Method and apparatus for predicting the onset of seizures based on features derived from signals indicative of brain activity

• US 6,658,287 B1
• Filed: 05/18/2001
• Issued: 12/02/2003
• Est. Priority Date: 08/24/1998
• Status: Expired due to Fees

• Current Assignee: Emory University, Georgia Tech Research Corporation (University System of Georgia)
• Original Assignee: Georgia Tech Research Corporation (University System of Georgia)
• Inventors: Vachtsevanos, George, Echauz, Javier, Litt, Brian, Esteller, Rosana
• Primary Examiner(s): Winakur, Eric F.
• Assistant Examiner(s): SZMAL, BRIAN SCOTT
• Application Number:
  US09/762,455
• Time in Patent Office: 928 Days
• Field of Search: 600/544, 600/545, 600/546, 600/300, 607/45, 607/62
• US Class Current: 600/544
• CPC Class Codes: A61B 5/369 Electroencephalography [EEG] devices for psychotechnics A61B5/16
• CPC Class Codes: A61B 5/37 Intracranial electroencephalography [IC-EEG], e.g. electrocorticography [ECoG]
• CPC Class Codes: A61B 5/372 Analysis of electroencephalograms
• 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 5/726 using Wavelet transforms
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16H 20/70 relating to mental therapies, e.g. psychological therapy or autogenous training
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 40/67 for remote operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders

Abstract

This invention is a method, and system for predicting the onset of a seizure prior to electrograph onset in an individual. During an “off-line” mode, signals representing brain activity of an individual (either stored or real time) are collected, and features are extracted from those signals. A subset of features, which comprise a feature vector, are selected by a predetermined process to most efficiently predict (and detect) a seizure in that individual. An intelligent prediction subsystem is also trained “off-line” based on the feature vector derived from those signals. During “on-line” operation, features are continuously extracted from real time brain activity signals to form a feacture vector, and the feature vector is continuously analyzed with the intelligent prediction subsystem to predict seizure onset in a patient. The system, and method are preferably implemented in an implanted device (102) that is capable of warning externally an individual of the probability of a seizure, and/or automatically taking preventative actions to abort the seizure. In addition, methods are provided for applying intervention measures to an animal to abort or modulat a seizure by adjusting the modality of an intervention measure; and/or parameters of an intervention measure based upon a probability measure indicative of a likelihood of seizure occurrence; and/or a predicted time to seizure onset.

Claims

1. A method for automatically predicting the onset of a seizure in an animal, comprising steps of:

(a) monitoring signals indicative of the activity of the brain of an animal;

(b) extracting a set of features from the signals;

(c) analyzing the set of features with a intelligent prediction subsystem; and

(d) generating, in response to analysis of the set of features by the intelligent prediction subsystem, an output indicative of the probability of occurrence of a seizure.

2. The method of claim 1, wherein the step of generating an output comprises generating a warning that a seizure is likely to occur.

3. The method of claim 1, wherein the step of generating an output comprises generating a measure of probability that a seizure will occur within a pre-identified time period.

4. The method of claim 3, and further comprising the steps of:

setting a probability threshold;

monitoring the probability and comparing it with the probability threshold; and

issuing an audible and/or visual warning alert when the probability exceeds the probability threshold.

5. The method of claim 3, wherein the step of generating an output comprises generating a plurality of probability measures each for a different time period.

6. The method of claim 5, and further comprising the step of applying an intervention measure, a character of which is based the probability measure and/or a predicted time to seizure occurrence.

7. The method of claim 1, and further comprising the step of applying an intervention measure beginning with an initial response when triggered in response to a relatively low probability measure and/or relatively remote predicted time to seizure onset, and escalating a character and/or modality of the intervention measure as the probability measure increases and/or predicted time to seizure onset is less remote.

8. The method of claim 7, wherein the step of applying an intervention measure comprises applying an intervention measure at a maximal intensity and/or combination of modalities when a feature identifies electrographic seizure onset.

9. The method of claim 1, and further comprising the step of applying intervention measures comprising pharmacological, cardiac pacing and/or electrical preventative measures to the animal when a seizure is predicted in order to terminate a seizure prior to its electrical or clinical onset, or to terminate a seizure after onset.

10. The method of claim 1, wherein the step (b) of extracting the set of features comprises extracting one or more instantaneous features.

11. The method of claim 1, wherein the step (b) of extracting the set of features comprises extracting one or more historical features.

12. The method of claim 11, wherein the step (b) of extracting the set of features comprises extracting one or more historical features using statistical process control techniques.

13. The method of claim 1, wherein the step (b) of extracting the set of features comprises artificially generating one or more features from the signals.

14. The method of claim 1, and further comprising the step of training the intelligent prediction subsystem to predict the onset of a seizure prior to its occurrence for a particular animal from data including signals indicative of the activity of the brain of a particular animal.

15. The method of claim 14, and further comprising the step of storing the data including signals indicative of the activity of the brain of a particular animal prior to and during a seizure event of the particular animal.

16. The method of claim 1, wherein the step (a) of monitoring signals comprises monitoring brain activity signals and other physiological signals indicative of the brain activity.

17. The method of claim 1, wherein the step of generating an output comprises generating a continuously updated probability measure indicative of the likelihood of occurrence of a seizure.

18. The method of claim 17, wherein the step of generating the probability measure comprises estimating the exact conditional probability function.

19. The method of claim 17, and further comprising the step of applying an intervention measure, a character of which is based on a mathematical function of the probability measure.

20. A method for automatically predicting the onset of a seizure in an animal, comprising steps of:

(a) monitoring signals indicative of the activity of the brain of an animal;

(b) extracting a set of features from the signals;

(c) forming a feature vector that is a combination of a plurality of features extracted from the signals;

(d) analyzing the set of features with a intelligent prediction subsystem; and

(e) generating an output indicative of the likelihood of occurrence of a seizure.

21. A system for predicting the onset of a seizure in an animal, comprising:

(a) at least one electrode for detecting signals indicative of the activity of the brain of animal;

(b) a processor coupled to the at least one electrode, the processor;

extracting a set of features from the brain activity signals;

forming a feature vector that is a combination of a plurality of features extracted from the signals;

continuously analyzing the set of features with a intelligent prediction process; and

generating as output a signal indicative of the likelihood of occurrence of a seizure.

28. The system of claim 21, wherein the processor generates an output comprising a probability measure representing the probability that a seizure will occur within a period of time.

29. The system of claim 28, wherein the processor generates as output a plurality of probability measures, each for a different time period.

30. The system of claim 29, wherein the processor generates a signal to cause application of an intervention measure, a character of which is based on the probability measure.

31. The system of claim 21, wherein the processor implements a trainable intelligence network as the intelligent prediction subsystem to analyze the set of features.

32. The system of claim 31, wherein the processor implements the intelligent prediction process with a wavelet neural network (WNN).

33. The system of claim 21, wherein the processor detects seizure onset from the set of features and generates a signal for causing the delivery of an intervention measure whose character is based upon multiple features of the feature set.

22. A system for predicting the onset of a seizure in an animal, comprising:

(a) at least one electrode for detecting signals indicative of the activity of the brain of animal;

(b) a processor coupled to the at least one electrode, the processor;

extracting a set of features from the brain activity signals;

continuously analyzing the set of features with a intelligent prediction process; and

generating as output, in response to continuous analysis of the set of features by the intelligent prediction process, a signal indicative of the probability of occurrence of a seizure.

23. The system of claim 22, wherein a predictor algorithm to be executed by the processor is trained “

off-line”

with data comprising signals indicative of brain activity obtained from a particular animal so as to operate “

on-line”

in real-time on signals obtained by the at least one electrode coupled to the particular individual.

24. The system of claim 22, wherein the processor is contained with an implantable unit for implantation in a body of an animal.

25. The system of claim 24, and further comprising a portable unit external of the body of the animal that communicates with the implantable unit via a communication link through the body of the animal, the portable unit comprising an alert device, the processor in the implantable unit generating as output a signal that activates the alert device in response to determining onset of a seizure.

26. The system of claim 25, wherein the portable unit comprises a display, wherein the processor generates a warning message that is transmitted via the communication link to the display.

27. The system of claim 25, wherein information stored in the implanted unit is uploaded through the portable unit for transmission to an external computer via a communications network or the Internet.

34. A method of automatically predicting the onset of a seizure comprising steps of:

(a) extracting a plurality of features from signals indicative of the brain activity of an animal;

(b) examining the plurality of features and selecting a subset of the plurality of features determined to predictive of seizure onset in the individual;

(c) training a intelligent prediction subsystem to predict a seizure in the individual based on the subset of features;

(d) continuously extracting the subset of features from real-time brain activity signals of an individual;

(e) continuously analyzing the subset of features with the intelligent prediction subsystem; and

(f) continuously generating as output a probability measure that a seizure will occur within a predetermined period of time.

35. The method of claim 34, wherein the step (f) of continuously generating comprises generating a plurality of probability measures each with respect to a different prediction time horizon.

36. The method of claim 34, wherein the step (c) of training comprises periodically training the intelligent prediction subsystem based on seizure and baseline data extracted for a particular animal to maintain performance of the intelligent prediction subsystem independent of conditions of the particular animal.

37. A method for applying intervention measures to an animal to abort or modulate a seizure comprising the step of adjusting the modality of an intervention measure and/or parameters of an intervention measure based upon a probability measure indicative of a likelihood of seizure occurrence and/or a predicted time to seizure onset.

38. The method of claim 37, wherein the step of adjusting comprises applying an intervention measure beginning with an initial response when triggered in response to a relatively low probability measure and/or relatively remote time to seizure onset, and escalating a character and/or modality of the intervention measure as the probability measure increases and/or time to seizure onset is less remote.

39. The method of claim 38, wherein the step of applying an intervention measure comprises applying an intervention measure at a maximal intensity and/or combination of modalities when a feature identifies electrographic seizure onset.