Forecasting arterial embolic and bleeding events
Systems, methods and computer-readable media are provided for monitoring patients and quantitatively predicting whether an event, such as a significant change in health status meriting intervention, is likely to occur within a future time interval subsequent to computing the prediction. Medical data for a patient is collected from one or more different inputs and used to determine time series data. From this, a forecasted numerical value is computed for one or more physiologic parameters associated with the patient, which may be used to further monitor the patient and facilitate decision making about a need for intensified monitoring or intervention to prevent or manage deterioration of hemostasis. An evolutionary algorithm, such as particle swarm optimization and/or differential evolution, may be used to determine the most probable value of the one or more physiologic parameters at one or more future times.
1 . One or more non-transitory media having computer-readable instructions that, when executed by one or more hardware processors (OMHPs), cause the OMHPs to perform a plurality of operations for preventing or managing clinical deterioration associated with at least one patient, the operations comprising:
generating a patient forecast via the OMHPs, wherein generating the patient forecast via the OMHPs comprises:
collecting physiologic data from a set of sensors that are in proximity to the at least one patient and that are configured to generate time-stamped physiological measurements relating to the at least one patient;
receiving the physiologic data as a set of patient data based on collecting the physiologic data, the received set of patient data comprising the time-stamped physiological measurements;
generating a time series, that corresponds to the received set of patient data, in a form of a Hankel matrix; and
producing a patient forecast using a set of evolutionary algorithms comprising at least one of particle swarm optimization or differential evolution, wherein:
noise is removed from the time series by utilizing the set of evolutionary algorithms to evaluate an objective function of the Hankel matrix comprising noise-compensating error terms, and
the patient forecast comprises a predicted physiologic parameter for the at least one patient at a future time based on the time series;
comparing the patient forecast with control limit information associated with the physiologic data;
determining, in response to the comparing, whether the patient forecast meets a condition associated with the control limit information; and
based on the predicted physiologic parameter and based further on the patient forecast meeting the condition, generating a set of instructions to present via an electronic user interface indicating notification content selected from a group comprising modifying a treatment program associated with the at least one patient and preventing a future occurrence corresponding to the patient forecast.
2 . The one or more non-transitory media of claim 1 , wherein the received set of patient data corresponds to measurements that were automatically determined using the set of sensors in proximity to the at least one patient.
3 . The one or more non-transitory media of claim 1 , wherein the received set of patient data corresponds to measurements obtained via one or more hemostasis sensors.
4 . The one or more non-transitory media of claim 1 , wherein the operations further comprise generating the time series using a plurality of serial measurements over time, the plurality of serial measurements represented via the OMHPs in the form of the Hankel matrix and comprising numerical heart rate data and systolic blood pressure data.
5 . The one or more non-transitory media of claim 1 , wherein the operations further comprise combining at least two values of a physiologic variable or at least two values of a composite variable for generating the patient forecast, and wherein the combining comprises an exponentially weighted moving average of the at least two values of the physiologic variable or the at least two values of the composite variable.
6 . The one or more non-transitory media of claim 1 , wherein the operations further comprise combining at least two values of a physiologic variable or at least two values of a composite variable for generating the patient forecast, and wherein the combining comprises a maximum of the at least two values of the physiologic variable or the at least two values of the composite variable.
7 . The one or more non-transitory media of claim 1 , wherein the operations further comprise providing an alert for the at least one patient to contact a health-care provider based on the patient forecast.
8 . The one or more non-transitory media of claim 1 , wherein the operations further comprise: based on the predicted physiologic parameter, increasing an intensity of the collecting, the receiving, or a combination of the collecting and the receiving.
9 . The one or more non-transitory media of claim 1 , wherein: based on presenting the set of instructions, the treatment program is modified for the at least one patient to prevent the future occurrence corresponding to the patient forecast.
10 . The one or more non-transitory media of claim 9 , wherein the operations further comprise: after the presenting, receiving a signal via the electronic user interface indicating an effectiveness of the treatment program following the modifying in preventing the future occurrence corresponding to the patient forecast.
11 . The one or more non-transitory media of claim 1 , wherein the operations further comprise: based on the patient forecast meeting the condition, increasing a frequency of the collecting, of the receiving, or of a combination of the collecting and the receiving.
12 . A computer-implemented method for preventing or managing clinical deterioration associated with at least one patient, the computer-implemented method performed by one or more hardware processors (OMHPs) and comprising:
generating a patient forecast via the OMHPs, wherein generating the patient forecast via the OMHPs comprises:
collecting physiologic data from a set of sensors that are in proximity to the at least one patient and that are configured to generate time-stamped physiological measurements relating to the at least one patient;
receiving the physiologic data as a set of patient data based on collecting the physiologic data, the received set of patient data comprising the time-stamped physiological measurements;
generating a time series, that corresponds to the received set of patient data, in a form of a Hankel matrix; and
producing a patient forecast using a set of evolutionary algorithms comprising at least one of particle swarm optimization or differential evolution, wherein:
noise is removed from the time series by utilizing the set of evolutionary algorithms to evaluate an objective function of the Hankel matrix comprising noise-compensating error terms, and
the patient forecast comprises a predicted physiologic parameter for the at least one patient at a future time based on the time series;
comparing the patient forecast with control limit information associated with the physiologic data;
determining, in response to the comparing, whether the patient forecast meets a condition associated with the control limit information; and
based on the predicted physiologic parameter and based further on the patient forecast meeting the condition, generating a set of instructions to present via an electronic user interface indicating notification content selected from a group comprising modifying a treatment program associated with the at least one patient and preventing a future occurrence corresponding to the patient forecast.
13 . The computer-implemented method of claim 12 , wherein the time series corresponds to a period of time in which the at least one patient was taking anticoagulants.
14 . The computer-implemented method of claim 12 , further comprising generating a second predicted physiologic parameter at a different future time and determining that information associated with the second predicted physiologic parameter does not meet the condition.
15 . The computer-implemented method of claim 12 , wherein generating the patient forecast comprises using a determined value of a composite variable at a plurality of future time points.
16 . The computer-implemented method of claim 12 , further comprising determining a length of the time series based on a variable of the predicted physiologic parameter and a frequency at which the variable is measured.
17 . A patient monitoring system having one or more hardware processors (OMHPs) configured to perform a plurality of operations for preventing or managing clinical deterioration associated with at least one patient, the operations comprising:
generating a patient forecast via the OMHPs, wherein generating the patient forecast via the OMHPs comprises:
collecting physiologic data from a set of sensors that are in proximity to the at least one patient and that are configured to generate time-stamped physiological measurements relating to the at least one patient;
receiving the physiologic data as a set of patient data based on collecting the physiologic data, the received set of patient data comprising the time-stamped physiological measurements;
generating a time series, that corresponds to the received set of patient data, in a form of a Hankel matrix; and
producing a patient forecast using a set of evolutionary algorithms comprising at least one of particle swarm optimization or differential evolution, wherein:
noise is removed from the time series by utilizing the set of evolutionary algorithms to evaluate an objective function of the Hankel matrix comprising noise- compensating error terms, and
the patient forecast comprises a predicted physiologic parameter for the at least one patient at a future time based on the time series;
comparing the patient forecast with control limit information associated with the physiologic data;
determining, in response to the comparing, whether the patient forecast meets a condition associated with the control limit information; and
based on the predicted physiologic parameter and based further on the patient forecast meeting the condition, generating a set of instructions to present via an electronic user interface indicating notification content selected from a group comprising modifying a treatment program associated with the at least one patient and preventing a future occurrence corresponding to the patient forecast.
18 . The patient monitoring system of claim 17 , wherein not being within a control limit range associated with the control limit information is indicative of hemodynamic deterioration.
19 . The patient monitoring system of claim 17 , wherein not being within a control limit range corresponding to the control limit information is indicative of acute deterioration.
20 . The patient monitoring system of claim 17 , wherein the received set of patient data comprises measurement information selected from a group comprising heart rate data, blood pressure data, and oxygen saturation data.
21 . The patient monitoring system of claim 17 , wherein generating the patient forecast comprises using an optimization fitness objective function.
22 . The patient monitoring system of claim 17 , wherein the set of evolutionary algorithms comprises a first differential evolutionary algorithm and a second differential evolutionary algorithm, and wherein the first differential evolutionary algorithm differs from the second differential evolutionary algorithm.