Accurate temporal event predictive modeling

1. A method for accurate temporal event predictive modeling by a processor, comprising:

monitoring medical data of a patient over a plurality of time-series windows;

detecting, based on the monitoring, one or more time-series events for the patient, each time-series event occurring within a time-series window, each time-series event being associated with a medical condition of the patient;

determining a reverse event delay for each of the plurality of time-series windows,

wherein the reverse event delay is an amount of time between the end of the time-series window and an occurrence of a corresponding time-series event;

setting, in response to determining a time-series event did not occur in a particular time-series window, a reverse event delay associated with the particular time-series window as zero for determining the average reverse event delay;

determining an average reverse event delay from the plurality of reverse event delays;

training a predictive model using the average reverse event delay and an average event delay, wherein the average event delay is an average time delay from the start of the time-series window until occurrence of a time-series event; and

predicting a future time-series event for the patient by applying the average reverse event delay in conjunction with one or more weighted factors in the predictive model.

2. The method of claim 1 , further including identifying the time-series window as a selected time period or one or more events.

3. The method of claim 1 , further including identifying an event delay in the time-series window as a time delay starting at a beginning of the time-series window until occurrence of an event.

4. The method of claim 1 , further including:

predicting a hypoglycemic event, as the time-series event, of the patient based on the learned predictive model.

5. A system for accurate temporal event predictive modeling,

comprising:

a processor configured to perform a method comprising:

monitoring medical data of a patient over a plurality of time-series windows;

detecting, based on the monitoring, one or more time-series events for the patient, each time-series event occurring within a time-series window, each time-series event being associated with a medical condition of the patient;

determining a reverse event delay for each of the plurality of time-series windows,

wherein the reverse event delay is an amount of time between the end of the time-series window and an occurrence of a corresponding time-series event;

setting, in response to determining a time-series event did not occur in a particular time-series window, a reverse event delay associated with the particular time-series window as zero for determining the average reverse event delay;

determining an average reverse event delay from the plurality of reverse event delays;

training a predictive model using the average reverse event delay and an average event delay, wherein the average event delay is an average time delay from the start of the time-series window until occurrence of a time-series event; and

predicting a future time-series event for the patient by applying the average reverse event delay in conjunction with one or more weighted factors in the predictive model.

6. The system of claim 5 , wherein the method further comprises identifying the time-series window as a selected time period or one or more events.

7. The system of claim 5 , wherein the method further comprises identifying an event delay in the time-series window as a time delay starting at a beginning of the time-series window until occurrence of an event.

8. The system of claim 5 , wherein the method further comprises:

predicting a hypoglycemic event, as the time-series event, of the patient based on the learned predictive model.

9. A computer program product for accurate temporal event predictive modeling by a processor, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by processor to cause the processor to perform a method comprising:

analyzing medical data for a patient, wherein the medical data includes data for a plurality of time-series windows;

identifying, based on the analyzing, one or more time-series events for the patient, each of the one or more time-series events occurring within a time-series window;

determining a reverse event delay for each of the time-series windows, wherein the reverse event delay for a respective time-series window is an amount of time between an occurrence of a time-series event in the respective time-series window and the end of the respective time-series window;

setting, in response to determining a time-series event did not occur in a particular time-series window, a reverse event delay associated with the particular time-series window as zero for determining the average reverse event delay;

determining an average reverse event delay from the plurality of reverse event delays;

training a predictive model using the average reverse event delay and an average event delay, wherein the average event delay is an average time delay from the start of the time-series window until occurrence of a time-series event; and

predicting a future time-series event for the patient by applying the average reverse event delay in conjunction with one or more weighted factors in the predictive model.

10. The computer program product of claim 9 , wherein the method further comprises identifying each of the plurality of time-series windows as a period between an occurrence of two or more events.

11. The computer program product of claim 9 , wherein the method further comprises identifying:

an event delay in each of the plurality of time-series windows, wherein the event delay for each respective time-series window is the time delay starting at a beginning of the respective time-series window until occurrence of a time-series event in the respective time-series window.

12. The computer program product of claim 9 , wherein the method further comprises:

predicting a hypoglycemic event, as the time-series event, of the patient based on the learned predictive model.

13. The computer program product of claim 9 , wherein identifying the reverse event delay for each of the plurality of time-series windows comprises:

identifying a first time-series window, wherein the first time-series window is a window that starts at an occurrence of a first event and ends a predetermined amount of time later;

determining whether a first time-series event occurs within the first time-series window;

if the first time-series event occurs within the first time-series window, calculating a first reverse event delay of the plurality of reverse event delays as an amount of time between an occurrence of the first time-series event and the end of the first time-series window; and

if the first time-series event does not occur within the first time-series window, setting the first reverse event delay as zero.

14. The computer program product of claim 13 , wherein the plurality of reverse event delays further includes a second reverse event delay associated with a second time-series window that starts at an occurrence of a second event, and wherein determining the average reverse event delay from the plurality of reverse event delays comprises determining an average of the first and second reverse event delays.

15. The computer program product of claim 14 , wherein the first event is a first bolus event and the second event is a second bolus event, wherein the first time-series event is a hypoglycemic event, and wherein determining whether the first time-series event occurs within the first time-series window includes:

determining whether a blood glucose level of the patient drops below 70 mg/dl during the first time-series window;

if the blood glucose level of the patient drops below 70 mg/dl during the first time-series window, determining that the first time-series event occurred; and

if the blood glucose level of the patient does not drop below 70 mg/dl during the first time-series window, determining that the first time-series event did not occur.