored blood glucose sources, and notification generation when intervention is indicated.
Metadata-driven model selection based on CGM versus self-monitored data
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, and selecting a first type of model for continuous glucose monitor data, a second type of model for self-monitored blood glucose data, and the second type of model when both data types are present.
Creating a library of models from patterns using unmeasured data category values
Processing initial data with one or more machine learning algorithms to identify patterns between blood glucose levels and other data related to the user's medical condition, assigning values to an unmeasured data category for which no measured data is received, processing the unmeasured data category to identify patterns, and creating a library of models containing a plurality of models based on the identified patterns.
Intervention notification based on applying selected models
Applying the selected one or more models and generating a notification when the application of the selected one or more model indicates an intervention is necessary.
System with server-based receipt of CGM and self-monitored glucose data for model selection
Receiving initial and additional blood glucose data wirelessly over a server from continuous glucose monitoring devices and self-monitored blood glucose sources, selecting the one or more models from the library of models based on the extracted metadata, applying the selected one or more models, and generating a notification when intervention is necessary.
Non-transitory computer-readable medium performing device-type metadata selection of model types
Receiving data measured from two different types of devices, extracting metadata to determine which device type was used, assigning values to a first data category for which no measured data is received, processing the initial data to identify patterns, creating a library of models, selecting model types based on the extracted metadata, applying the selected one or more models, and generating a notification when intervention is necessary.
Across the independent claims, the core claim coverage is the combination of creating a library of models from identified patterns that include assigning values for an unmeasured data category, extracting metadata to determine CGM versus self-monitored data streams and selecting model types accordingly, and applying the selected model(s) to generate a notification when an intervention is necessary.
Stated Advantages
Generates a notification when the application of the selected one or more model indicates an intervention is necessary.
Selects model types based on extracted metadata indicating whether blood glucose data is continuous glucose monitor data or self-monitored blood glucose data.
Creates a library of models containing a plurality of models based on identified patterns between blood glucose levels and other data, including patterns involving values assigned for an unmeasured data category.
Documented Applications
Predicting medical conditions and generating notifications for intervention when model application indicates an intervention is necessary.
Interested in licensing this patent?