System and method for timely notification of treatments to healthcare providers and patient

1. A computer-implemented method to timely provide notifications of treatments, the method comprising:

generating, based on machine-learning techniques, a graphical model that processes data about a plurality of notification records and treatment data comprising de-identified records identifying a treatment received by a patient as being prescribed to the patient by a healthcare provider, each de-identified record including an identifier that uniquely distinguishes the patient from other patients, wherein:

each notification record in the plurality of notification records is associated with a respective notification for the treatment and identifies a respective channel through which the notification for the treatment was provided, and

the respective channel provides data indication a recipient of the notification;

learning, by the graphical model, a function that identifies relationships between treatments for the patient and each notification received by a recipient, wherein the function is based on values computed for variables in at least the treatment data processed by the graphical model;

computing, based at least on the graphical model and the function, an impact parameter that indicates whether notifications received by the healthcare provider and the patient resulted in i) the healthcare provider prescribing the treatment to the patient and ii) the patient receiving the treatment prescribed by the healthcare provider; and

determining, based on the impact parameter, a plan comprising a timing of notifications that causes an increase or prescribed treatments received by a plurality of patients and a reduction in usage of processing and storage resources of a system that includes the graphical model.

2. The method of claim 1 , comprising:

identifying a first identifier that uniquely distinguishes the patient from other anonymized patients in a de-identified longitudinal medical record; and

determining that the first identifier matches a second identifier that uniquely distinguishes the patient from other anonymized patients in a treatment prescription record.

3. The method of claim 1 , wherein the graphical model represents a patient impact model and the method comprises:

determining healthcare provider related parameters for a joint patient-healthcare provider impact model using the patient impact model, the plurality of notification records, and treatment prescription records of the healthcare providers; and

determining patient related parameters for the joint patient-healthcare provider impact model using the plurality of notification records, the de-identified records, and the healthcare provider related parameters.

4. The method of claim 3 , wherein determining the healthcare provider related parameters and the patient related parameters are sequentially iterated multiple times.

5. The method of claim 1 , comprising:

providing the plurality of notification records, information indicating relationships identified between a plurality of healthcare providers and anonymized patients to the graphical model to output a joint patient-healthcare provider impact model.

6. The method of claim 1 , wherein determining the plan comprises:

obtaining information describing notification constraints;

obtaining information describing computing resources necessary to provide notifications through respective channels to patients;

obtaining information describing resources necessary to provide notifications through respective channels to healthcare providers; and

determining the plan based at least on the information describing notification constraints, information describing resources necessary to provide notifications through respective channels to patients, the information describing resources necessary to provide notifications through respective channels to healthcare providers, and the impact parameter.

7. The method of claim 1 , comprising: receiving prescription records of the healthcare provider in response to providing a request for prescription records of healthcare providers with particular attributes.

8. The method of claim 1 , comprising:

receiving de-identified longitudinal medical records in response to providing a request for de-identified longitudinal medical records of patients with particular attributes.

9. The method of claim 1 , wherein the notification records encode information identifying notifications provided through channels including a digital channel and a non-digital channel.

10. The method of claim 1 , comprising:

scheduling, based at least on the plan, a first set of notifications to be provided through respective channels to only the patients, a second set of notifications to provided through respective channels to only the healthcare providers, and a third set of notifications to be provided to both the healthcare providers and the patients.

11. A system comprising one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:

generating, based on machine-learning techniques, a graphical model that processes data about a plurality of notification records and treatment data comprising de-identified records identifying a treatment received by a patient as being prescribed to the patient by a healthcare provider, each de-identified record including an identifier that uniquely distinguishes the patient from other patients, wherein,

each notification record in the plurality of notification records is associated with a respective notification for the treatment, and identifies a respective channel through which the notification for the treatment eras provided, and

the respective channel provides data indicating a recipient of the notification;

learning, by the graphical model, a function that identifies relationships between treatments for the patient and each notification received by a recipient, wherein the function is based on values computed for variables in at least the treatment data processed by the graphical model;

computing, based at least on the graphical model and the function, an impact parameter that indicates whether notifications received by the healthcare provider and the patient resulted in: i) the healthcare provider prescribing the treatment to the patient and ii) the patient receiving the treatment prescribed by the healthcare provider; and

determining, based on the impact parameter, a plan comprising a timing of notifications that causes an increase of prescribed treatments received by a plurality of patients and a reduction in usage of processing and storage resources of a system that includes the graphical model.

12. The system of claim 11 , wherein the operations comprise:

identifying a first identifier that uniquely distinguishes the patient from other anonymized patients in a de-identified longitudinal medical record; and determining that the first identifier matches a second identifier that uniquely distinguishes the patient from other anonymized patients in a treatment prescription record.

13. The system of claim 11 , wherein the graphical model represents a patient impact model and the operations comprises:

determining healthcare provider related parameters for a joint patient-healthcare provider impact model using the patient impact model, the plurality of notification records, and treatment prescription records of the healthcare providers; and

determining patient related parameters for the joint patient-healthcare provider impact model using the plurality of notification records, the de-identified records, and the healthcare provider related parameters.

14. The system of claim 13 , wherein determining the healthcare provider related parameters and the patient related parameters are sequentially iterated multiple times.

15. The system of claim 11 , wherein the operations comprise:

providing the plurality of notification records, information indicating relationships identified between a plurality of healthcare providers and anonymized patients to the graphical model to output a joint patient-healthcare provider impact model.

16. The system of claim 11 , wherein determining the plan comprises:

obtaining information describing notification constraints;

obtaining information describing computing resources necessary to provide notifications through respective channels to patients;

obtaining information describing resources necessary to provide notifications through respective channels to healthcare providers; and

determining the plan based at least on the information describing notification constraints, information describing resources necessary to provide notifications through respective channels to patients, the information describing resources necessary to provide notifications through respective channels to healthcare providers, and the impact parameter.

17. The system of claim 11 , wherein the operations comprise:

receiving prescription records of the healthcare provider in response to providing a request for prescription records of healthcare providers with particular attributes.

18. The system of claim 11 , wherein the operations comprise:

receiving de-identified longitudinal medical records response to providing a request for de-identified longitudinal medical records of patients with particular attributes.

19. The system of claim 11 , wherein the notification records encode information identifying notifications provided through channels including a digital channel and a non-digital channel.

20. A non-transitory computer-readable medium storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to perform operations comprising:

generating, based on machine-learning techniques, a graphical model that processes data about a plurality of notification records and treatment data comprising de-identified records identifying a treatment received by a patient as being prescribed to the patient by a healthcare provider, each de-identified record Including an Identifier that uniquely distinguishes the patient from other patients, wherein:

each notification record in the plurality of notification records is associated with a respective notification for the treatment and identifies a respective channel through which the notification for the treatment was provided, and

the respective channel provides data indicating a recipient of the notification;

learning, by the graphical model, a function that identifies relationships between treatments for the patient and each notification received by a recipient, wherein the function is based on values computed for variables in at least the treatment data processed by the graphical model;

comprising, based at least on the graphical model and the function, an impact parameter that indicates whether notifications received by the healthcare provider and the patient resulted in: i) the healthcare provider prescribing the treatment to the patient and ii) the patient receiving the treatment prescribed by the healthcare provider; and

determining, based on the impact parameter, a plan comprising a timing of notifications that causes an increase of prescribed treatments received by a plurality of patients and a reduction in usage of processing and storage resources of a system that includes the graphical model.