Pre-emptive chronic obstructive pulmonary disease risk notifications based on medicament device monitoring
A system and method for providing chronic obstructive pulmonary disease (COPD) exacerbation risk notifications in real time or near-real time is described. Rescue and controller medication events are detected by sensors associated with the patient's medicament device/s, and provide a basis to determine to rescue and controller medication use trends for the patient. This data is analyzed to determine the patient's risk for COPD exacerbation after each event, and is used to send notifications to one or both of the patient and their health care provider.
1. A computer-executed method comprising:
receiving, by a remote server over a network, a plurality of rescue medication events from a plurality of client computing devices, each client computing device associated with a patient of a plurality of patients, wherein the plurality rescue medication events are detected by a sensor of a medicament device associated with each patient of the plurality of patients during a first time window;
storing, in a storage device of the remote server, the plurality of rescue medication events in association with a patient profile of each patient;
receiving, by the remote server over the network, a plurality of current rescue medication events at the remote server, wherein the current rescue medication events are detected by the sensor of the medicament device during a second time window occurring later in time than the first time window, wherein the plurality of current rescue medication events include a geographical location of the plurality of current rescue medication events recorded detected by the sensor;
responsive to receiving the plurality of current rescue medication events, applying, at the remote server, a machine-learned model to the plurality of rescue medication events and the current plurality of current rescue medication events together to determine a real-time risk of an outbreak of rescue medication usage for a geographic region; and
transmitting, by the remote server over the network, a real-time notification to the plurality of client computing devices associated with each patient, the real-time notification comprising the outbreak of rescue medication usage and information describing the one or more nearby rescue medication events in the geographical region.
2. The method of claim 1 , applying, at the remote server, a machine-learned model to the plurality of rescue medication events and the plurality of current rescue medication events comprises:
determining a rescue medication baseline based on at least one historical rescue medication event for a plurality of patients in the geographic region;
determining a rescue medication trend by comparing the rescue medication baseline to a count of the current rescue medication events received during the second time window; and
determining the real-time risk of the outbreak of rescue medication usage based on the rescue medication trend.
3. The method of claim 1 , wherein the real-time notification comprises at least one of a numerical risk of outbreak of rescue medication usage, and a categorical risk of outbreak of rescue medication usage.
4. The method of claim 1 , wherein the patient profile of each patient of the plurality of patients comprises an identifier of the client computing device associated with each patient.
5. The method of claim 1 , wherein the patient profile of each patient of the plurality of patients comprises a type of controller medication and a type of rescue medication prescribed to the first patient.
6. The method of claim 2 , wherein determining the rescue medication baseline comprises: summing a count of the historical rescue medication events occurring during the second time window; and dividing the count of the rescue medication events occurring during the second time window by a duration of the second time window.
7. The method of claim 5 , wherein determining the rescue medication trend comprises: labeling the patient as at risk of rescue medication usage responsive to a function of the count of current rescue medication events received during the second time window and the rescue medication baseline being above a threshold value; and labeling the patient as not at risk of rescue medication usage responsive to the function of the count of current rescue medication events received during the second time window and the rescue medication baseline being below the threshold value.
8. The method of claim 1 , further comprising:
storing, in the storage device of the remote server, a controller medication adherence plan for each patient;
receiving, over a third time window occurring earlier in time than the second time window, at least one historical controller medication event from the client computing device associated with each patient;
determining controller medication adherence based on the at least one historical controller medications event and the controller medication adherence plan; and
determining the real-time risk of rescue medication usage further based on the controller medication adherence.
9. The method of claim 8 , wherein determining controller medication adherence comprises: dividing a count of the historical controller medication events occurring during the first time window by a number of controller medication doses prescribed in the controller medication adherence plan.
10. The method of claim 8 , wherein determining the real-time risk of outbreak of rescue medication usage comprises: determining a numerical risk score based on the controller medication adherence and the rescue medication trend, wherein the controller medication adherence and the rescue medication trend are weighted relative to each other such that the rescue medication trend is more determinative of the numerical risk score than the controller medication adherence.
11. A non-transitory computer-readable storage medium comprising computer program instructions that, when executed by a computer processor, cause the computer processor to:
receive, by a remote server over a network, a plurality of rescue medication events from a plurality of client computing devices, each client computing device associated with of a plurality of patients, wherein the plurality of rescue medication events are detected by a sensor of a medicament device associated with each patient during a first time window;
store, in a storage device of the remote server, the plurality of rescue medication events in association with a patient profile of each patient;
receive, by the remote server over the network, a plurality of current rescue medication events at the remote server, wherein the current rescue medication events are detected by the sensor of the medicament device during a second time window occurring later in time than the first time window, wherein the plurality of current rescue medication events include a geographical location of the plurality of current rescue medication events recorded detected by the sensor;
responsive to receiving the plurality of current rescue medication events, apply, at the remote server, a machine-learned model to the plurality of rescue medication events and the current plurality of current rescue medication events together to determine a real-time risk of an outbreak of rescue medication usage for a geographic region; and
transmit, by the remote server over the network, a real-time notification to the plurality of client computing devices associated with each patient, the real-time notification comprising an outbreak of rescue medication usage and information describing the one or more nearby rescue medication events in the geographical region.
12. The storage medium of claim 11 , wherein the instructions being configured to apply, at the remote server, a machine-learned model to the plurality of rescue medication events and the plurality of current rescue medication events comprises:
determining a rescue medication baseline based on at least one historical rescue medication event for a plurality of patients in the geographic region;
determining a rescue medication trend by comparing the rescue medication baseline to a count of the current rescue medication events received during the second time window; and
determining the real-time risk of the outbreak of rescue medication usage based on the rescue medication trend.
13. The storage medium of claim 11 , wherein the real-time notification comprises at least one of a numerical risk of outbreak of rescue medication usage, and a categorical risk of outbreak of rescue medication usage.
14. The storage medium of claim 11 , wherein the patient profile of each patient comprises an identifier of the client computing device associated with each patient of the plurality of patients.
15. The storage medium of claim 11 , wherein the patient profile of the first patient comprises a type of controller medication and a type of rescue medication prescribed to each patient of the plurality of patients.
16. The storage medium of claim 12 , wherein the instructions are further configured to:
store, in the storage device of the remote server a controller medication adherence plan for each patient;
receive, over a third time window occurring earlier in time than the second time window, at least one historical controller medication event from the client computing device associated with each patient;
determine controller medication adherence based on the at least one historical controller medications event and the controller medication adherence plan; and
determine the real-time risk of rescue medication usage further based on the controller medication adherence.
17. The storage medium of claim 16 , wherein the instructions being configured to determine controller medication adherence comprises: dividing a count of the historical controller medication events occurring during the first time window by a number of controller medication doses prescribed in the controller medication adherence plan.
18. The storage medium of claim 16 , wherein the instructions being configured to determine the real-time risk of outbreak of rescue medication usage comprises: determining a numerical risk score based on the controller medication adherence and the rescue medication trend, wherein the controller medication adherence and the rescue medication trend are weighted relative to each other such that the rescue medication trend is more determinative of the numerical risk score than the controller medication adherence.
19. The storage medium of claim 16 , wherein the instructions being configured to determine the rescue medication baseline comprises: summing a count of the historical rescue medication events occurring during the second time window; and dividing the count of the rescue medication events occurring during the second time window by a duration of the second time window.
20. The storage medium of claim 12 , wherein the instructions being configured to determine the rescue medication trend comprises: labeling the patient as at risk of rescue medication usage responsive to a function of the count of current rescue medication events received during the second time window and the rescue medication baseline being above a threshold value; and labeling the patient as not at risk of rescue medication usage responsive to the function of the count of current rescue medication events received during the second time window and the rescue medication baseline being below the threshold value.