Back to results
US 12,755,848 B2Granted
Machine learning for effective patient planning
AssigneeMatrixCare, Inc.
InventorsVivek Kumar, Coleen Danielson
Life of this patent
Estimating the term…
Abstract
Techniques for improved machine learning are provided. Resident data describing a first condition of a first resident of a residential care facility is received. An optimized treatment plan is selected for the first resident by extracting a first plurality of resident attributes, from the resident data, generating a first approach to remediate the first condition, and generating a first predicted recovery score by inputting the first approach and the first plurality of resident attributes into a trained machine learning model. The optimized treatment plan is implemented for the first resident.
Claims20
A method, comprising:
- receiving treatment data describing a historical treatment plan for a historical resident of a residential care facility;
- generating a historical recovery score based on the treatment data, wherein the historical recovery score indicates success of the historical treatment plan in treating the historical resident, wherein the historical recovery score is inversely related to an amount of time that elapsed while the historical resident used the historical treatment plan;
- training one or more machine learning models based on the historical recovery score;
- deploying the trained one or more machine learning models;
- receiving resident data describing a first condition of a first resident of the residential care facility, wherein the resident data comprises only updated data that has changed since a last time a treatment plan was generated for the first resident;
- in response to determining, at a first point in time, that a current time is not within a specified off-peak time, refraining from processing the resident data at the first point in time;
- in response to determining, at a second point in time subsequent to the first point in time, that the current time is within a specified off-peak time when additional computational resources are available as compared to a specified peak time, automatically selecting, by a computing system, a first treatment plan for the first resident using the trained one or more machine learning models, comprising:
- extracting a first plurality of resident attributes from the resident data for the first resident;
- determining a first goal for a first condition of the first resident;
- generating a first achievability score based on processing the first goal and the first plurality of resident attributes using a first machine learning model;
- generating a first approach to achieve the first goal;
- generating a first efficacy score by processing the first goal, the first approach, and the first plurality of resident attributes using a second machine learning model, wherein the first efficacy score indicates an amount of time that is predicted to elap se prior to achieving the first goal;
- generating a first predicted recovery score based on the first achievability score and the first efficacy score;
- generating a second predicted recovery score for a second treatment plan;
- and
- determining that the first predicted recovery score is greater than the second predicted recovery score;
- storing, by the computing system, the second treatment plan and the second predicted recovery score;
- automatically implementing, bythe computing system, the first treatment plan for the first resident, wherein the first treatment plan comprises the first approach, comprising allocating increased resources to an area where the first treatment plan will be provided to the first resident;
- and
- subsequent to implementing the first treatment plan, implementing the stored second treatment plan without additional execution of the second machine learning model.
17 more elements
A method, comprising:
- receiving resident data describing a first condition of a first resident of a residential care facility, wherein the resident data comprises only updated data that has changed since a last time a treatment plan was generated for the first resident;
- in response to determining, at a first point in time, that a current time is not within a specified off- peak time, refraining from processing the resident data at the first point in time;
- in response to determining, at a second point in time subsequent to the first point in time, that the current time is within a specified off-peak time when additional computational resources are available as compared to a specified peak time, automatically selecting, by a computing system, a first treatment plan for the first resident, comprising:
- extracting a first plurality of resident attributes, from the resident data, for the first resident;
- determining a first goal for the first condition;
- generating a first achievability score based on processing the first goaland the first plurality of resident attributes using a first machine learning model;
- generating a first approach to achieve the first goal;
- generating a first efficacy score by processing the first goal, the first approach, and the first plurality of resident attributes using a second machine learning model, wherein the first efficacy score indicates an amount of time that is predicted to elap se prior to achieving the first goal;
- generating a first predicted recovery score based on the first achievability score and the first efficacy score;
- generating a second predicted recovery score for a second treatment plan;
- and
- determining that the first predicted recovery score is greater than the second predicted recovery score;
- storing, by the computing system, the second treatment plan and the second predicted recovery score;
- automatically implementing, bythe computing system, the first treatment plan for the first resident, wherein the first treatment plan comprises the first approach, comprising allocating increased resources to an area where the first treatment plan will be provided to the first resident;
- and
- subsequent to implementing the first treatment plan, implementing the stored second treatment plan without additional execution of the second machine learning model.
13 more elements
The method of claim6, wherein selecting the first treatment plan comprises:
- generating a plurality of predicted recovery scores by processing a plurality of alternate approaches using the second machine learning model;
- and
- selecting the first approach based on determining that the first predicted recovery score is greater than each of the plurality of predicted recovery scores.
A non-transitory computer-readable storage medium comprising computer-readable program code that, when executed using one or more computer processors, performs an operation comprising:
- receiving resident data describing a first condition of a first resident of a residential care facility, wherein the resident data comprises only updated data that has changed since a last time a treatment plan was generated for the first resident;
- in response to determining, at a first point in time, that a current time is not within a specified off- peak time, refraining from processing the resident data at the first point in time;
- in response to determining, at a second point in time subsequent to the first point in time, that the current time is within a specified off-peak time when additional computational resources are available as compared to a specified peak time, automatically selecting, by a computing system, a first treatment plan for the first resident, comprising:
- extracting a first plurality of resident attributes, from the resident data, for the first resident;
- determining a first goal for the first condition;
- generating a first achievability score based on processing the first goaland the first plurality of resident attributes using a first machine learning model;
- generating a first approach to achieve the first goal;
- generating a first efficacy score by processing the first goal, the first approach, and the first plurality of resident attributes using a second machine learning model, wherein the first efficacy score indicates an amount of time that is predicted to elapse prior to achieving the first goal;
- and
- generating a first predicted recovery score based on the first achievability score and the first efficacy score;
- generating a second predicted recovery score for a second treatment plan;
- and
- determining that the first predicted recovery score is greater than the second predicted recovery score;
- storing, by the computing system, the second treatment plan and the second predicted recovery score;
- automatically implementing, bythe computing system, the first treatment plan forthe first resident, wherein the first treatment plan comprises the first approach, comprising allocating increased resources to an area where the first treatment plan will be provided to the first resident;
- and
- subsequent to implementing the first treatment plan, implementing the stored second treatment plan without additional execution of the second machine learning model.
14 more elements
Dependencies are inferred from the claim text.