Seizure detection methods, apparatus, and systems using an autoregression algorithm

• US 10,206,591 B2
• Filed: 07/20/2012
• Issued: 02/19/2019
• Est. Priority Date: 10/14/2011
• Status: Active Grant

• Current Assignee: Flint Hills Scientific, LLC
• Original Assignee: Flint Hills Scientific, LLC
• Inventors: Osorio, Ivan, Lyubushin, Alexey, Sornette, Didier
• Primary Examiner(s): Sims, Jason M
• Application Number:
  US13/554,694
• Publication Number:
  US 20130096840A1
• Time in Patent Office: 2,405 Days
• Field of Search: None
• CPC Class Codes: A61B 5/0205 Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
• CPC Class Codes: A61B 5/02055 Simultaneously evaluating both cardiovascular condition and temperature
• CPC Class Codes: A61B 5/024 Detecting, measuring or recording pulse rate or heart rate A61B5/0205, A61B5/021 take precedence
• CPC Class Codes: A61B 5/0816 Measuring devices for examining respiratory frequency measuring frequency of electric signals G01R23/00
• CPC Class Codes: A61B 5/0833 Measuring rate of oxygen consumption
• CPC Class Codes: A61B 5/1118 Determining activity level
• CPC Class Codes: A61B 5/14542 for measuring blood gases A61B5/14551 takes precedence
• CPC Class Codes: A61B 5/16 Devices for psychotechnics using teaching or educational appliances G09B1/00 - G09B7/00; Testing reaction times ; Devices for evaluating the psychological state
• CPC Class Codes: A61B 5/165 Evaluating the state of mind, e.g. depression, anxiety
• CPC Class Codes: A61B 5/24 Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
• CPC Class Codes: A61B 5/369 Electroencephalography [EEG] devices for psychotechnics A61B5/16
• CPC Class Codes: A61B 5/374 Detecting the frequency distribution of signals, e.g. detecting delta, theta, alpha, beta or gamma waves
• CPC Class Codes: A61B 5/4094 Diagnosing or monitoring seizure diseases, e.g. epilepsy
• CPC Class Codes: A61B 5/4836 Diagnosis combined with treatment in closed-loop systems or methods A61B5/0036 takes precedence
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/746 Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms alarms per se G08B
• CPC Class Codes: G06F 17/10 Complex mathematical operations function generation by table look-up G06F1/03; evaluation of elementary functions by calculation G06F7/544

Abstract

A method, comprising receiving a time series of patient body signal, determining first and second sliding time windows for the time series; applying an autoregression algorithm, comprising: applying an autoregression analysis to each of the first and second windows, yielding autoregression coefficients and a residual variance for each window; estimating a parameter vector for each window based on the autoregression coefficients and residual variances; and determining a difference between the parameter vectors; and determining seizure onset and seizure termination based on the difference between the parameter vectors. A non-transitory computer readable program storage unit encoded with instructions that, when executed by a computer, perform the method.

Claims

1. A system, comprising:

a body data collection module configured to collect body data comprising a time series of a first body signal of a patient, wherein the first body signal is a cardiac signal or a kinetic signal, and

a non-transitory computer readable program storage unit encoded with instructions that, when executed by a processor, performs a method, comprising;

receiving the time series of the first body signal of the patient from the body data collection module,

determining a first sliding time window ending at a time τ and

a second sliding time window beginning at the time τ

for the time series of the first body signal;

applying an autoregression algorithm, comprising;

applying an autoregression analysis to the first sliding time window and the second sliding time window, wherein the autoregression analysis comprises presenting each sample as a weighted sum of P previous values with weights given by autoregression coefficients, wherein the autoregression coefficients are determined by a technique selected from an ordinary least squares procedure, a method of moments, Yule-Walker equations, a maximum entropy spectra estimation, or a maximum likelihood estimation;

plus a shift, and generating a first residual value for the first sliding time window, a second residual value for the second sliding time window, a first variance for the first sliding time window, and a second variance for the second sliding time window;

estimating a first parameter vector based at least in part on the autoregression coefficients for the first sliding time window and a second parameter vector based at least in part on the autoregression coefficients for the second sliding time window;

determining a non-stationarity measure and a third residual value by computing a first matrix in the first sliding time window and a second matrix in the second sliding time window using a Fisher'"'"'s matrix function;

determining an onset of a seizure based on the non-stationarity measure exceeding a threshold and a second variance of the residuals in the second sliding time window is larger than a first variance of the residuals in the first sliding time window; and

determining a termination of the seizure based on the non-stationarity measure exceeding the threshold and the first variance of the residuals in the first sliding time window is larger than the second variance of the residuals in the second sliding time window.

2. The system of claim 1, including data that when executed by a processor performs the method of claim 1, wherein the autoregression model is of second order, and parameters of said autoregression model comprise second time window length of 1 second;

first time window length of 1 second; and

a detection threshold of 3 for determining seizure onset and termination.

3. The system of claim 1, including data that when executed by the processor performs the method of claim 1, wherein parameters of the autoregression analysis are selected based on at least one of a clinical application of a detection;

a level of safety risk associated with an activity;

at least one of an age, a physical state, or a mental state of the patient;

a length of a window available for a warning;

a degree of efficacy of a therapy and a latency of the therapy;

a degree of seizure control;

a degree of circadian and ultradian fluctuations of a patient'"'"'s seizure activity;

a performance of the detection method as a function of a patient'"'"'s sleep/wake cycle or a vigilance level;

a dependence of a patient'"'"'s seizure activity on at least one of a level of consciousness, a level of cognitive activity, or a level of physical activity;

a site of a seizure origin;

a seizure type suffered by the patient;

a desired sensitivity of detection of the seizure, a desired specificity of detection of the seizure, a desired speed of detection of the seizure, an input provided by the patient or provided by a sensor.

4. The system of claim 1, further comprising:

a therapy unit configured to deliver a therapy for the seizure at a particular time, wherein at least one of the therapy, the particular time, or both is based upon the determination of the onset of the seizure.

5. The system of claim 4, including data that when executed by the processor performs the method of claim 4, wherein at least one of the delivered therapy or the issued warning is based at least in part on at least one of a type of activity engaged in by the patient at a seizure onset time, a seizure type, a seizure severity, or a time elapsed from a last seizure.

6. The system of claim 1, including data that when executed by a processor performs the method of claim 1, wherein the method further comprises determining at least one of:

a timing of delivery of therapy, a duration of a therapy, a type of therapy, at least one parameter of the therapy, a timing of sending a warning, a type of warning, or a duration of the warning;

based upon said seizure onset, said seizure termination, or both.

7. The system of claim 1, including data that when executed by a processor performs the method of claim 1, wherein the method further comprises:

determining at least one value selected from the duration of said seizure, the severity of said seizure, the intensity of said seizure, the extent of spread of said seizure, an inter-seizure interval between said seizure and a prior seizure, a patient impact of said seizure, or a time of occurrence of said seizure; and

logging said at least one value.

8. The system of claim 1, including data that when executed by a processor performs the method of claim 1, wherein the method further comprises at least one of:

determining an occurrence of a seizure based on the output of applying an autoregression algorithm on at least one second body signal,

determining an occurrence of a seizure based on the output of at least one second algorithm on said first body signal, or

determining an occurrence of a seizure based on the output of at least one second algorithm on said at least one second body signal.

9. The system of claim 8, including data that when executed by a processor performs the method of claim 8, wherein the second body signal is selected from an EKG signal, an accelerometer signal, or a signal indicative of a loss of responsiveness.

10. The system of claim 1, including data that when executed by a processor performs the method of claim 1, wherein the method further comprises estimating the degree of nonstationarity of said first body signal.

21. The system of claim 1, including data that when executed by the processor performs the method of claim 1, wherein the method further comprises at least one responsive action selected from:

determining an efficacy of a therapy;

or

issuing a warning for the seizure, wherein the warning is based upon a determination of the onset of the seizure, a determination of a duration of the seizure, a determination of a seizure type, or two or more thereof.

11. A method, comprising:

collecting, by a body data collection module, body data comprising a time series of a first body signal of a patient, wherein the first body signal is a cardiac signal or a kinetic signal,

determining a first sliding time window ending at a time τ and

a second sliding time window beginning at the time τ

for the time series of the first body signal;

applying an autoregression algorithm, comprising;

applying an autoregression analysis to the first sliding time window and the second sliding time window, wherein the autoregression analysis comprises presenting each sample as a weighted sum of P previous values with weights given by autoregression coefficients, wherein the autoregression coefficients are determined by a technique selected from an ordinary least squares procedure, a method of moments, Yule-Walker equations, a maximum entropy spectra estimation, or a maximum likelihood estimation;

plus a shift, and generating a first residual value for the first sliding time window, a second residual value for the second sliding time window, a first variance for the first sliding time window, and a second variance for the second sliding time window;

estimating a first parameter vector based at least in part on the autoregression coefficients for the first sliding time window and a second parameter vector based at least in part on the autoregression coefficients for the second sliding time window;

determining a difference between the first parameter vector and the second parameter vector by computing a first matrix in the first sliding time window and a second matrix in the second sliding time window using a Fisher'"'"'s matrix function;

determining an onset of a seizure based on the difference between the first parameter vector and the second parameter vector indicating that a second variance in the second sliding time window is larger than a first variance in the first sliding time window; and

determining a termination of the seizure based on the difference between the first parameter vector and the second parameter vector indicating that the first variance in the first sliding time window is larger than the second variance in the second sliding time window.

12. The method of claim 11, wherein p=2, and parameters of the autoregression algorithm comprise a second sliding time window length of 1 second;

a first sliding time window length of 1 second; and

a threshold equals 3 for determining the onset of the seizure and the termination of the seizure.

13. The method of claim 11, wherein parameters of the autoregression model are selected based on at least one of a clinical application of a detection;

a level of safety risk associated with an activity;

at least one of an age, a physical state, or a mental state of the patient;

a length of a window available for a warning;

a degree of efficacy of a therapy and a latency of the therapy;

a degree of seizure control;

a degree of circadian and ultradian fluctuations of a patient'"'"'s seizure activity;

a performance of the detection method as a function of a patient'"'"'s sleep/wake cycle or a vigilance level;

a dependence of a patient'"'"'s seizure activity on at least one of a level of consciousness, a level of cognitive activity, or a level of physical activity;

a site of a seizure origin;

a seizure type suffered by the patient;

a desired sensitivity of detection of the seizure, a desired specificity of detection of the seizure, a desired speed of detection of the seizure, an input provided by the patient or provided by a sensor.

14. The method of claim 11, further comprising at least one responsive action selected from:

delivering, by a therapy unit, a therapy for the seizure at a particular time, wherein at least one of the therapy, the particular time, or both is based upon the

determination of the onset of the seizure;

determining an efficacy of the therapy;

or issuing a warning for the seizure, wherein the warning is based upon the determination of the onset of the seizure, a determination of duration of the seizure, a determination of a seizure type, or two or more thereof.

15. The method of claim 14, wherein at least one of the delivered therapy or the issued warning may be based at least in part on at least one of a type of activity engaged in by the patient at a seizure onset time, the seizure type, a seizure severity, or a time elapsed from a last seizure.

16. The method of claim 11, further comprising determining at least one of a timing of delivery of therapy, a type of therapy, a duration of a therapy, at least one parameter of the therapy, a timing of sending a warning, a type of warning, or a duration of the warning, based upon said seizure onset, said seizure termination, or both.

17. The method of claim 11, further comprising:

determining at least one value selected from the duration of said seizure, the severity of said seizure, the intensity of said seizure, the extent of spread of said seizure, an inter-seizure interval between said seizure and a prior seizure, a patient impact of said seizure, or a time of occurrence of said seizure; and

logging said at least one value.

18. The method of claim 11, further comprising at least one of:

determining an occurrence of a seizure based on the output of applying an autorgression algorithm on at least one second body signal,

determining an occurrence of a seizure based on the output of at least one second algorithm on said first body signal, or

determining an occurrence of a seizure based on the output of at least one second algorithm on said at least one second body signal.

19. The method of claim 18, wherein the second body signal is selected from an EKG signal, an accelerometer signal, or a signal indicative of a loss of responsiveness.

20. The method of claim 11, further comprising estimating the degree of nonstationarity of said first body signal.