Managing medical events using visual patterns generated from multivariate medical records

1. A system for managing medical events, comprising:

at least one medical sensor, configured to monitor one or more patients and output a continuous data stream comprising a plurality of sensor measurements;

at least one display;

at least one hardware processor connected to said at least one medical sensor and said at least one display, and adapted to:

receive a plurality of events, each having a time of occurrence, comprising said plurality of sensor measurements and a plurality of external events pertaining to said one or more patients and received from an external database;

identify a target sequence of target events in said plurality of events, wherein the target sequence comprises a first target event followed by a second target event, the first target event comprising an administration of a medication and the second target event comprising a medical occurrence;

identify in said plurality of events a plurality of matching sequences, each comprising the first target event followed by the second target event;

augment each of said plurality of matching sequences with some temporally related events of said plurality of events;

cluster said plurality of augmented matching sequences in a plurality of event clusters, wherein a portion of the plurality of augmented matching sequences are clustered according to a temporal distribution of events of each augmented matching sequence in a plurality of time intervals according to an event's time of occurrence relative to a time of occurrence of said matching sequence, wherein the plurality of time intervals comprise different durations, wherein a duration of a first time interval closer to a target event of the target sequence is smaller than a duration of a second time interval further from the target event;

display on said at least one display a visual representation of said plurality of event clusters according to a predefined set of similarity criteria, wherein the predefined set of similarity criteria comprises the temporal distribution of events and at least one additional similarity criterion, wherein the display is populated with two or more cluster selection tools, and wherein selecting a first cluster selection tool displays a first cluster pattern comprising the temporal distribution of events, and selecting a second cluster selection tool displays a second cluster pattern comprising an additional similarity criterion of the at least one additional similarity criterion; and

wherein a first medication is associated with a matching sequence and a second medication is not associated with any matching sequences.

2. The system of claim 1 , wherein said at least one hardware processor is adapted to cluster said plurality of augmented matching sequences by:

assigning said sequence time of occurrence to each augmented matching sequence of said plurality of augmented matching sequences according to a time of occurrence of at least one event in said plurality of augmented matching sequences;

computing for each event in each augmented matching sequence of said plurality of augmented matching sequences a relative time of occurrence, relative to said sequence time of occurrence of plurality of augmented matching sequences;

sorting, in each of said plurality of augmented matching sequences, said matching sequence's plurality of events into said plurality of time intervals according to said relative time of occurrence; and

clustering the plurality of augmented matching sequences into a plurality of event clusters according to a predefined similarity test applied to a distribution of said plurality of augmented matching sequence's events in said plurality of time intervals.

3. The system of claim 1 , wherein each of said plurality of events has a plurality of values of a plurality of attributes selected from a group of attributes consisting of: a patient identifier, a day of week, a time of day, said medical occurrence, a measured blood sugar value, a measured blood pressure value, a medication identifier, and a medication dosage.

4. The system of claim 1 , wherein said medical occurrence is one of: a blood sugar value exceeding a threshold high blood sugar value, a blood sugar value less than a threshold low blood sugar value, a blood pressure value exceeding a threshold high blood pressure value, a blood pressure value less than a threshold low blood pressure value, a blood sugar value greater than a previous blood sugar value, a blood sugar value less than a previous blood sugar value, a blood pressure value greater than a previous blood pressure value, and a blood pressure value less than a previous blood pressure value.

5. The system of claim 3 , wherein said at least one identified matching criterion comprises one or more of a group consisting of: a patient identifier of an event of said sequence of events is equal to another patient identifier of a target event of said target sequence, a day of week of an event of said sequence of events is equal to another day of week of a target event of said target sequence, a time of day of an event of said sequence of events is equal to another time of day of a target event of said target sequence, and a medical occurrence of an event of said sequence of events is equal to another medical occurrence of a target event of said target sequence.

6. The system of claim 2 , wherein at least one time interval of said plurality of time intervals is shorter than at least one other time interval of said plurality of time intervals; and

wherein a first plurality of absolute values of a first plurality of times in said at least one other time interval are greater than a second plurality of absolute values of a second plurality of times in said at least one time interval.

7. The system of claim 1 , wherein said at least one hardware processor is further adapted to produce a report of said plurality of event clusters.

8. The system of claim 1 , wherein said at least one hardware processor is further adapted to:

execute an electronic medical record (EMR) management system for storing a plurality of events comprising one or more sensor measurements;

store said plurality of sensor measurements in said EMR management system; and

retrieve said plurality of sensor measurements from said EMR management system.

9. A method for managing medical events, comprising:

receiving a plurality of events, each having a time of occurrence, comprising a plurality of sensor measurements received from at least one medical sensor and a plurality of external events pertaining to one or more patients and received from an external database;

identifying a target sequence of target events in said plurality of events, wherein the target sequence comprises a first target event followed by a second target event, the first target event comprising an administration of a medication and the second target event comprising a medical occurrence;

identifying in said plurality of events a plurality of matching sequences, each comprising the first target event followed by the second target event, wherein a first medication is associated with a matching sequence and a second medication is not associated with any matching sequences;

augmenting each of said plurality of matching sequences with some temporally related events of said plurality of events;

clustering said plurality of augmented matching sequences in a plurality of event clusters, wherein a portion of the plurality of augmented matching sequences are clustered according to a temporal distribution of events of each augmented matching sequence in a plurality of time intervals according to an event's time of occurrence relative to a time of occurrence of said matching sequence, wherein the plurality of time intervals comprise different durations, wherein a duration of a first time interval closer to a target event of the target sequence is smaller than a duration of a second time interval further from the target event;

displaying on at least one display a visual representation of said plurality of event clusters according to a predefined set of similarity criteria, wherein the predefined set of similarity criteria comprises the temporal distribution of events and at least one additional similarity criterion, wherein the display is populated with two or more cluster selection tools, and wherein selecting a first cluster selection tool displays a first cluster pattern comprising the temporal distribution of events, and selecting a second cluster selection tool displays a second cluster pattern comprising an additional similarity criterion of the at least one additional similarity criterion; and

identifying the second medication for a patient over the first medication in response to the first medication being associated with a matching sequence and the second medication not being associated with any matching sequences.

10. The method of claim 9 , wherein said visual representation comprises: for each of one or more of said plurality of event clusters displaying in a common display area of said at least one display a cluster visual representation of some of said event cluster's plurality of events.

11. The method of claim 10 , further comprising:

receiving an instruction from a user to expand one of said plurality of event clusters; and

displaying some more of said plurality of events of at least one of said plurality of event clusters.

12. The method of claim 10 , wherein for each cluster of said plurality of event clusters said cluster visual representation comprises displaying a first graphic object for a first plurality of events of a first of said plurality of augmented matching sequences of said cluster, and a second graphic object for a second plurality of events of a second of said plurality of augmented matching sequences of said cluster; and

wherein said cluster visual representation is displayed on a graph having one axis representing a time according to said relative time of occurrence of each of said plurality of events.