System and method for timely notification of treatment
A computer-assisted method to timely provide notifications of treatments, the method including receiving de-identified longitudinal medical records, receiving notification data, identifying anonymized patients that received the treatment, identifying notifications for the treatment that were received by the recipients, determining, for each of the identified notifications, whether the recipient is an anonymized patient identified as having received the treatment, determining, for each of the identified notifications for the treatment determined to be received by a recipient that is an anonymized patient identified as having received the treatment, a time relationship between the time when the treatment was received by the anonymized patient and the time that the notification was received by the recipient that is the anonymized patient, and determining, for each of the anonymized patients that received the treatment, associations between one or more time relationships for notifications received by the anonymized patient.
1. A computer-implemented method comprising:
identifying a set of notifications, wherein:
a recipient received a notification,
an anonymized patient received a treatment,
the recipient patient based on (i) a set of first identifiers for a set of anonymized patients and (ii) a set of second identifiers for a set of recipients;
determining a set of time relationships between a set of treatments that were received by the set of anonymized patients and the set of notifications;
training a time decay model to identify, for each time relationship in the set of time relationships, a set of coefficients for a decay function representing an impact of the notification on the corresponding treatment relative to the time relationship;
computing, based on the set of coefficients and for each notification in the set of notifications, a score representing a likelihood that the notification being provided to the recipient impacted the corresponding treatment being received by the anonymized patient;
determining a notification plan for notifying potential patients of treatments based on the scores for each notification included in the set of notifications, wherein the notification plan is determined to increase a number of treatments to be subsequently received by the anonymized patients; and
scheduling the notifications based on the notification plan.
2. The method of claim 1 , comprising:
determining a forecast model based on scores computed for the set of notifications, where the forecast model stores, for each type of notification from among the set of notifications, a notification type label that is indicative of a relationship between a number of anonymized patients receiving the treatment and a number of recipients receiving notifications for the treatment.
3. The method of claim 2 , comprising:
scheduling notifications based on the forecast model.
4. The method of claim 1 , wherein identifying the set of notifications comprises:
determining that a second identifier for the recipient that received the notification and a first identifier for the anonymized patient that received the treatment both refer to the same person.
5. The method of claim 1 , wherein determining time relationships between the treatments that were received by the set of anonymized patients and the notifications that were received by the set of recipients comprises:
determining a first time when a treatment was received by a particular anonymized patient;
determining a second time when a notification for the treatment was received by a particular recipient corresponding to the particular anonymized patient; and
determining a time relationship that represents a time length between the first time and the second time.
6. The method of claim 1 , comprising:
determining for a particular patient that a first notification for the treatment was received by the particular patient and a second notification for the treatment was received by the particular patient.
7. The method of claim 1 , wherein computing, based on the set of coefficients and for each notification included in the set of notifications, a score representing a likelihood that the notification being provided to the recipient impacted the corresponding treatment being received by the anonymized patient comprises:
generating an impact model by applying a random survival forest analysis to the set of coefficients.
8. The method of claim 7 , wherein the impact model represents an impact of a notification on a treatment being received exponentially decayed from time when the notification was initially provided to a patient to a time the treatment was received by the patient.
9. The method of claim 1 , further comprising:
determining an impact for a first notification to a recipient through a first type of notification based on a time relationship for the first notification; and
determining an impact for a second notification to the recipient through a second type of notification based on a time relationship for the second notification.
10. The method of claim 1 , further comprising:
aggregating impacts of a particular type of notification on a treatment being received over each anonymized patient included in the set of anonymized patients.
11. The method of claim 2 , wherein determining a forecast model based on scores computed for the set of notifications, where the forecast model stores, for each type of notification from among the set of notifications, a notification type label that is indicative of a relationship between a number of anonymized patients receiving the treatment and a number of recipients receiving notifications for the treatment comprises:
determining, for each type of notification from among the set of notifications, a curve fitting a number of anonymized patients that previously received the treatment with a number of notifications previously provided to the set of recipients.
12. The method of claim 2 , wherein determining a forecast model based on scores computed for the set of notifications, where the forecast model stores, for each type of notification from among the set of notifications, a notification type label that is indicative of a relationship between a number of anonymized patients receiving the treatment and a number of recipients receiving notifications for the treatment comprises:
determining the forecast model as a sigmoid function representing a diminishing effect of a number of notifications received by the set of recipients on a number of anonymized patients that receive the treatment.
13. The method of claim 2 , wherein determining a forecast model based on scores computed for the set of notifications, where the forecast model stores, for each type of notification from among the set of notifications, a notification type label that is indicative of a relationship between a number of anonymized patients receiving the treatment and a number of recipients receiving notifications for the treatment comprises:
determining an initial forecast model for model patients based on the scores indicated by the impact model for the set of notifications; and
projecting the initial forecast model to potential patients.
14. The method of claim 1 , wherein determining the notification plan for notifying potential patients of treatments based on the scores for the notifications included in the set of notifications comprises:
receiving notification constraints for the notification plan;
determining that increasing a number of notifications for a particular channel increases a number of forecasted treatments to be received by anonymized patients included in the set of anonymized patients and satisfies the notification constraints; and
determining a total number of notifications of each type that increases the number of forecasted treatments to be received by anonymized patients included in the set of anonymized patients.
15. The method of claim 3 , wherein scheduling the notifications based on the notification plan comprises:
providing notifications for the treatment to the set of recipients.
16. The method of claim 1 , wherein each anonymized patient included in the set of anonymized patients comprises a patient for which a patient identity cannot be determined but is distinguishable from other anonymized patients.
17. A computer system comprising one or more processors, configured to perform operations comprising:
identifying a set of notifications, wherein:
a recipient received a notification, an anonymized patient received a treatment,
the recipient corresponds to the anonymized patient based on (i) a set of first identifiers for a set of anonymized patients and (ii) a set of second identifiers for a set of recipients;
determining a set of time relationships between a set of treatments that were received by the set of anonymized patients and the set of notifications;
training a time decay model to identify, for each time relationship in the set of time relationships, a set of coefficients for a decay function representing an impact of the notification on the corresponding treatment relative to the time relationship;
computing, based on the set of coefficients and for each notification in the set of notifications, a score representing a likelihood that the notification being provided to the recipient impacted the corresponding treatment being received by the anonymized patient;
determining a notification plan for notifying potential patients of treatments based on the scores for each notification included in the set of notifications, wherein the notification plan is determined to increase a number of treatments to be subsequently received by the anonymized patients; and
scheduling the notifications based on the notification plan.
18. A computer-readable medium, comprising software instructions, which when executed by a processor of a computer, causes the computer to perform operations comprising:
identifying a set of notifications, wherein:
a recipient received a notification,
an anonymized patient received a treatment,
the recipient corresponds to the anonymized patient based on (i) a set of first identifiers for a set of anonymized patients and (ii) a set of second identifiers for a set of recipients;
determining a set of time relationships between a set of treatments that were received by the set of anonymized patients and the set of notifications;
training a time decay model to identify, for each time relationship in the set of time relationships, a set of coefficients for a decay function representing an impact of the notification on the corresponding treatment relative to the time relationship;
computing, based on the set of coefficients and for each notification in the set of notifications, a score representing a likelihood that the notification being provided to the recipient impacted the corresponding treatment being received by the anonymized patient;
determining a notification plan for notifying potential patients of treatments based on the scores for each notification included in the set of notifications, wherein the notification plan is determined to increase a number of treatments to be subsequently received by the anonymized patients; and
scheduling the notifications based on the notification plan.