MODEL-BASED ADVANCE PREDICTION OF ADVERSE EVENTS USING MONITORING DEVICE DATA
In some embodiments, a model (such as a formulaic model or machine-learning model) is used to predict adverse patient events. The model is developed based on patient data derived by monitoring patients of a patient population over a period of time, such as one year. Patient parameters, such as heart rate, are obtained from the patient data. Selected patient parameters and verified occurrences of the adverse patient events are used to develop the model to have a specified sensitivity and/or specificity. Key patient parameters, determined through analysis of the patient to have higher predictive power, are used to increase the sensitivity and/or specificity of the model. Once developed, new patients are monitored and corresponding patient parameters are obtained for the new patients. The model is then able to predict if one of the new patients will have an adverse patient event.
1 . A method of predicting the onset of a condition comprising:
gathering, using a nighttime monitoring device and over a plurality of nights, respective nighttime monitoring data of a body, wherein the nighttime monitoring device operates using signals indicative of movements of the body, the signals affected by a reactive near-field coupling of the body to a generated electric field generated by the nighttime monitoring device;
computing, on a daily basis, from the respective nighttime monitoring data, a respective plurality of nighttime physiological parameters, wherein at least one of the respective plurality of nighttime physiological parameters is computed over a window of a plurality of sequential nights ending on a current night;
determining, on a daily basis, from the respective plurality of nighttime physiological parameters, a respective daily probability of the occurrence of a condition;
determining, on a daily basis from the respective daily probabilities over a window of at least three sequential days ending on a current day immediately following the current night, an improved prediction of the occurrence of the condition; and
transmitting, over a network, an indication of a prediction of the condition according to the improved prediction of the occurrence of the condition.
2 . The method of claim 1 , wherein the nighttime monitoring device is a passive, nighttime monitoring device.
3 . The method of claim 2 , wherein the passive, nighttime monitoring device comprises an electric field generator operating at a nominal frequency, and circuitry to compute one or more properties of the generated electric field, the one or more properties of the generated electric field changing over time due to interactions with the body in the reactive near-field region of the generated electric field.
4 . The method of claim 1 , further comprising:
computing a first one or more of the respective plurality of nighttime physiological parameters based at least in part on one or more periodic behaviors determined from the signals indicative of movements of the body, and
computing a second one or more of the respective plurality of nighttime physiological parameters based at least in part on one or more non-periodic behaviors determined from the signals indicative of movements of the body.
5 . The method of claim 4 , wherein the one or more periodic behaviors comprise a heart rate metric and a respiration rate metric, and wherein the one or more non-periodic behaviors comprise a movement metric of the body.
6 . The method of claim 1 , wherein the respective nighttime monitoring data comprises a heart waveform superimposed on a respiratory waveform and the respective plurality of nighttime physiological parameters comprises the heart waveform and the respiratory waveform.
7 . The method of claim 1 , wherein the determining the improved prediction of the occurrence of the condition is according to a sum of the respective daily probabilities in the window, wherein the window comprises at least three sequential days ending on the current day, and wherein the transmitted indication of the prediction of the condition is according to the sum exceeding a determined threshold.
8 . The method of claim 1 , wherein the window of at least three sequential days ending on the current day comprises at least nine sequential days ending on the current day.
9 . The method of claim 8 , wherein the improved prediction of the occurrence of the condition is an advance prediction of a Chronic Obstructive Pulmonary Disease (COPD) exacerbation event.
10 . The method of claim 9 , wherein the advance prediction of the COPD exacerbation event predicts an occurrence of a COPD exacerbation event at least five days in advance.
11 . The method of claim 1 , wherein the respective plurality of nighttime physiological parameters comprises a heart rate volatility computed over a window of at least seven sequential nights ending on the current night and a respiration rate volatility computed over a window of at least four sequential nights ending on the current night.
12 . The method of claim 1 , wherein the window of the plurality of sequential nights ending on the current night begins seventeen nights prior to the current night.
13 . The method of claim 1 , wherein at least one of the respective plurality of nighttime physiological parameters is computed based at least in part on baseline data of the body; and wherein the baseline data of the body comprises baseline values of one or more of the nighttime physiological parameters of the body.
14 . The method of claim 13 , further comprising:
determining the baseline data of the body from a plurality of preceding nights of the respective nighttime monitoring data, wherein each of the plurality of preceding nights is prior to any night in the window of the plurality of sequential nights ending on the current night, and wherein the plurality of preceding nights is at least seven nights.
15 . The method of claim 14 , wherein the determining the baseline data of the body is repeated periodically.
16 . The method of claim 14 , wherein the determining the baseline data of the body is repeated after a recovery period from an adverse event.
17 . The method of claim 14 , wherein the determining the baseline data of the body comprises computing new baseline data of the body from the plurality of preceding nights of the respective nighttime monitoring data of the body and selectively updating the baseline data of the body with at least some of the new baseline data of the body.
18 . A system for predicting the onset of a condition, the system comprising:
a nighttime monitoring device enabled to monitor a body via a generated electric field generated by the nighttime monitoring device;
a processor; and
a memory device that stores program code structured to cause the processor to:
gather, from the nighttime monitoring device and over a plurality of nights, respective nighttime monitoring data of the body, wherein the nighttime monitoring device operates using signals indicative of movements of the body, the signals affected by a reactive near-field coupling of the body to the generated electric field;
compute, on a daily basis, from the respective nighttime monitoring data, a respective plurality of nighttime physiological parameters, wherein at least one of the respective plurality of nighttime physiological parameters is computed over a window of a plurality of sequential nights ending on a current night;
determine, on a daily basis, from the respective plurality of nighttime physiological parameters, a respective daily probability of the occurrence of a condition;
determine, on a daily basis from the respective daily probabilities over a window of at least three sequential days ending on a current day immediately following the current night, an improved prediction of the occurrence of the condition; and
transmit, over a network, an indication of a prediction of the condition according to the improved prediction of the occurrence of the condition.
19 . The system of claim 18 , wherein the nighttime monitoring device is a passive, nighttime monitoring device.
20 . The system of claim 19 , wherein the passive, nighttime monitoring device comprises an electric field generator to generate the generated electric field at a nominal frequency, and circuitry to compute one or more properties of the generated electric field, the one or more properties of the generated electric field changing over time due to interactions with the body in a reactive near-field region of the generated electric field.
21 . The system of claim 20 , wherein the program code is further structured to cause the processor to:
compute a first one or more of the respective plurality of nighttime physiological parameters based at least in part on one or more periodic behaviors determined from the one or more properties of the generated electric field, and
compute a second one or more of the respective plurality of nighttime physiological parameters based at least in part on one or more non-periodic behaviors determined from the one or more properties of the generated electric field, and wherein the one or more properties of the generated electric field comprise a frequency of the generated electric field.
22 . The system of claim 21 , wherein the one or more periodic behaviors comprise a heart rate metric and a respiration rate metric, and wherein the one or more non-periodic behaviors comprise a movement metric of the body.
23 . The system of claim 18 , wherein the program code is further structured to cause the processor to determine the improved prediction of the occurrence of the condition according to a sum of the respective daily probabilities in the window, wherein the window comprises at least three sequential days ending on the current day, and wherein the transmitted indication of the prediction of the condition is according to the sum exceeding a determined threshold.
24 . The system of claim 18 , wherein the window of at least three sequential days ending on the current day comprises at least nine sequential days ending on the current day.
25 . The system of claim 24 , wherein the improved prediction of the occurrence of the condition is an advance prediction of a Chronic Obstructive Pulmonary Disease (COPD) exacerbation event that predicts an occurrence of a COPD exacerbation event at least five days in advance.
26 . The system of claim 18 , wherein the window of the plurality of sequential nights ending on the current night begins seventeen nights prior to the current night.
27 . The system of claim 18 , wherein at least one of the respective plurality of nighttime physiological parameters is computed based at least in part on baseline data of the body; and wherein the baseline data of the body comprises baseline values of one or more of the nighttime physiological parameters of the body.
28 . The system of claim 27 , wherein the program code is further structured to cause the processor to:
determine the baseline data of the body from a plurality of preceding nights of the respective nighttime monitoring data, wherein each of the plurality of preceding nights is prior to any night in the window of the plurality of sequential nights ending on the current night, and wherein the plurality of preceding nights is at least seven nights.