System and method for timely multi-channel notification of treatment
A computer-assisted method to timely provide notifications of treatments, the method including receiving de-identified longitudinal medical records, each de-identified longitudinal medical record representing a record of a different anonymized patient and encoding information identifying a treatment received by the anonymized patient and receiving notification data including notification records, each notification record encoding information identifying a channel through which the notification was provided. The method includes determining a first channel impact model representing an impact of a notification provided through a first channel on a treatment being received, a second channel impact model representing an impact of a notification provided through a second channel on a treatment being received, and determining a multi-channel impact model representing an impact of notifications being provided through both the first channel and the second channel on a treatment being received.
1. A computer-implemented method comprising:
identifying, from a set of recipients, a first subset of recipients that (i) received a notification through a first channel, (ii) have not received the notification through a second channel, and (iii) have received a treatment, wherein the identifying is based on (i) de-identified medical records indicating treatments received by a set of anonymized patients and (ii) notification records indicating channels through which notifications were provided to the first subset of recipients and a first set of time points when the notifications were received by the first subset of recipients stored in one or more database systems;
identifying, from the set of recipients, a second subset of recipients that (i) have received the notification through the second channel, (ii) have not received the notification through the first channel, and (iii) have received the treatment, wherein the identifying is based on (i) the de-identified medical records indicating treatments received by the set of anonymized patients and (ii) the notification records indicating channels through which notifications were provided to the second subset of recipients and a second set of time points when the notifications were received by the second subset of recipients stored in the one or more database systems;
generating a multi-channel impact model based on a time relationship between when the treatment was received by each patient of the set of anonymized patients and when a notification was received by each recipient of the first subset of recipients and the second subset of recipients, wherein the multi-channel impact model represents an impact of notifications being provided through the first channel and the second channel on a likelihood of one or more recipients in the set of recipients receiving the treatment;
generating a notification plan for timely notifying potential patients of treatments based on the impact represented by the multi-channel impact model, wherein the notification plan indicates one or more notifications to be provided through a corresponding channel to increase a number of treatments to be subsequently received; and
sending the one or more notifications through the first channel or the second channel to respective recipients associated with potential patients based on the notification plan.
2. The method of claim 1 , wherein the first channel comprises a digital channel and the second channel comprises a non-digital channel.
3. The method of claim 1 , further comprising:
generating a first channel impact model based on the first subset of recipients, wherein the first channel impact model represents a first impact of the notification received through the first channel and treatments received by the first subset of recipients; and
generating a second channel impact model based on the second subset of recipients, wherein the second channel impact model represents a second impact of the notification received through the second channel and treatments received by the second subset of recipients.
4. The method of claim 3 , wherein the multi-channel impact model is generated based on the first impact represented by the first channel impact model and the second impact represented by the second channel impact model.
5. The method of claim 3 , wherein generating the first channel impact model comprises:
determining a time decay factor associated with the first impact represented by the first channel impact model; and
determining, based on the time decay factor, an expected time delay before the notification received through the first channel is expected to impact whether treatments are received by the first subset of recipients.
6. The method of claim 1 , wherein each recipient included in the first subset of recipients and the second subset of recipients is associated with a unique identifier that distinguishes a recipient from other recipients.
7. The method of claim 1 , wherein the multi-channel impact model is generated based on:
oversampling data associated with the first subset of recipients and the second subset of recipients; and
undersampling data associated with other recipients not included in either the first subset of recipients and the second subset of recipients.
8. A system comprising:
one or more processors; and
one or more storing devices storing instructions that cause the one or more processors to perform operations comprising:
identifying, from a set of recipients, a first subset of recipients that (i) received a notification through a first channel, (ii) have not received the notification through a second channel, and (iii) have received a treatment, wherein the identifying is based on (i) de-identified medical records indicating treatments received by a set of anonymized patients and (ii) notification records indicating channels through which notifications were provided to the first subset of recipients and a first set of time points when the notifications were received by the first subset of recipients stored in one or more database systems;
identifying, from the set of recipients, a second subset of recipients that (i) have received the notification through the second channel, (ii) have not received the notification through the first channel, and (iii) have received the treatment, wherein the identifying is based on (i) the de-identified medical records indicating treatments received by the set of anonymized patients and (ii) the notification records indicating channels through which notifications were provided to the second subset of recipients and a second set of time points when the notifications were received by the second subset of recipients stored in the one or more database systems;
generating a multi-channel impact model based on a time relationship between when the treatment was received by each patient of the set of anonymized patients and when a notification was received by each recipient of the first subset of recipients and the second subset of recipients, wherein the multi-channel impact model represents an impact of notifications being provided through the first channel and the second channel on a likelihood of one or more recipients in the set of recipients receiving the treatment;
generating a notification plan for timely notifying potential patients of treatments based on the impact represented by the multi-channel impact model, wherein the notification plan indicates one or more notifications to be provided through a corresponding channel to increase a number of treatments to be subsequently received; and
sending the one or more notifications through the first channel or the second channel to respective recipients associated with potential patients based on the notification plan.
9. The system of claim 8 , wherein the first channel comprises a digital channel and the second channel comprises a non-digital channel.
10. The system of claim 8 , wherein the operations comprise:
generating a first channel impact model based on the first subset of recipients, wherein the first channel impact model represents a first impact of the notification received through the first channel and treatments received by the first subset of recipients; and
generating a second channel impact model based on the second subset of recipients, wherein the second channel impact model represents a second impact of the notification received through the second channel and treatments received by the second subset of recipients.
11. The system of claim 10 , wherein the multi-channel impact model is generated based on the first impact represented by the first channel impact model and the second impact represented by the second channel impact model.
12. The system of claim 10 , wherein generating the first channel impact model comprises:
determining a time decay factor associated with the first impact represented by the first channel impact model; and
determining, based on the time decay factor, an expected time delay before the notification received through the first channel is expected to impact whether treatments are received by the first subset of recipients.
13. The system of claim 8 , wherein each recipient included in the first subset of recipients and the second subset of recipients is associated with a unique identifier that distinguishes a recipient from other recipients.
14. The system of claim 8 , wherein the multi-channel impact model is generated based on:
oversampling data associated with the first subset of recipients and the second subset of recipients; and
undersampling data associated with other recipients not included in either the first subset of recipients and the second subset of recipients.
15. At least one non-transitory computer-readable storage device storing instructions that cause one or more processors to perform operations comprising:
identifying, from a set of recipients, a first subset of recipients that (i) received a notification through a first channel, (ii) have not received the notification through a second channel, and (iii) have received a treatment, wherein the identifying is based on (i) de-identified medical records indicating treatments received by a set of anonymized patients and (ii) notification records indicating channels through which notifications were provided to the first subset of recipients and a first set of time points when the notifications were received by the first subset of recipients stored in one or more database systems;
identifying, from the set of recipients, a second subset of recipients that (i) have received the notification through the second channel, (ii) have not received the notification through the first channel, and (iii) have received the treatment, wherein the identifying is based on (i) the de-identified medical records indicating treatments received by the set of anonymized patients and (ii) the notification records indicating channels through which notifications were provided to the second subset of recipients and a second set of time points when the notifications were received by the second subset of recipients stored in the one or more database systems;
generating a multi-channel impact model based on a time relationship between when the treatment was received by each patient of the set of anonymized patients and when a notification was received by each recipient of the first subset of recipients and the second subset of recipients, wherein the multi-channel impact model represents an impact of notifications being provided through the first channel and the second channel on a likelihood of one or more recipients in the set of recipients receiving the treatment;
generating a notification plan for timely notifying potential patients of treatments based on the impact represented by the multi-channel impact model, wherein the notification plan indicates one or more notifications to be provided through a corresponding channel to increase a number of treatments to be subsequently received; and
sending the one or more notifications through the first channel or the second channel to respective recipients associated with potential patients based on the notification plan.
16. The at least one non-transitory computer-readable storage device of claim 15 , wherein the first channel comprises a digital channel and the second channel comprises a non-digital channel.
17. The at least one non-transitory computer-readable storage device of claim 15 , wherein the operations further comprise:
generating a first channel impact model based on the first subset of recipients, wherein the first channel impact model represents a first impact of the notification received through the first channel and treatments received by the first subset of recipients; and
generating a second channel impact model based on the second subset of recipients, wherein the second channel impact model represents a second impact of the notification received through the second channel and treatments received by the second subset of recipients.
18. The at least one non-transitory computer-readable storage device of claim 17 , wherein the multi-channel impact model is generated based on the first impact represented by the first channel impact model and the second impact represented by the second channel impact model.
19. The at least one non-transitory computer-readable storage device of claim 17 , wherein generating the first channel impact model comprises:
determining a time decay factor associated with the first impact represented by the first channel impact model; and
determining, based on the time decay factor, an expected time delay before the notification received through the first channel is expected to impact whether treatments are received by the first subset of recipients.
20. The at least one non-transitory computer-readable storage device of claim 15 , wherein each recipient included in the first subset of recipients and the second subset of recipients is associated with a unique identifier that distinguishes a recipient from other recipients.