Adaptive method and apparatus for forecasting and controlling neurological disturbances under a multi-level control
First Claim
1. A method for predicting and controlling the electrographic and clinical onset of a seizure and other neurological events in an individual, comprising the acts of:
- generating data that is acquired from a plurality of input signals obtained from at least one sensor located in or on the individual;
fusing the data to combine information from the at least one sensor that is connected to at least one transducer;
selecting and extracting a plurality of features from the fused data;
determining from the extracted features if a seizure or other neurological event is likely to occur within a plurality of specified time frames, and the probability of having a seizure for each specified time frame;
providing an alarm to the individual to inform him of an imminent seizure or neurological event when the probability of seizure is higher than an adaptive threshold; and
applying a control rule to initiate an intervention measure that is commensurate with the probability of the electrographical onset of a seizure for each specified time frame.
1 Assignment
0 Petitions
Accused Products
Abstract
A method and apparatus for forecasting and controlling neurological abnormalities in humans such as seizures or other brain disturbances. The system is based on a multi-level control strategy. Using as inputs one or more types of physiological measures such as brain electrical, chemical or magnetic activity, heart rate, pupil dilation, eye movement, temperature, chemical concentration of certain substances, a feature set is selected off-line from a pre-programmed feature library contained in a high level controller within a supervisory control architecture. This high level controller stores the feature library within a notebook or external PC. The supervisory control also contains a knowledge base that is continuously updated at discrete steps with the feedback information coming from an implantable device where the selected feature set (feature vector) is implemented. This high level controller also establishes the initial system settings (off-line) and subsequent settings (on-line) or tunings through an outer control loop by an intelligent procedure that incorporates knowledge as it arises. The subsequent adaptive settings for the system are determined in conjunction with a low-level controller that resides within the implantable device. The device has the capabilities of forecasting brain disturbances, controlling the disturbances, or both. Forecasting is achieved by indicating the probability of an oncoming seizure within one or more time frames, which is accomplished through an inner-loop control law and a feedback necessary to prevent or control the neurological event by either electrical, chemical, cognitive, sensory, and/or magnetic stimulation.
-
Citations
103 Claims
-
1. A method for predicting and controlling the electrographic and clinical onset of a seizure and other neurological events in an individual, comprising the acts of:
-
generating data that is acquired from a plurality of input signals obtained from at least one sensor located in or on the individual;
fusing the data to combine information from the at least one sensor that is connected to at least one transducer;
selecting and extracting a plurality of features from the fused data;
determining from the extracted features if a seizure or other neurological event is likely to occur within a plurality of specified time frames, and the probability of having a seizure for each specified time frame;
providing an alarm to the individual to inform him of an imminent seizure or neurological event when the probability of seizure is higher than an adaptive threshold; and
applying a control rule to initiate an intervention measure that is commensurate with the probability of the electrographical onset of a seizure for each specified time frame. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58)
rhythmic electrical pacing that changes in frequency, intensity and distribution as the probability of a seizure onset reaches and exceeds a threshold;
chaos control pacing;
random electrical stimulation to interfere with developing coherence in activity in a region of, and surrounding, an epileptic focus;
depolarization or hyperpolarization stimuli to silence or suppress activity in actively discharging regions, or regions at risk for seizure spread.
-
-
15. The method for predicting and controlling the electrographic onset of a seizure of claim 14 wherein the intervention measure is delivered to a plurality of electrodes to provide a surround inhibition to prevent a progression of a seizure precursor.
-
16. The method for predicting and controlling the electrographic onset of a seizure of claim 14 wherein the intervention measure is delivered sequentially in a wave that covers a cortical or subcortical region of tissue so as to progressively inhibit normnal or pathological neuronal function in the covered region.
-
17. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the intervention measure application is an infusion of a therapeutic chemical agent into a brain region where seizures are generated, or to which they may spread.
-
18. The method for predicting and controlling the electrographic onset of a seizure of claim 17 wherein the chemical agent is delivered in greater quantity, concentration or spatial distribution as the probability of seizure increases.
-
19. The method for predicting and controlling the electrographic onset of a seizure of claim 17 wherein the intervention measure is applied to at least one of an epilectic focus, an area surrounding the epilectic focus, a region involved in an early spread, and a central or deep brain region to modulate seizure propagation.
-
20. The method for predicting and controlling the electrographic onset of a seizure of claim 17 wherein the therapeutic chemical agent is activated by oxidative stress and increases in concentration and distribution as the probability of seizure increases.
-
21. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the intervention measure is delivered to central nerves or blood vessels in a graduated manner as the probability of seizure increases.
-
22. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the intervention measure is a plurality of artificial neuronal signals delivered to disrupt eletrochemical traffic on at least one neuronal network that includes or communicates with an ictal onset zone.
-
23. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the alarm is any one of a visual signal, an audio signal and a tactile sensation.
-
24. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the plurality of features are selected for each individual.
-
25. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the same plurality of features are selected for each individual.
-
26. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein parameters of the selected features are tuned for each individual.
-
27. The method for predicting and controlling the electrographic onset of a seizure of claim 26 wherein one of the parameters that is used for each selected feature is a running window length that is used in feature extraction.
-
28. The method for predicting and controlling the electrographic onset of a seizure of claim 27 wherein a determination of the running window length and a starting time for feature extraction over an input signal for every feature includes the acts of:
-
determining a window range based on stationarity criteria and a minimum length to compute a feature under analysis;
determining a feature value for each of a plurality of different window sizes;
calculating a feature effectiveness measure based on class distinguishability for the plurality of different window sizes used for every feature;
determining the window length that corresponds to a best class distinguishability as indicated by a maximum value or minimum value of the feature effectiveness measure; and
aligning the plurality of windows with the window having the maximum length such that the right edge of all windows coincide.
-
-
29. The method for predicting and controlling the electrographic onset of a seizure of claim 28 wherein the maximum or minimum values of the feature effectiveness measure that provides the best class distinguishability depends on the feature effectiveness measure in use.
-
30. The method for predicting and controlling the electrographic onset of a seizure of claim 28 wherein the feature effectiveness measure determines the window length that maximizes the distinguishability between a preictal/ictal class and a baseline class.
-
31. The method for predicting and controlling the electrographic onset of a seizure of claim 30 wherein the act of selecting and extracting a plurality of features comprises the acts of:
-
extracting a set of candidate features from the feature library;
ranking the extracted features by the feature effectiveness measure; and
determining a smallest subset of features that satisfies a performance criterion.
-
-
32. The method for predicting and controlling the electrographic onset of a seizure of claim 31 further comprising the acts of:
-
performing an initial pre-selection from the feature library to discard a plurality of features with inferior class separability; and
evaluating individual feature performance using at least one criterion for every feature that is not discarded during the initial pre-selection.
-
-
33. The method for predicting and controlling the electrographic onset of a seizure of claim 31 wherein the act or ranking the extracted features by the feature effectiveness measure uses an overlap measure criterion, a modified add-on algorithm and heuristics to select a final feature set.
-
34. The method for predicting and controlling the electrographic onset of a seizure of claim 33 wherein the overlap measure criterion is based on functions proportional to the estimated conditional probability distributions of the features under analysis for both a pre-seizure class and a non-pre-seizure class.
-
35. The method for predicting and controlling the electrographic onset of a seizure of claim 31 further comprising the acts of constructing and evaluating two-dimensional feature spaces to validate qualitatively that the final feature set is complementary and has low correlation among the final features.
-
36. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein a plurality of features are extracted at an analog level.
-
37. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein a plurality of features are extracted at a digital level.
-
38. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the plurality of features are extracted over a pre-established window length.
-
39. The method for predicting and controlling the electrographic onset of a seizure of claim 38 further comprising shifting of the window over the plurality of input signals to allow at least a partial overlap with a previous window, reusing the extracted features in the overlap portion and repeating the extraction of the plurality of features on a new input portion within the window.
-
40. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the act of fusing the data comprises the act of combining the plurality of signals from at least one sensor using an intelligent tool including a neural network or a fuzzy logic algorithmn.
-
41. The method for predicting and controlling the electrographic onset of a seizure of claim 40 wherein the neural network or fuzzy logic algorithm include at least one of a probabilistic neural network, a k-nearest neighbor neural network, a wavelet network, and a combination probabilistic/k-nearest neighbor neural network.
-
42. The method for predicting and controlling the electrographic onset of a seizure of claim 1 wherein the plurality of features is selected from a feature library including a plurality of historical and instantaneous features.
-
43. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the plurality of instantaneous features are generated directly from preprocessed and fused input signals through a running observation window.
-
44. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the historical features are based on a historical evolution of features over time.
-
45. The method for predicting and controlling the electrographic onset of a seizure of claim 44 wherein at least one historical feature is generated as a feature of other features by a second or higher level of feature extraction.
-
46. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the historical and instantaneous features are limited to a focus region in the brain of an individual.
-
47. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the historical and instantaneous features are derived as a spatial feature from a combination of a plurality of regions in the brain of an individual.
-
48. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the feature library includes a collection of custom routines to compute the features.
-
49. The method for predicting and controlling the electrographic onset of a seizure of claim 42 wherein the plurality of features are extracted from different domains.
-
50. The method for predicting and controlling the electrographic onset of a seizure of claim 49 wherein at least one feature is a ratio of a short term value and a long term value of that feature.
-
51. The method for predicting and controlling the electrographic onset of a seizure of claim 49 wherein the different domains include at least two of time, frequency, wavelet, fractal geometry, stochastic processes, statistics, and information theory domains.
-
52. The method for predicting and controlling the electrographic onset of a seizure of claim 51 wherein the time domain features include at least one of an average power, a power derivative, a fourth-power indicator, an accumulated energy, an average non-linear energy, a thresholded non-linear energy, a duration of thresholded non-linear energy, and a ratio of short term and long term power feature.
-
53. The method for predicting and controlling the electrographic onset of a seizure of claim 52 wherein the fractal geometry features include at least one of a fractal dimension of analog signal, a curve length, a fractal dimension of digital signals, a ratio of short term and long term curve length, an a ratio of short term and long term fractal dimensions of digital signals.
-
54. The method for predicting and controlling the electrographic onset of a seizure of claim 52 wherein the frequency domain features include at least one of a power spectrum, a power on frequency bands, a coherence between intracranial channels, a mean crossings and a zero crossings feature.
-
55. The method for predicting and controlling the electrographic onset of a seizure of claim 52 wherein the wavelet domain features include at least one of a spike detector, a density of spikes over time, and an absolute value of a wavelet coefficient.
-
56. The method for predicting and controlling the electrographic onset of a seizure of claim 52 wherein the statistics and stochastic process domains include at least one of a mean frequency index, a cross-correlation between different intracranial channels, and autoregressive coefficients.
-
57. The method for predicting and controlling the electrographic onset of a seizure of claim 52 wherein the information theory features include at least one of an entropy feature and an average mutual information feature.
-
58. The method for predicting and controlling the electrographic onset of a seizure of claim 1 further comprising the act of fusing the selected features to include establishing an individual-tuned variable normalization level that uses an individual'"'"'s state of awareness to normalize an accumulated energy or other feature and decide if a seizure is approaching when a normalized threshold value is exceeded.
-
59. A computer readable medium containing a computer program product for predicting and controlling the electrographic and clinical onset of a seizure and other neurological events in an individual, the computer program product comprising:
-
program instructions that generate data acquired from a plurality of input signals obtained from at least one sensor located in or on the individual;
program instructions that fuse the data to combine information from the at least one sensor that is connected to at least one transducer;
program instructions that select and extract a plurality of features from the fused data;
program instructions that determine from the extracted features if a seizure or other neurological event is likely to occur within a plurality of specified time frames, and the probability of having a seizure for each specified time frame;
program instructions that generate an alarm to the individual to inform him of an imminent seizure or neurological event when the probability of seizure is higher than an adaptive threshold; and
program instructions that apply a control rule to initiate an intervention measure that is commensurate with the probability of the electrographical onset of a seizure. - View Dependent Claims (60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103)
program instructions that determine a window range based on stationarity criteria and a minimum length to compute a feature under analysis;
program instructions that determine a feature value for each of a plurality of different window sizes;
program instructions that calculate a feature effectiveness measure for each feature for the plurality of different window sizes;
program instructions that determine the optimal window length for each feature from the plurality of windows examined that corresponds to a value of the feature effectiveness measure wherein the distinguisability between a preictal class and a non-preictal class is maximized; and
program instructions that align the plurality of optimal windows determined for each feature with the feature window having the maximum length.
-
-
77. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 76 further comprising program instructions that initiate re-execution of the program instructions that determine a feature value and the program instructions that calculate a feature effectiveness measure for each selected feature.
-
78. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 76 further comprising program instructions that maximize the distinguishability between a preictal/ictal class and a baseline class as the feature effectiveness measure.
-
79. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 78 wherein the program instructions that select and extract a plurality of features comprise:
-
program instructions that extract a set of candidate features from the feature library;
program instructions that rank the extracted features by the feature effectiveness measure; and
program instructions that determine a smallest subset of features that satisfies a performance criterion.
-
-
80. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 79 further comprising:
-
program instructions that perform an initial pre-selection from the feature library to discard a plurality of features with inferior class separability; and
program instructions that evaluate individual feature performance using at least one criterion for every feature that is not discarded during the initial pre-selection.
-
-
81. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 79 wherein the program instructions that rank the extracted features by the feature effectiveness measure use an overlap measure criterion, a modified add-on algorithm and heuristics to select a final feature set.
-
82. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 81 further comprising program instructions that construct and evaluate two-dimensional feature spaces to validate qualitatively that the final feature set is complementary and has low correlation among the final features.
-
83. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 81 further comprising program instructions that base the overlap measure criterion on estimated conditional probability distributions of each particular feature under analysis for both a pre-seizure class and non-pre-seizure class.
-
84. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 59 further comprising program instructions that extract a plurality of features at an analog level.
-
85. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 59 further comprising program instructions that extract a plurality of features at a digital level.
-
86. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 59 further comprising program instructions that combine the plurality of signals from at least one sensor using an intelligent tool that includes a neural network or a fuzzy logic algorithm.
-
87. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 86 further comprising program instructions that determine at least one of a probabilistic neural network, a k-nearest neighbor neural network, a wavelet network, and a combination probabilistic/k-nearest neighbor neural network.
-
88. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 59 further comprising program instructions that select a plurality of features from a feature library that includes a plurality of historical and instantaneous features.
-
89. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions that generate a plurality of instantaneous features directly from pre-processed and fused input signals through a running observation window.
-
90. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions that generate historical features based on a historical evolution of features over time.
-
91. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions that limit the historical and instantaneous features to a focus region in the brain of an individual.
-
92. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions that derive historical and instantaneous features as a spatial feature from a combination of a plurality of regions in the brain of an individual.
-
93. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions collected as custom routines within the feature library to compute the features.
-
94. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 88 further comprising program instructions that extract a plurality of features from different domains.
-
95. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 94 further comprising program instructions that determine at least one feature as a ratio of a short term value and a long term value of that feature.
-
96. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 94 wherein the different domains include at least two of time, frequency, wavelet, fractal geometry, stochastic processes, statistics, and information theory domains.
-
97. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 96 further comprising program instructions that determine at least one of an average power, a power derivative, a fourth-power indicator, an accumulated energy, and average non-linear energy, a thresholded non-linear energy, a duration of thresholded non-linear energy, and a ratio of short term and long term power as time domain features.
-
98. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 97 further comprising program instructions that determine at least one of a fractal dimension of analog signals, a curve length, a fractal dimension of digital signals, a ratio of a short term and a long term fractal dimension of digital signals, and a ratio of short term and long term curve length as fractal geometry features.
-
99. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 97 further comprising program instructions that determine at least one of a power spectrum, a power on frequency bands, a coherence between intracranial channels, a mean crossings and a zero crossings feature.
-
100. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 97 further comprising program instructions that determine at least one of a spike detector, a density of spikes over time, and an absolute value of a wavelet coefficient as wavelet domain features.
-
101. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 97 further comprising program instructions that determine at least one of a mean frequency index, a cross-correlation between different intracranial channels, and autoregressive coefficients as features in the statistics and stochastic process domains.
-
102. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 97 further comprising program instructions that determine at least one of an entropy feature and an average mutual information feature as information theory features.
-
103. The computer program product for predicting and controlling the electrographic onset of a seizure of claim 59 further comprising program instructions that fuse the selected features to include establishing an individual-tuned variable normalization level that uses an individual'"'"'s state of awareness to normalize an accumulated energy or other feature and decide if a seizure is approaching when a normalized threshold value is exceeded.
Specification