JP7138995B2 - A method, computer system, and program for performing burst analysis on waveform data acquired from the brain, and a method, computer system, and program for predicting the state of a target using burst analysis - Google Patents

Description

The present invention relates to a method, computer system, and program for performing burst analysis on waveform data acquired from the brain, and a method, computer system, and program for predicting the state of a target using burst analysis.

In non-clinical studies, neuronal network activity of human-derived neurons and the like is obtained using a micro-electrode array (MEA) or the like, and studies are being conducted to examine the effects of pharmaceuticals (Non-Patent Document 1).

A. Odawara, H. Katoh, N. Matsuda & I. Suzuki, "Physiological maturation and drug responses of human induced pluripotent stem cell-derived cortical neuronal networks in long-term culture", Scientific Reports volume 6, Article number: 26181 ( 2016)

An object of the present invention is to provide a novel technique for predicting unknown properties of target compounds.

In one embodiment, the present invention provides, for example, the following items.
(Item 1)
A method of performing burst analysis on waveform data acquired from the brain, comprising:
dividing the waveform data into a plurality of frequency bands;
acquiring spike data for each frequency band by detecting spikes in waveform data for each of the plurality of frequency bands;
detecting bursts in the acquired spike data;
deriving data for burst parameters for each frequency band based on the detected bursts.
(Item 2)
The step of dividing the waveform data into a plurality of frequencies includes: a θ waveband of about 5 to about 8 Hz; The method of item 1, comprising dividing into at least five bands including a gamma waveband and a high-gamma waveband of about 70 to about 150 Hz.
(Item 3)
detecting the spike includes detecting a spike having a predetermined duration;
3. The method according to item 1 or 2, wherein the predetermined time duration is a reciprocal of a representative frequency of the divided frequency bands.
(Item 4)
The burst parameters include firing frequency, burst frequency, burst interval, burst duration, number of spikes within burst, maximum amplitude within burst, maximum amplitude interval, CV value of burst interval, CV value of burst duration, CV value of number of spikes within burst. , the CV value of the maximum intra-burst amplitude, and the CV value of the maximum amplitude interval.
(Item 5)
A method of predicting a state of a target, comprising:
obtaining first waveform data acquired from the brain of a subject having a first condition;
obtaining second waveform data obtained from the brain of a subject having a second condition;
Multivariate analysis is performed on a combination of burst parameter data derived from the first waveform data and the second waveform data according to the method according to any one of items 1 to 4, and a prediction parameter set is obtained. a step of identifying;
predicting the state based on the predictive parameter set from target waveform data obtained from the target brain.
(Item 6)
further comprising obtaining third waveform data obtained from the brain of a third subject;
Item 5, wherein identifying the predictive parameter set includes performing multivariate analysis of the first waveform data, the second waveform data, and the third waveform data to identify a predictive parameter set. The method described in .
(Item 7)
The first state is a state having a predetermined symptom,
The second state is a state that does not have the predetermined symptom,
The third state is a state having a precursor of the predetermined symptom,
7. The method of item 6, wherein the predictive parameter set includes a combination of parameters that can separate the first state, the second state and the third state.
(Item 8)
8. The method of item 7, wherein the predetermined symptom is a convulsive symptom.
(Item 9)
A computer system for performing burst analysis on waveform data acquired from the brain,
dividing means for dividing the waveform data into a plurality of frequency bands;
Acquisition means for acquiring spike data in each frequency band by detecting a spike in waveform data in each frequency band of the plurality of frequency bands;
detection means for detecting bursts in the acquired spike data;
derivation means for deriving data for burst parameters for each frequency band based on said detected bursts.
(Item 9A)
9A. The computer system of item 9A, including the features of one or more of the above items.
(Item 10)
A program for performing burst analysis on waveform data acquired from the brain, said program being executed in a computer system comprising a processor, said program comprising:
dividing the waveform data into a plurality of frequency bands;
acquiring spike data for each frequency band by detecting spikes in waveform data for each of the plurality of frequency bands;
detecting bursts in the acquired spike data;
deriving data for burst parameters for each frequency band based on the detected bursts.
(Item 10A)
11. Program according to item 10, including features according to one or more of the above items.
(Item 10B)
A storage medium for storing the program according to item 10 or item 10A.
(Item 11)
A computer system for predicting the state of a target, comprising:
receiving means for receiving first waveform data obtained from the brain of a subject having a first state and second waveform data obtained from the brain of a subject having a second state;
Multivariate analysis is performed on a combination of burst parameter data derived from the first waveform data and the second waveform data according to the method according to any one of items 1 to 4, and a prediction parameter set is obtained. a specific means to identify;
a prediction means for predicting said state based on said prediction parameter set from target waveform data obtained from said target's brain.
(Item 11A)
The computer system of item 11A, including the features of one or more of the above items.
(Item 12)
A program for predicting the state of a target, said program being executed in a computer system comprising a processor, said program comprising:
obtaining first waveform data acquired from the brain of a subject having a first condition;
obtaining second waveform data obtained from the brain of a subject having a second condition;
Multivariate analysis is performed on a combination of burst parameter data derived from the first waveform data and the second waveform data according to the method according to any one of items 1 to 4, and a prediction parameter set is obtained. a step of identifying;
predicting the state based on the prediction parameter set from the target waveform data obtained from the brain of the target.
(Item 12A)
13. A program according to item 12, including the features described in one or more of the above items.
(Item 12B)
A storage medium for storing the program according to item 12 or item 12A.

According to the present invention, it is possible to provide a method of performing burst analysis that can be used to predict the state of a target. Further, according to the present invention, it is possible to provide a method for predicting the state of a target using burst analysis. This makes it possible to predict unknown properties of target compounds.

A diagram showing an example of the configuration of a computer system 100 for predicting the state of a target. FIG. 4 is a diagram showing an example of the configuration of the processor 120; A diagram explaining a method for detecting bursts in spike data A diagram explaining a method for detecting bursts in spike data A diagram explaining a method for detecting bursts in spike data A diagram explaining a method for detecting bursts in spike data Flowchart showing an example of processing in the computer system 100 for predicting the state of the target Diagram showing an example of electroencephalogram data divided into multiple frequency bands Flowchart showing an example of processing in the computer system 100 for predicting the state of the target Flowchart showing an example of a process in which the processor 120 predicts the state of the target in step S604 An example of principal component plots of target waveform data, first waveform data, and second waveform data Flowchart showing an example of processing in the computer system 100 for predicting the state of the target Figure showing the results of Example 1 The figure which shows the result of Example 2. The figure which shows the result of Example 3 The figure which shows the result of Example 4. The figure which shows the result of Example 5 The figure which shows the result of Example 5 The figure which shows the result of Example 5 The figure which shows the result of Example 6 The figure which shows the result of Example 6

The present invention will be described below. It should be understood that the terms used herein have the meanings commonly used in the art unless otherwise specified. Thus, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. In case of conflict, the present specification (including definitions) will control.

1. Definition As used herein, the term "target" refers to a living organism whose state is to be predicted. The target may be human, non-human animal, or both human and animal.

As used herein, the term "target compound" refers to a compound whose properties are to be predicted. The target compound may be an unknown compound or a known compound. Properties of a compound of interest include, but are not limited to, efficacy, toxicity, and mechanism of action.

As used herein, the term "subject" refers to a living organism to which a target compound is administered. The subject and the target may be humans, animals other than humans, or both humans and animals.

As used herein, "medicinal effect" refers to the resulting effect when a drug is applied to a subject. For example, if the drug is an anti-cancer drug, the drug effect will be a direct effect on the object such as reduction of the cancer area under X-ray observation, delay of cancer progression, and extension of cancer patient's survival period. It can be an effect or an indirect effect such as a reduction in biomarkers that correlate with cancer progression. As used herein, "medicinal efficacy" contemplates efficacy under any applicable conditions. For example, if the drug is an anticancer drug, the efficacy may be an effect in a specific subject (e.g., a man aged 80 or over) or a specific application condition (e.g., other anticancer therapy (under combined use) may be the effect. In one embodiment, the drug may have a single medicinal effect or may have multiple medicinal effects. In one embodiment, the drug may have different efficacy under different application conditions. In general, efficacy refers to the effect that is intended to be achieved.

As used herein, "toxicity" is an undesirable effect that occurs when a drug is applied to a subject. In general, toxicity is an effect that differs from the intended effect of a drug. Toxicity can occur through a different mechanism of action than the drug effect or through the same mechanism of action as the drug effect. For example, if the drug is an anticancer drug, it may cause hepatotoxicity by killing normal hepatocytes at the same time as the efficacy of killing cancer cells through the mechanism of action of suppressing cell proliferation. Efficacy in killing cancer cells via an inhibitory mode of action may be accompanied by neuronal dysfunction toxicity via a membrane stabilizing mode of action.

As used herein, "mode of action" is the manner in which a drug interacts with biological mechanisms. For example, if the drug were an anticancer drug, the mechanism of action would be activation of the immune system, killing of rapidly proliferating cells, blocking of proliferative signaling, blocking of specific receptors, blocking of specific genes. There may be different levels of events such as transcriptional inhibition. Once the mechanism of action is identified, efficacy, toxicity and/or appropriate mode of use can be predicted based on accumulated information.

As used herein, "about" means ±10% of the numerical value that follows.

2. Target State Prediction The inventors of the present invention have developed a target state prediction method using multivariate analysis to predict unknown properties of target compounds. In the method of the present invention, a prediction parameter set suitable for separating known states is derived by using multivariate analysis for multiple (preferably all) combinations in a parameter set containing multiple candidate parameters. can do. The state of the target can be predicted by analyzing the waveform data obtained from the brain of the target when the compound of interest is administered to the target using the predictive parameter set. Based on this prediction, properties of the compound of interest can be predicted.

For example, by multivariate analysis of waveform data obtained from the brain of a subject having a predetermined symptom and waveform data obtained from the brain of a subject not having a predetermined symptom, A predictive parameter set is derived that is capable of separating the condition of having a condition from the condition of not having a given symptom. By analyzing waveform data acquired from the brain of a target having an unknown state using this predictive parameter set, it is possible to determine whether the target is in a state of having a predetermined symptom or has a predetermined symptom. It is possible to predict whether it is in a state where it is not

For example, it can be used to predict efficacy, toxicity, and mechanism of action of a compound of interest. For example, a plurality of known compounds each having a property to be separated (for example, when separating the property of "presence or absence of toxicity", a known compound A having a property of "having toxicity" and a known compound having a property of "no toxicity" By multivariate analysis of the waveform data obtained from the brain when B) is administered to separate subjects, a predictive parameter set that can separate the state of toxicity and the state of non-toxicity is derived. do. Using this predictive parameter set, by analyzing the waveform data acquired from the target's brain when the target compound is administered to the target, it is possible to determine whether the target's state is toxic or not. state, and based on this prediction, it is possible to predict whether the property of the target compound is "toxic" or "non-toxic". For example, by deriving a prediction parameter set according to the state to be separated, the target compound has what efficacy, or what toxicity, or what mechanism of action It will be possible to predict whether This makes it possible to predict the efficacy, toxicity, and mechanism of action of target compounds.

For example, data obtained from the brain can more accurately predict the identification of target compounds in vivo than data obtained directly from cultured neurons. Data obtained directly from cultured neurons are information obtained when the target compound was directly exposed to neurons, while data obtained from the brain were obtained through pharmacokinetics in vivo. This is because it is the information when the target compound is exposed to the cells.

Predicting the state of a target using such multivariate analysis can be realized, for example, by a computer system for predicting the state of a target described below.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings.

3. Configuration of Computer System for Predicting Target State FIG. 1 shows an example configuration of a computer system 100 for predicting target state, according to one embodiment of the present invention.

Computer system 100 comprises receiving means 110 , processor 120 , memory 130 and output means 140 . Computer system 100 may be connected to database unit 200 .

The receiving means 110 is configured to be able to receive data from outside the computer system 100 . The receiving means 110 may, for example, receive data from outside the computer system 100 via a network, or receive data from a storage medium (eg, USB memory, optical disk, etc.) connected to the computer system 100 or from the database unit 200. may be received. When data is received via a network, the type of network does not matter. The receiving means 110 may receive data using, for example, a wireless LAN such as Wi-fi, or may receive data via the Internet.

The receiving means 110 is configured to receive waveform data obtained from the subject's brain. For example, the receiving means 110 can receive waveform data acquired from the subject's brain when the known compound is administered. For example, the receiving means 110 can receive waveform data obtained from the subject's brain when no compound has been administered. The receiving means 110 may be further configured to receive waveform data acquired from the target brain.

The waveform data obtained from the brain is time-series potential data representing brain activity. The waveform data acquired from the brain includes electroencephalogram data, brain slice waveform data, and brain organoid waveform data. For example, scalp electroencephalogram (EEG) data is body surface potential time series data due to nerve activity measured using electrodes attached to the scalp. Cortical electroencephalogram (ECoG) data are cortical potential time series data due to neural activity measured using electrodes placed on the intracranial cortex.

The data received by the receiving means 110 are passed to the processor 120 for subsequent processing.

Processor 120 controls the overall operation of computer system 100 . Processor 120 reads a program stored in memory 130 and executes the program. This allows the computer system 100 to function as a device that executes desired steps. Processor 120 may be implemented by a single processor or by multiple processors. Data processed by processor 120 is passed to output means 140 for output.

The memory 130 stores programs for executing processes in the computer system 100, data required for executing the programs, and the like. The memory 130 stores, for example, a program for performing burst analysis on waveform data acquired from the brain (for example, a program for realizing the processing shown in FIG. 4, which will be described later), and a program for predicting the state of an object. (for example, a program that implements processes shown in FIGS. 6, 7, and 9, which will be described later). Memory 130 may store an application that implements any functionality. Here, it does not matter how the program is stored in the memory 130 . For example, the program may be preinstalled in memory 130 . Alternatively, the program may be installed in memory 130 by being downloaded via a network. Alternatively, the program may be stored in a machine-readable non-transitory storage medium. Memory 130 may be implemented by any storage means.

The output means 140 is configured to be able to output data to the outside of the computer system 100 . It does not matter in what manner the output means 140 enables information to be output from the computer system 100 . For example, if the output means 140 is a display screen, the information may be output to the display screen. Alternatively, if the output means 140 is a speaker, information may be output by sound from the speaker. Alternatively, if the output means 140 is a data writing device, information may be output by writing the information in a storage medium connected to the computer system 100 or in the database section 200 . Alternatively, if the output means 140 is a transmitter, the transmitter may output information by transmitting it to the outside of the computer system 100 via a network. In this case, the type of network does not matter. For example, a transmitter may transmit information over the Internet or transmit information over a LAN. For example, the output means 140 converts the data into a format that can be handled by the hardware or software of the data output destination, or adjusts the response speed so that it can be handled by the hardware or software of the data output destination, and outputs the data. You may make it

A database unit 200 connected to the computer system 100 may store, for example, waveform data acquired from the brain in a plurality of known states of a subject. Waveform data obtained from a subject's brain in a plurality of known states may be stored in database unit 200 in association with known compounds that induce the state, for example. The database unit 200 may store, for example, data output by the computer system 100 (eg, predicted target states).

In the example shown in FIG. 1, the database section 200 is provided outside the computer system 100, but the present invention is not limited to this. It is also possible to provide the database unit 200 inside the computer system 100 . At this time, the database unit 200 may be implemented by the same storage means as the storage means in which the memory 130 is installed, or by a storage means different from the storage means in which the memory 130 is installed. In any event, database unit 200 is configured as a storage unit for computer system 100 . The configuration of database unit 200 is not limited to a specific hardware configuration. For example, the database unit 200 may be configured with a single hardware component, or may be configured with a plurality of hardware components. For example, the database unit 200 may be configured as an external hard disk device of the computer system 100, or configured as a cloud storage connected via a network.

FIG. 2 shows an example of the configuration of the processor 120. As shown in FIG.

Processor 120 comprises at least burst analysis means 121 .

The burst analysis means 121 is configured to perform burst analysis on the waveform data received by the reception means 110 .

The burst analysis means 121 comprises division means, acquisition means, detection means, and derivation means.

The dividing means is configured to divide the waveform data into a plurality of frequency bands. The dividing means can divide the waveform data into a plurality of frequency bands by any known filtering process. The dividing means can divide the waveform data into a plurality of frequency bands using, for example, a plurality of bandpass filters.

The plurality of frequency bands includes at least five frequency bands. The at least five frequency bands are a theta waveband from about 5 to about 8 Hz, an alpha waveband from about 8 to about 14 Hz, a beta waveband from about 15 to about 30 Hz, a gamma waveband from about 30 to about 50 Hz, and It includes the high-gamma waveband from about 70 to about 150 Hz. By subdividing to include these frequency bands, it becomes possible to capture the unique features of a particular condition (or disease) that may appear in at least one of these frequency bands. The plurality of frequency bands may include a band from about 150 to about 200 Hz in addition to the at least five frequency bands mentioned above.

The acquisition means is configured to acquire spike data of each frequency band by detecting spikes in waveform data of each frequency band of the plurality of frequency bands. A spike in waveform data is defined as an oscillation in which the waveform data amplitude (or amplitude deviation) exceeds a predetermined threshold, and the start point of a spike is defined as the time when the amplitude (or amplitude deviation) exceeds a predetermined threshold. and the end point of the spike is defined as the time when the spike acquisition width has passed from the start point of the spike. Here, the predetermined threshold can be any value. For example, the predetermined threshold value may be the median amplitude value, the average amplitude value, or a value obtained by multiplying the median amplitude value or the average amplitude value by a coefficient. The spike acquisition width can be any duration. The spike acquisition width may be constant or may vary. The spike acquisition width, for example, preferably varies for each of a plurality of frequency bands. This is because the spike time width (or spike duration) of the spikes in the waveform data varies depending on the frequency band, and thus erroneous detection of spikes can be reduced or prevented by appropriately setting the spike acquisition width. . The spike acquisition width can be, for example, the reciprocal of the representative frequency of each frequency band. Here, the representative frequency of each frequency band can be, for example, any frequency within the frequency band, such as the minimum value, maximum value, average value, etc. within the frequency band. For example, for the high-gamma waveband of about 70 to about 150 Hz, the spike acquisition width can be about 1/70 sec, about 1/150 sec, about 1/105 sec, or about 1/100 sec (10 msec). For example, when the spike in the high-gamma waveband of about 70 to about 150 Hz has a spike width of 10 msec, and the spike acquisition width is set to 1 msec, which is the reciprocal of 1000 Hz (i.e., when the spike acquisition width is not appropriate ), 10 spikes are erroneously detected every 1 msec for 10 msec in which one spike should be detected.

Acquisition means, for example, from the waveform data of each frequency band of a plurality of frequency bands, by detecting the time when the amplitude (or amplitude deviation) of the waveform data exceeds a predetermined threshold and the time after the spike acquisition width has passed, Spikes can be detected. The acquisition means can acquire spike data based on a plurality of spikes detected within the signal acquisition time. Spike data may include total spike number N and total spike time T, as described below.

The detection means is configured to detect bursts in the spike data acquired by the acquisition means. The detection means may detect bursts in the spike data, for example, by techniques described with reference to FIGS. 3A-3D. In Figures 3A-3D, the horizontal axis represents the time axis and the spikes are plotted as black dots.

First, the detection means calculates an average ISI (Inter spike interval) from spike data of total spike number N and total spike time T, as shown in FIG. 3A. In the example shown in FIG. 3A, the first spike occurs at t ₁ and the Nth spike occurs at t _N , total spike time T=t _N −t ₁ . The average ISI is calculated as T/(N-1).

The detection means then identifies consecutive spikes with ISI less than or equal to the average ISI as burst candidates, as shown in FIG. 3B. A burst candidate has a start time and an end time. In the example shown in FIG. 3B, two burst candidates (a first burst candidate and a second burst candidate) have been identified, the first burst candidate and the second burst candidate having start times start ₁ , respectively. It has start ₂ and end times end ₁ , end ₂ .

Next, the detection means calculates each Poisson surprise while changing the tail spike of each burst candidate to the direction of the leading spike, as shown in FIG. Determined as Here, the Poisson surprise is an index representing how low the probability of an event is in the Poisson distribution, and the lower the probability, the larger the value.

The detection means then calculates the respective Poisson surprise while polarizing the leading spike of each burst candidate in the trailing spike direction, and selects the spike with the maximum Poisson surprise as the leading spike of that burst candidate, as shown in FIG. 3D. is determined as Burst candidates having leading and trailing spikes thus determined are detected as bursts.

The deriving means is configured to derive data for burst parameters for each frequency band based on the bursts detected by the detecting means. Burst parameters are e.g.
(1) firing frequency (Spike rate),
(2) Burst rate,
(3) burst interval (IBI);
(4) burst period (Duration),
(5) Spikes in burst,
(6) maximum amplitude in burst (Peak amplitude),
(7) maximum amplitude interval (IPI);
(8) CV value of burst interval (CV of IBI) (coefficient of variation (CV) = standard deviation / average value),
(9) CV value of burst period (CV of Duration),
(10) CV value of spikes in burst (CV of spikes in burst),
(11) CV value of maximum amplitude in burst (CV of Peak amplitude), (12) CV value of maximum amplitude interval (CV of IPI)
at least one of

(1) The firing frequency represents the frequency of spikes and is expressed by (total number of spikes)/(signal acquisition time). (2) Burst frequency represents the frequency of bursts and is expressed by (total number of bursts)/(signal acquisition time). (3) The burst interval represents the interval between bursts, and is represented by the average value of the time from the time of the last spike of the previous burst to the time of the first spike of the next burst for all bursts. (4) The burst period represents the duration of the burst, and is represented by the average value of all bursts from the time of the leading spike to the time of the trailing spike. (5) Number of spikes in burst represents the average number of spikes in a burst over all bursts. (6) Maximum intra-burst amplitude represents the average value of the maximum amplitude in all bursts within the burst period. 7) Maximum Amplitude Interval represents the average value of the interval between maximum intra-burst amplitudes over all bursts.

The burst parameters are preferably (8) CV value of burst interval (CV of IBI) (coefficient of variation (CV) = standard deviation/mean value), (9) CV value of burst duration (CV of Duration), (10 ) CV value of spike number in burst (CV of spikes in burst), (11) CV value of maximum amplitude in burst (CV of Peak amplitude), (12) CV value of maximum amplitude interval (CV of IPI) The burst parameters, further preferably comprising at least one of (8) burst interval CV value (CV of IBI) (coefficient of variation (CV) = standard deviation/mean value), (9) burst duration CV value ( (10) CV value of the number of spikes in burst (CV of Spikes in burst), (11) CV value of maximum amplitude within burst (CV of Peak amplitude), and (12) CV value of maximum amplitude interval (CV of IPI). This is because these burst parameters have a higher probability of becoming burst parameters that can separate the target state than other burst parameters, and can be effective parameters.
including.

The data for the derived burst parameters for each frequency band can be utilized by processor 120 to predict target conditions. Waveform data obtained from the brain can only be analyzed qualitatively as it is. , quantitative analysis (for example, multivariate analysis) can be performed on the waveform data acquired from the brain. Multivariate analysis, which is an example of quantitative analysis, will be described later.

Referring again to FIG. 2 , processor 120 may further comprise multivariate analysis means 122 , prediction parameter identification means 123 and prediction means 124 .

The multivariate analysis means 122 is configured to perform multivariate analysis on the input data. The multivariate analysis means 122 performs multivariate analysis on the waveform data acquired from the brain of the first subject and the waveform data acquired from the brain of the second subject analyzed by the burst analysis means 121 . The first target and the second target have different states. Multivariate analysis is, for example, principal component analysis. Principal component analysis can calculate at least a first principal component and a second principal component of the data. For example, the 3rd principal component, the 4th principal component, . Multivariate analysis is, for example, cluster analysis. Cluster analysis allows data to be classified into multiple clusters. As an algorithm for cluster analysis, for example, the ward method (internal square distance) can be used, and average distance, distance between centroids, maximum distance, and shortest distance can also be used. As an index for determining the degree of similarity, for example, Euclidean distance, Mahalanobis distance, and Spearman's rank correlation can be used. For example, cosine similarity can be used as an index for determining similarity. As an index for determining similarity, for example, a combination of Euclidean distance and cosine similarity can be used. The threshold value of the index for determining similarity can be set to any value according to the data.

Data input to the multivariate analysis means 122 may include, for example, data obtained from different brains when the same compound is administered to different subjects. By processing data from a plurality of different subjects with the multivariate analysis means 122, the prediction parameter specifying means 123, which will be described later, can specify a prediction parameter set that does not depend on inter-specimen differences.

The multivariate analysis means 122, for example, for a combination of burst parameter data derived by the burst analysis means 121 from the waveform data acquired from the brain of the first subject and the waveform data acquired from the second brain Perform multivariate analysis. The multivariate analysis means 122 may, for example, perform multivariate analysis on a combination of burst parameter data derived from waveform data acquired from brains of three or more subjects. The multivariate analysis means 122 may, for example, perform multivariate analysis on a combination of burst parameter data in a specific frequency band derived by the burst analysis means 121, or Alternatively, multivariate analysis may be performed on combinations of burst parameter data for all frequency bands.

A parameter set for multivariate analysis must include three or more parameters because multivariate analysis cannot be performed with two or less parameters. For example, if the burst parameters consist of the 12 burst parameters mentioned above, then all combinations of each burst parameter result in 4017 parameter sets (each parameter set containing 3-12 burst parameters). At this time, the multivariate analysis means 122 performs multivariate analysis on the waveform data acquired from the brain of the first subject and the waveform data acquired from the second brain for each of the 4017 parameter sets. .

The prediction parameter identification means 123 is configured to identify a prediction parameter set based on the results of multivariate analysis by the multivariate analysis means 122 . The predictive parameter specifying means 123 specifies, as a predictive parameter set, a combination of burst parameters that can separate the first state and the second state, for example, based on the results of multivariate analysis. For example, the first condition may be a condition with a given symptom, and the second condition may be a condition without a given symptom. For example, the first condition can be a condition having a given symptom and the second condition can be a condition having a precursor to a given symptom. For example, the first condition can be a condition without a given symptom and the second condition can be a condition that has a precursor to a given symptom.

The predictive parameter specifying means 123 may specify, as a predictive parameter set, a combination of parameters capable of separating three or more states, for example, based on the results of multivariate analysis. For example, the first state is a state with a predetermined symptom, the second state is a state without the predetermined symptom, and the third state is a precursor of the predetermined symptom. can be in a state of having

For example, if the multivariate analysis is a principal component analysis, the result of the principal component analysis for a combination of burst parameters is the first principal component of waveform data obtained from the brain of a subject having a first known state. Significance test is performed on the first principal component score of the waveform data acquired from the brain of the subject having the score and the second known state, and if a significant difference is found, the combination of the burst parameters is used as a prediction parameter set. can be specified as For example, from the results of principal component analysis for a certain burst parameter combination, a first principal component score of waveform data obtained from the brain of a subject with a first known state, a subject with a second known state The first principal component score of the waveform data obtained from the brain of . If a significant difference is recognized, the combination of burst parameters can be identified as a predictive parameter set. Such a significance test is performed for all combinations of burst parameters for which multivariate analysis has been performed, and one or more combinations of burst parameters for which a significant difference is observed are specified as prediction parameter sets.

The prediction parameter specifying means 123 can perform a significance test using any significance test method. The significance test can be performed using, for example, ANOVA (analysis of variance), and in the case of multivariate, MANOVA (multivariate analysis of variance). For example, ANOVA (analysis of variance) or MANOVA (multivariate analysis of variance) can be performed and a significant difference can be recognized if the resulting p-value is 0.05 or less. The p-value threshold at which a significant difference is recognized is not limited to 0.05, and can be any value.

For example, if the multivariate analysis is a cluster analysis, the results of the cluster analysis for a combination of burst parameters are waveform data obtained from the brain of a subject having a first known state and a second known state. When the waveform data acquired from the brain of the subject having the parameters are classified into different clusters, the combination of burst parameters can be specified as the predictive parameter set. For example, cluster analysis results for a combination of burst parameters, waveform data obtained from the brain of a subject with a first known state, waveform data obtained from the brain of a subject with a second known state, , . . . , if all the waveform data acquired from the brain of the subject with the nth known state are classified into another cluster, that combination of burst parameters can be identified as the predictive parameter set. In this way, cluster analysis is performed on all combinations of burst parameters, and one or more combinations of burst parameters in which all data are classified into different clusters are identified as predictive parameter sets.

It may be preferable for the prediction parameter specifying means 123 to specify a plurality of prediction parameter sets that can separate the states of the plurality of sets. This is because it becomes possible to predict a plurality of states. It may be further preferable that the prediction parameter identifying means 123 identify multiple prediction parameter sets each separable from multiple sets of related states. This is because stepwise separation of a plurality of sets of related states enables step-by-step prediction of the state, thus enabling highly accurate state prediction.

The prediction means 124 is configured to predict the state of the target based on the prediction parameter set identified by the prediction parameter identification means 123 from the input data. The prediction means 124 receives the prediction parameter set from the prediction parameter set 123 and also receives waveform data acquired from the target brain analyzed by the burst analysis means 121 .

The predictor 124 may, for example, determine whether the waveform data obtained from the brain of the target is waveform data obtained from the brain of the subject having a first known state or the brain of the subject having the second known state. The first known state or the second known state thus determined is predicted as the target state.

Predictor 124 performs, for example, a multivariate analysis on the waveform data obtained from the target brain for the prediction parameter set, and based on the results, the waveform data obtained from the subject's brain having a first known state. A determination may be made as to whether the waveform data or waveform data obtained from the brain of a subject having a second known condition is similar. For example, when the multivariate analysis is a principal component analysis, the prediction means 124 calculates the principal component scores of the waveform data obtained from the target brain for the prediction parameter set, and plots the principal component scores. Determine similarity based on For example, similarity is determined by Euclidean distance between principal component plots, cosine similarity, or a combination thereof. For example, if the multivariate analysis is a cluster analysis, the prediction means 124 includes waveform data obtained from the brain of a subject having a first known state, a second known state, for the prediction parameter set. Waveform data obtained from the brain of the subject and waveform data obtained from the brain of the subject are cluster-analyzed to obtain waveform data obtained from the brain of the subject having a first known state and a second known state. and the waveform data obtained from the brain of the target are classified into clusters. Classifying the waveform data acquired from the brain of the target into the same cluster as the waveform data acquired from the brain of the subject having the first known state or the waveform data acquired from the brain of the subject having the second known state. determine similarity based on whether or not The waveform data obtained from the brain of the target is then obtained from the brain of the subject with the first known state or the cluster into which the waveform data obtained from the brain of the subject with the first known state is sorted. If the acquired waveform data did not fall into any of the classified clusters, then the waveform data acquired from the brain of the target is the waveform data acquired from the brain of the subject having the first known condition or It will not resemble any waveform data acquired from the brain of a subject with a second known condition.

In the above example, it was explained that the processor includes the multivariate analysis means 122 and the prediction parameter identification means 123, but when the prediction parameter set is determined in advance, the processor includes the multivariate analysis means 122 and the prediction The parameter specifying means 123 may not be provided. In this case, the processor receives a predetermined prediction parameter set and predicts the characteristic based on the received prediction parameter set.

For example, the processor generates waveform data obtained from the target brain, waveform data obtained from the subject brain having a first known state, and a second known state for a predetermined prediction parameter set. performing a multivariate analysis on the waveform data obtained from the brain of the subject having a first known condition, the waveform data obtained from the target brain based on the results obtained from the brain of the subject having determining whether it is similar to either the waveform data or the waveform data obtained from the brain of the subject having a second known state, and determining the first known state or the second known state determined to be similar, Predict as target state. For example, for a predetermined predictive parameter set, waveform data obtained from a target brain, waveform data obtained from a subject having a first known state, and waveform data obtained from a subject having a second known state. performing principal component analysis on the waveform data obtained from the brain, and based on the principal component plot (e.g., based on the similarity S), waveform data obtained from the brain of the subject having a first known state; or determine which of the waveform data acquired from the brain of a subject having a second known state is similar, and determine whether the first known state or the second known state determined to be similar to the target Predict as a state. For example, for a predetermined predictive parameter set, waveform data obtained from a target brain, waveform data obtained from a subject having a first known state, and waveform data obtained from a subject having a second known state. performing a cluster analysis on the waveform data obtained from the brain, wherein the waveform data obtained from the brain of the target has a first known state or the waveform data obtained from the brain of the subject has a second known state; Waveform data obtained from the brain of a subject having a first known state or having a second known state based on whether they fall into the same cluster as the waveform data obtained from the brain of the subject having It is determined which of the waveform data acquired from the subject's brain is similar, and the first known state or second known state determined to be similar is predicted as the target state.

In the example described above, each component of the computer system 100 is provided within the computer system 100, but the present invention is not limited to this. Any of the components of computer system 100 may be provided external to computer system 100 . For example, if the processor 120 and the memory 130 are each composed of separate hardware components, each hardware component may be connected via any network. At this time, the type of network does not matter. Each hardware component may be connected via a LAN, wirelessly, or wired, for example. Computer system 100 is not limited to any particular hardware configuration. For example, it is within the scope of the invention for processor 120 to be implemented with analog circuitry rather than digital circuitry. The configuration of computer system 100 is not limited to the above as long as it can implement its functions.

Although each component of the processor 120 is provided in the same processor 120 in the above example, the present invention is not limited to this. It is also within the scope of the present invention for each component of processor 120 to be distributed across multiple processor units.

4. Processing by Computer System for Predicting Target State FIG. 4 shows an example of processing in the computer system 100 for predicting target state. The example shown in FIG. 4 describes a process 400 for performing burst analysis on waveform data acquired from the brain.

After the receiving means 110 receives the waveform data obtained from the brain, in step S401 the burst analysis means 121 of the processor 120 divides the waveform data obtained from the brain into a plurality of frequency bands. The burst analysis means 121 can divide the waveform data into a plurality of frequency bands by any known filtering process, for example. The burst analysis means 121 can divide the waveform data into a plurality of frequency bands using, for example, a plurality of bandpass filters.

The plurality of frequency bands includes at least five frequency bands. The at least five frequency bands are a theta waveband from about 5 to about 8 Hz, an alpha waveband from about 8 to about 14 Hz, a beta waveband from about 15 to about 30 Hz, a gamma waveband from about 30 to about 50 Hz, and high-gamma waveband from about 70 to about 150 Hz. By subdividing to include these frequency bands, it becomes possible to capture the unique features of a particular condition (or disease) that may appear in at least one of these frequency bands. The plurality of frequency bands may include a band from about 150 to about 200 Hz in addition to the at least five frequency bands mentioned above.

FIG. 5 shows an example of electroencephalogram data divided into multiple frequency bands. In FIG. 5, waveform data obtained from the brain of subjects administered vehicle (top of FIG. 5) and 4-aminopyridine (4-AP) 6 mg/kg. FIG. 5 shows divided waveform data when the waveform data (lower side of FIG. 5) is divided into six frequency bands. The six frequency bands are a theta wave band of about 5 to about 8 Hz, an alpha wave band of about 8 to about 14 Hz, a beta wave band of about 15 to about 30 Hz, a gamma wave band of about 30 to about 50 Hz, and a gamma wave band of about 70 to about a high-gamma waveband at 150 Hz and a band from about 150 to about 200 Hz. Waveform data is divided using band-pass filters corresponding to these frequency bands.

Referring again to FIG. 4, in step S402, burst analysis means 121 detects spikes in the waveform data of each of the plurality of frequency bands divided in step S401, thereby generating spike data of each frequency band. to get For example, the burst analysis means 121 detects the time when the amplitude (or amplitude deviation) of the waveform data exceeds a predetermined threshold and the time after the spike acquisition width has elapsed from the waveform data of each frequency band of a plurality of frequency bands. can detect spikes and acquire spike data based on multiple spikes detected within the signal acquisition time. Here, the spike acquisition width may be constant or may vary. The spike acquisition width, for example, preferably varies for each of a plurality of frequency bands. This is because the spike width (or spike duration) of the spikes in the waveform data varies depending on the frequency band, and thus erroneous detection of spikes can be reduced or prevented by appropriately setting the spike acquisition width. The spike acquisition width can be, for example, the reciprocal of the representative frequency of each frequency band.

In step S403, the burst analysis means 121 detects bursts in the spike data acquired in step S402 for each of the plurality of frequency bands. The detection means may detect bursts in the spike data, for example, by the techniques described above with reference to Figures 3A-3D.

In step S404, the burst analysis means 121 derives data for burst parameters for each of the plurality of frequency bands based on the burst detected in step S403. Burst parameters are e.g.
(1) firing frequency (Spike rate),
(2) Burst rate,
(3) burst interval (IBI);
(4) burst period (Duration),
(5) Spikes in burst,
(6) maximum amplitude in burst (Peak amplitude),
(7) maximum amplitude interval (IPI);
(8) CV value of burst interval (CV of IBI) (coefficient of variation (CV) = standard deviation / average value),
(9) CV value of burst period (CV of Duration),
(10) CV value of spikes in burst (CV of spikes in burst),
(11) CV value of maximum amplitude in burst (CV of Peak amplitude), (12) CV value of maximum amplitude interval (CV of IPI)
at least one of

The data for the derived burst parameters for each frequency band can be utilized, such as in process 600 described below, to predict target conditions.

FIG. 6 shows an example of a process in computer system 100 for predicting the state of a target. The example shown in FIG. 6 describes a process 600 for predicting the state of a target.

At step S601, the computer system 100 receives, via the receiving means 110, first waveform data obtained from the brain of a subject having a first state. The received first waveform data is passed to processor 120 .

At step S602, the computer system 100 receives, via the receiving means 110, the second waveform data acquired from the brain of the subject having the second state. The received second waveform data is passed to processor 120 .

The second state is a state different from the first state. For example, the first condition may be a condition with a given symptom, and the second condition may be a condition without a given symptom. For example, the first condition can be a condition having a given symptom and the second condition can be a condition having a precursor to a given symptom. For example, the first condition can be a condition without a given symptom and the second condition can be a condition that has a precursor to a given symptom.

A given condition can be, for example, a neurological disease. Neurological disorders include, but are not limited to, convulsions, epilepsy, ADHD, dementia, autism, schizophrenia, depression, and the like.

In step S603, the processor 120 performs the processing 400 described above to obtain data for burst parameters for each of the first waveform data and the second waveform data, and performs multivariate analysis on a combination of these data. to identify the prediction parameter set. For example, the burst analysis means 121 of the processor 120 obtains data for burst parameters for each of the first waveform data and the second waveform data, and the multivariate analysis means 122 of the processor 120 analyzes the combination of burst parameter data. A multivariate analysis is performed on the set of prediction parameters, and the prediction parameter specifying means 123 of the processor 120 specifies a prediction parameter set based on the results of the multivariate analysis.

Once the predictive parameter set has been identified and the computer system 100 receives, via the receiving means 110, the target waveform data obtained from the brain of the target, at step S604 the processor 120 determines from the target waveform data identified at step S603 Predict the state of the target based on the predicted parameter set. For example, predictor 124 of processor 120 predicts the state of the target.

FIG. 7 shows an example of the process by which the processor 120 predicts the state of the target in step S604.

At step S701, the prediction means 124 of the processor 120 performs multivariate analysis of the target waveform data obtained from the target brain for the set of prediction parameters identified at step S703. This allows the target waveform data to be compared with the results of multivariate analysis on predictive parameter sets previously performed on the first waveform data and the second waveform data. Multivariate analysis is, for example, principal component analysis. For example, the prediction means 124 calculates the principal component score of the target waveform data by principal component analysis. Multivariate analysis is, for example, cluster analysis. For example, the predictor 124 classifies the target waveform data into clusters by cluster analysis.

The multivariate analysis performed in step S701 may be the same method as the multivariate analysis performed in step S603, or may be a different method. For example, when the prediction parameter set is specified by performing principal component analysis in step S603, in step S701, the same principal component analysis may be performed, or a different multivariate analysis (eg, cluster analysis) may be performed. good. For example, when the prediction parameter set is identified by performing cluster analysis in step S603, in step S701, the same cluster analysis may be performed, or a different multivariate analysis (e.g., principal component analysis) may be performed. .

At step S602, the prediction means 124 of the processor 120 determines whether the target waveform data is more similar to the first waveform data than to the second waveform data.

The prediction means 124 can determine whether or not they are similar based on the principal component scores calculated in step S701, for example. For example, the degree of similarity between the target waveform data and the first waveform data and the degree of similarity between the waveform data and the second waveform data can be calculated based on the principal component plot created based on the principal component scores.

For example, plot the first and second principal component scores of the first waveform data, plot the first and second principal component scores of the second waveform data, and target After plotting the first principal component score and the second principal component score of the waveform data, the Euclidean distance between the plot of the target waveform data and the first waveform data, and the distance between the plot of the target waveform data and the second waveform data. By comparing the Euclidean distance with the plot, it can be determined whether the target waveform data is more similar to the first waveform data than the second waveform data. For example, by comparing the cosine similarity between the target waveform data plot and the first waveform data plot and the cosine similarity between the target waveform data plot and the second waveform data plot, the target waveform data is more similar to the first waveform data than the second waveform data. For example, comparing the Euclidean distance and cosine similarity between the target waveform data plot and the first waveform data plot and the Euclidean distance and cosine similarity between the target waveform data plot and the second waveform data plot. Thus, it can be determined whether the target waveform data is more similar to the first waveform data than the second waveform data. When a plurality of points are plotted by a plurality of first waveform data or a plurality of second waveform data, the average value of all plots is calculated, and the Euclidean Distance or cosine similarity may be obtained.

で表される。Preferably, similarity is determined by a combination of Euclidean distance and cosine similarity of principal component plots. For example, as shown in FIG. 8, the Euclidean distance d ₁ between the plot of the target (target waveform data) and the plot of the target having the first state (first waveform data) and the target (target waveform data ) and the plot of the object (second waveform data) having the _second state are equal, the similarity can be determined by the cosine similarity. Using a combination of the Euclidean distance and cosine similarity of the principal component plot, the similarity S is

is represented by

The prediction means 124 determines whether the target waveform data is more similar to the first waveform data than to the second waveform data, for example, based on which cluster the target waveform data was classified in step S801. can be done.

If it is determined in step S702 that the target waveform data is more similar to the first waveform data than to the second waveform data, then flow proceeds to step S703 to identify the first state as the target state. For example, if the first condition is "a condition having symptoms of neurological disease", the target condition is identified as "a condition having symptoms of neurological disease". For example, if the first condition is "a condition with symptoms of convulsions", the target condition is identified as "a condition with symptoms of convulsions".

It was determined in step S702 that the target waveform data is less similar to the first waveform data than the second waveform data, i.e., the target waveform data is more similar to the second waveform data than the first waveform data. If so, the process proceeds to step S704 to identify the second state as the target state. For example, if the second condition is "a condition without symptoms of neurological disease", the target condition is identified as "a condition without symptoms of neurological disease". For example, if the second condition is an "asymptomatic condition of convulsions", the target condition is identified as an "asymptomatic condition of convulsions".

In the example described above, process 600 was performed using first waveform data obtained from the brain of a subject having a first state and second waveform data obtained from the brain of a subject having a second state. but in addition to this, the third waveform data obtained from the brain of the subject having the third state, the fourth waveform data obtained from the brain of the subject having the fourth state, . Process 600 may be performed using the n-th waveform data (where n is an integer of 3 or more) obtained from the brain of a subject having the state of . At this time, in step S702, the degree of similarity between the target waveform data and each of the first waveform data, second waveform data, . . . nth waveform data is compared.

After the target state is identified in step S703 or step S704, step S604 may be repeated using a second set of prediction parameters. At this time, the second prediction parameter set, for example, using the waveform data obtained from the brain of the subject having the third state and the waveform data obtained from the brain of the subject having the fourth state, steps S601 to It may be a prediction parameter set specified by performing the process of step S603. Alternatively, the second predictive parameter set comprises waveform data obtained from the brain of the subject having the first state and the third state if the first state was identified as the target state in step S703. It may be a prediction parameter set identified by performing the processing of steps S601 to S603 using waveform data acquired from the brain of the target, or the second state in step S704 was identified as the target state. In this case, the waveform data obtained from the brain of the subject having the second state and the waveform data obtained from the brain of the subject having the third state are used to perform the processing of steps S601 to S603. It may be a prediction parameter set that has been set. This causes the second prediction parameter set to correspond to the state identified in step S703 or step S704. This can be preferable in that it allows the status to be identified step by step.

For example, if the target condition is identified as "neurological disorder condition" in step S703 or step S704, electroencephalogram data obtained from the brain of a subject with "ADHD" symptoms and "integrated Step S604 is repeated using the second prediction parameter set specified by performing the processing of steps S601 to S603 using the waveform data acquired from the brain of the subject having the symptoms of "ataxia" can be done. As a result, it is possible to specify whether the "condition of having a neurological disease" has a symptom of "ADHD" or a symptom of "schizophrenia". This can be utilized for target diagnosis or as an indicator for target diagnosis.

FIG. 9 shows an example of a process in computer system 100 for predicting the state of a target. The example shown in FIG. 9 describes a process 900 for predicting the state of a target when the prediction parameter set is predetermined. The predetermined prediction parameter set includes at least one of the burst parameters described above.

In step S901, the computer system 100, via the receiving means 110, first waveform data obtained from the brain of the subject having the first state, second waveform data obtained from the brain of the subject having the second state, Receiving waveform data, target waveform data obtained from the brain of the target. The received waveform data is passed to processor 120 .

In step S902, the burst analysis means 121 of the processor 120 derives data for burst parameters included in the prediction parameter set from each waveform data, and the prediction means 124 of the processor 120 performs multivariate analysis on each data. I do. The burst analysis means 121 can derive data for burst parameters by process 400 . Multivariate analysis is, for example, principal component analysis. For example, the prediction unit 124 calculates the principal component score of the first waveform data, the principal component score of the second waveform data, and the principal component score of the target waveform data by principal component analysis. Multivariate analysis is, for example, cluster analysis. For example, the predictor 124 classifies the target waveform data into clusters by cluster analysis.

At step S903, the prediction means 124 of the processor 120 determines whether the target waveform data is more similar to the first waveform data than to the second waveform data. The prediction means 124 can determine whether or not they are similar based on the principal component scores calculated in step S902, for example. For example, the degree of similarity between the target waveform data and the first waveform data and the degree of similarity between the target waveform data and the second waveform data can be calculated based on the principal component plot created based on the principal component scores. .

For example, plot the first and second principal component scores of the first waveform data, plot the first and second principal component scores of the second waveform data, and target After plotting the first principal component score and the second principal component score of the waveform data, the Euclidean distance between the plot of the target waveform data and the first waveform data, and the distance between the plot of the target waveform data and the second waveform data. By comparing the Euclidean distance with the plot, it can be determined whether the target waveform data is more similar to the first waveform data than the second waveform data. For example, by comparing the cosine similarity between the target waveform data plot and the first waveform data plot and the cosine similarity between the target waveform data plot and the second waveform data plot, the target waveform data is more similar to the first waveform data than the second waveform data. For example, comparing the Euclidean distance and cosine similarity between the target waveform data plot and the first waveform data plot and the Euclidean distance and cosine similarity between the target waveform data plot and the second waveform data plot. Thus, it can be determined whether the target waveform data is more similar to the first waveform data than the second waveform data. When a plurality of points are plotted by a plurality of first waveform data or a plurality of second waveform data, the average value of all plots is calculated, and the Euclidean Distance or cosine similarity may be obtained.

The prediction means 124 determines whether the target waveform data is more similar to the first waveform data than to the second waveform data, for example, based on which cluster the target waveform data was classified in step S902. can be done.

If it is determined in step S903 that the target waveform data is more similar to the first waveform data than the second waveform data, then flow proceeds to step S904 to identify the first state as the target state. Identify the first state as the target state. For example, if the first condition is "a condition having symptoms of neurological disease", the target condition is identified as "a condition having symptoms of neurological disease". For example, if the first condition is "a condition with symptoms of convulsions", the target condition is identified as "a condition with symptoms of convulsions".

It was determined in step S903 that the target waveform data is less similar to the first waveform data than the second waveform data, i.e., the target waveform data is more similar to the second waveform data than the first waveform data. If so, the process proceeds to step S905 to identify the second state as the target state. For example, if the second condition is "a condition without symptoms of neurological disease", the target condition is identified as "a condition without symptoms of neurological disease". For example, if the second condition is an "asymptomatic condition of convulsions", the target condition is identified as an "asymptomatic condition of convulsions".

In the example described above, process 900 was performed using first waveform data obtained from the brain of a subject having a first state and second waveform data obtained from the brain of a subject having a second state. but in addition to this, the third waveform data obtained from the brain of the subject having the third state, the fourth waveform data obtained from the brain of the subject having the fourth state, . The process 900 may be performed using the n-th waveform data (where n is an integer of 3 or more) acquired from the brain of a subject having the state of . At this time, in step S903, the degree of similarity between the first waveform data, the second waveform data, . . . and the n-th waveform data is compared.

It should be noted that although the above examples describe operations occurring in a particular order, the order of operations is not limited to that described and may occur in any order that is logically possible.

In the examples described above with reference to FIGS. 4, 6, 7, and 9, the processing of each step shown in FIGS. , the present invention is not limited to this. At least one of the processing of each step shown in FIGS. 4, 6, 7, and 9 may be implemented by a hardware configuration such as a control circuit.

In the above example, the computer system 100 is used to predict the state of the target, but the present invention is not limited to this. It is also within the scope of the present invention for a person to manually predict the state of a target without using the computer system 100 . In this case, for example, processes 600 and 900 may include obtaining data from a subject instead of receiving data (steps S601, S602, S901). Based on the data obtained, subsequent steps can be calculated by hand.

(Example 1)
Electroencephalograms of the corticofrontal lobe were acquired when the convulsion-positive drug 4-aminopyridine (4-AP) was administered to the specimen. Rats were used as specimens. n=5 subjects received vehicle, n=3 subjects received 3 mg/kg 4-AP, and n=3 subjects received 6 mg/kg 4-AP. No seizures were observed in subjects receiving vehicle. No seizures were observed in subjects dosed with 3 mg/kg 4-AP. Convulsive seizures were seen in subjects receiving 6 mg/kg of 4-AP.

The acquired electroencephalograms were analyzed using the computer system 100 of the present invention to derive data for frequency band burst parameters for the high-γ waveband.

FIG. 10A shows this result. Data derived from the EEG for subjects dosed with vehicle are labeled vehicle_ip, data derived from EEG for subjects dosed with 3 mg/kg 4-AP are labeled 4AP-1, 4-AP at 6 mg/kg - Data derived from electroencephalograms of AP-treated subjects are labeled 4AP-2.

From this result, 4AP-1, in which seizures were not observed, was separably distinguished from vehicle_ip, in which seizures were not observed, in parameters such as Spike rate, Burst rate, and IBI. It can be seen that by capturing changes in parameters, it can be used as an indicator of a convulsive sign. That is, it has been suggested that these parameters can serve as a predictive parameter set for predicting whether or not there is a preconvulsive state in the frequency band of the high-γ waveband. By using these parameter sets for prediction, it is believed that early detection of precursory states can be useful for diagnosis, early treatment, and prevention.

(Example 2)
Hippocampal electroencephalograms were obtained when the convulsion-positive drug 4-aminopyridine (4-AP) was administered to the specimen. Rats were used as specimens. n=5 subjects received vehicle, n=5 subjects received 3 mg/kg 4-AP, and n=5 subjects received 6 mg/kg 4-AP. No seizures were observed in subjects receiving vehicle. No seizures were observed in subjects dosed with 3 mg/kg 4-AP. Convulsive seizures were seen in subjects receiving 6 mg/kg of 4-AP.

The acquired electroencephalograms were analyzed using the computer system 100 of the present invention to derive data for frequency band burst parameters for the high-γ waveband.

FIG. 10B shows this result. Data derived from the EEG for subjects dosed with vehicle are labeled vehicle_ip, data derived from EEG for subjects dosed with 3 mg/kg 4-AP are labeled 4AP-1, 4-AP at 6 mg/kg - Data derived from electroencephalograms of AP-treated subjects are labeled 4AP-2.

From this result, 4AP-1, in which seizures were not observed, was separably distinguished from vehicle_ip, in which seizures were not observed, in parameters such as Spike rate, Burst rate, and IBI. It can be seen that by capturing changes in parameters, it can be used as an indicator of a convulsive sign. That is, it has been suggested that these parameters can serve as a predictive parameter set for predicting whether or not there is a preconvulsive state in the frequency band of the high-γ waveband. By using these parameter sets for prediction, it is believed that early detection of precursory states can be useful for diagnosis, early treatment, and prevention.

(Example 3)
Electroencephalograms of the corticofrontal lobe were acquired when the convulsion-positive drug Isoniazid was administered to the specimen. Rats were used as specimens. n=5 subjects received vehicle, n=4 subjects received 150 mg/kg Isoniazid, and n=5 subjects received 300 mg/kg Isoniazid. No seizures were observed in subjects receiving vehicle. No seizures were observed in subjects receiving 150 mg/kg Isoniazid. Seizures were seen in subjects receiving 300 mg/kg Isoniazid.

The acquired electroencephalograms were analyzed using the computer system 100 of the present invention to derive data for frequency band burst parameters for the high-γ waveband.

FIG. 10C shows this result. EEG-derived data for subjects administered vehicle is labeled vehicle_ip, EEG-derived data for subjects administered 150 mg/kg Isoniazid is labeled Iso-1, and 300 mg/kg Isoniazid is labeled Data derived from electroencephalograms of the subjects tested are labeled Iso-2.

From this result, Iso-1, in which seizures were not observed, was separably distinguished from vehicle_ip, in which seizures were not observed, in parameters such as Spike rate, Burst rate, and IBI. It can be seen that by capturing changes in parameters, it can be used as an indicator of a convulsive sign. That is, it has been suggested that these parameters can serve as a predictive parameter set for predicting whether or not there is a preconvulsive state in the frequency band of the high-γ waveband. By using these parameter sets for prediction, it is believed that early detection of precursory states can be useful for diagnosis, early treatment, and prevention.

(Example 4)
10 seizure-positive compounds (4-AP, Isoniazid, Philocapine, PTZ, Strychnine, Aminophylline, Picrotoxin, Tramadol, Bupropion, Venlafaxine), 3 seizure-negative compounds (Acetaminophen, Amoxicillin, Aspirin), and administered vehicle to the specimen. EEG of corticofrontal lobe was obtained. Rats were used as specimens. Electroencephalograms were acquired 3 hours after compound administration. Each compound was administered at the following doses.
・4-AP (3 mg/kg, ip)
- Isoniazid (150 mg/kg, ip)
・Pilocarpine (150 mg/kg, ip)
・PTZ (30 mg/kg, sc)
・Strychnine (1 mg/kg, sc)
・Aminophyline (100 mg/kg, ip)
・Picrotoxin (5 mg/kg, ip)
- Tramadol (50 mg/kg, ip)
- Bupropion (200 mg/kg, sc)
・Venlafaxine (400 mg/kg, po)
- Acetaminophen (1000, 3000 mg/kg, po)
- Amoxicillin (4000 mg/kg, po)
- Aspirin (3000 mg/kg, po)
For convulsion-positive compounds, experiments were conducted at doses that do not lead to convulsions, in order to examine whether a potential convulsive risk can be detected.

The acquired electroencephalograms were analyzed using the computer system 100 of the present invention to derive data for frequency band burst parameters for the gamma waveband and the high-gamma waveband.

FIG. 10D shows this result. Values for each parameter are normalized by the value for vehicle administration. That is, the value of each parameter is shown as a percentage with the value when the vehicle is administered as 100%. In FIG. 10D , darker toward black indicates more than the value when vehicle was administered, lighter toward white indicates less than the value when vehicle was administered, and middle gray indicates It shows that it was similar to the value when vehicle was administered.

For example, when focusing on a single parameter such as Spike rate and Burst rate in the γ wave band, some spasm positive compounds (e.g., 4-AP, Strychnine, etc.) and some negative spasm compounds (Amoxicillin) It could be separably distinguished, but not from other seizure-negative compounds. It has been suggested that it is difficult to detect potential convulsive risk only by focusing on a single parameter because the change tendency of the burst parameter differs depending on the type of compound.

On the other hand, by focusing on a combination of multiple burst parameters, seizure-positive compounds and seizure-negative compounds can be separably distinguished. For example, focusing on a parameter set including the Spike rate in the gamma waveband and the CV value of the burst interval (CV of IBI) separably distinguishes between the seizure-positive compound (4-AP) and all negative seizure-negative compounds. be able to. That is, it is suggested that the Spike rate in the gamma waveband and the CV value of the burst interval (CV of IBI) can be a predictive parameter set for predicting whether 4-AP is convulsive toxicity or convulsiveness. It is Other prediction parameter sets may also be derivable from these results.

(Example 5: Comparison with conventional method)
A comparison was made between the conventional method of utilizing FFT spectral intensity for detecting seizure precursors and seizures, which is conventionally performed in electroencephalogram analysis, and the method of predicting the target state using burst analysis of the present invention.

EEGs were acquired when subjects were administered vehicle and three seizure-positive agents (4-AP, Isoniazid, Pilocarpine). Here, rats were used as specimens. Three seizure-positive agents (4-AP, Isoniazid, Pilocarpine) were administered at 3 mg/kg, 150 mg/kg, and 150 mg/kg, respectively, to induce a proconvulsive state. Three seizure-positive drugs (4-AP, Isoniazid, Pilocarpine) were administered at 6 mg/kg, 300 mg/kg, and 400 mg/kg, respectively, to induce a seizure state. Based on the general condition observation records before and after drug administration, electroencephalogram data were classified into (1) before administration, (2) immediately after administration (before drug effects appear), (3) preconvulsive state, and (4) convulsive seizure state.

In the conventional method, FFT (Fast Fourier Transform) is performed on each of the classified electroencephalogram data to calculate the frequency spectrum. Of the calculated frequency spectra, the total spectral intensity was calculated for the frequency spectrum in the 4-200 Hz band. The sum of spectral intensities of frequency spectra obtained from data before drug administration was set to 100%, and the sum of spectral intensities was normalized.

In the method of predicting the state of a target using burst analysis of the present invention, each acquired electroencephalogram is analyzed using the computer system 100 of the present invention, and bursts of frequency bands for the gamma wave band (30 to 50 Hz) Data were derived for the parameters. Specifically, the acquired electroencephalogram was divided into a plurality of frequency bands, and spike data was acquired by detecting spikes in the waveform data for the gamma wave band (30 to 50 Hz). We then detected bursts in the spike data and derived data for burst parameters based on the detected bursts. In this example, among multiple burst parameters, data were derived for the burst interval (IBI), the maximum amplitude interval (IPI), and the number of spikes in burst.

Using the derived data, an attempt was made to detect the proconvulsant state and the seizure state by comparing the data for the burst parameters during vehicle administration, the proconvulsive state, and the seizure state (Experiment 1).

Furthermore, the effect of vehicle administration on burst parameters was investigated by comparing data on burst parameters derived from data immediately after vehicle administration and data 60 minutes after vehicle administration (Experiment 2).

FIG. 11 shows the results of the conventional method.

FIG. 11(a) shows the results immediately after and 60 minutes after administration of vehicle and three seizure-positive agents (4-AP, Isoniazid, Pilocarpine) administered at 3 mg/kg, 150 mg/kg, and 150 mg/kg, respectively. FIG. 11(b) shows results immediately after and 60 minutes after administration of vehicle and three seizure-positive agents (4-AP, Isoniazid, Pilocarpine), respectively. Results are shown immediately after administration and at the time of convulsive seizure when 6 mg/kg, 300 mg/kg and 400 mg/kg were administered. FIG. 11(c) shows the results at the onset of convulsions when three seizure-positive drugs (4-AP, Isoniazid, Pilocarpine) were administered at 3 mg / kg, 150 mg / kg, and 150 mg / kg, respectively. The results of convulsive seizures when 6 mg/kg, 300 mg/kg and 400 mg/kg of drugs (4-AP, Isoniazid and Pilocarpine) were administered are shown. In each figure, the vertical axis indicates the spectral intensity normalized by the spectral intensity before administration, and the horizontal axis indicates the label. Figures 11(a) and 11(b) show whether there was a significant difference between the results immediately after administration and the results 60 minutes after administration, and between the results immediately after administration and the results at the onset of convulsions. It is shown. "*" indicates that there was a significant difference, and "NS" indicates that there was no significant difference. Figures 11(c) and 11(d) show whether there was a significant difference between the pre-seizure results and the results immediately after vehicle administration. "*" indicates that there was a significant difference, and "NS" indicates that there was no significant difference.

As can be seen from FIGS. 11(c) and 11(d), there were drugs that did not differ significantly from the results immediately after vehicle administration both during preconvulsive and seizure. Drugs with no significant difference could not distinguish between vehicle administration and preconvulsive and seizure times. Conventional methods prove inadequate for detecting preconvulsive and seizure states for a wide range of drugs.

As can be seen from Figures 11(a) and 11(b), there is a significant difference between the results immediately after vehicle administration and the results 60 minutes after vehicle administration. It can be seen that the conventional method is not a stable evaluation system.

Figures 12A and 12B show results from a method of predicting target states using burst analysis of the present invention.

FIG. 12A shows the results for Experiment 1. FIG. 12A(a) is a graph comparing IBI upon administration of 4-AP to IBI upon administration of vehicle, normalized to IBI upon administration of vehicle as 100%. The vertical axis indicates the IBI, and the horizontal axis indicates the label. FIG. 12A(b) is a graph comparing IPI upon administration of Isoniazid to IPI upon administration of vehicle, normalized to IPI upon administration of vehicle as 100%. The vertical axis indicates IPI, and the horizontal axis indicates labels. FIG. 12A(c) is a graph comparing the number of spikes in burst upon administration of Pilocarpine with the number of spikes in burst upon administration of vehicle. is normalized as 100%. The vertical axis indicates the number of spikes within a burst, and the horizontal axis indicates the label. FIG. 12A shows that there was a significant difference between the preconvulsive and seizure results and the vehicle administration results. "*" indicates a significant difference at p<0.05 and "**" indicates a significant difference at p<0.01.

As can be seen from FIG. 12A, for each of 4-AP, Isoniazid, and Pilocarpine, there was a significant difference between the results of vehicle administration and the results during aura and seizures, indicating that burst The burst parameters obtained from the analysis allowed detection of proconvulsive and seizure states for each of the three seizure-positive agents. That is, it can be said that the method of predicting the target state using burst analysis of the present invention is more advantageous than conventional methods in that it can detect preconvulsive states and seizure states for a wide range of drugs. .

In addition, in the conventional method, the parameter is only the spectral intensity, but in the burst analysis, for example, 12 or more types of burst parameters can be calculated for, for example, 5 or more frequency bands. It can be said that the detection sensitivity of the seizure state is excellent. Furthermore, in the method of predicting the target state using burst analysis of the present invention, it is possible to estimate the mechanism of action of a wide range of target compounds using a large number of parameters, and to compare cultured neurons using the same parameters. is. From these points as well, it can be said that the method is more advantageous than the conventional method.

FIG. 12B shows the results for Experiment 2. FIG. 12B(a) is a graph comparing IBI derived from data immediately after vehicle administration and IBI derived from data 60 minutes after vehicle administration. The vertical axis indicates IPI, and the horizontal axis indicates labels. FIG. 12B(b) is a graph comparing IPI derived from data immediately after vehicle administration and IPI derived from data 60 minutes after vehicle administration. The vertical axis indicates IPI, and the horizontal axis indicates labels. FIG. 12B(c) is a graph comparing the number of intra-burst spikes derived from data immediately after vehicle administration and the number of intra-burst spikes derived from data 60 minutes after vehicle administration. The vertical axis indicates the number of spikes within a burst, and the horizontal axis indicates the label. Figures 12B(a)-12B(c) show whether there was a significant difference between the results immediately after vehicle administration and the results 60 minutes after vehicle administration, where "NS" indicates significant. indicates that there was no difference. As can be seen from Figures 12B(a) to 12B(c), no significant difference was observed between the results immediately after vehicle administration and the results 60 minutes after vehicle administration. The method of predicting the state of the target using the burst analysis of the present invention is a stable evaluation system, and it can be said that this point is also more advantageous than the conventional method.

(Example 6)
Five types of seizure-positive compounds (4-AP, Isoniazid, Philocapine, PTZ, Strychnine) and three types of seizure-negative compounds (Acetaminophen, Amoxicillin, Aspirin) were administered to subjects, and electroencephalograms of the corticofrontal lobe were obtained. Rats were used as specimens. Electroencephalograms were acquired 3 hours after compound administration. Each compound was administered at the following doses.
・4-AP (3 mg/kg, ip)
- Isoniazid (150 mg/kg, ip)
・Pilocarpine (150 mg/kg, ip)
・PTZ (30 mg/kg, sc)
・Strychnine (1 mg/kg, sc)
- Acetaminophen (1000 mg/kg, 3000 mg/kg, po)
- Amoxicillin (4000 mg/kg, po)
- Aspirin (3000 mg/kg, po)
For convulsion-positive compounds, experiments were conducted at doses that do not lead to convulsions, in order to examine whether a potential convulsive risk can be detected. As above, acetaminophen was tested at two doses.

The acquired electroencephalograms are analyzed using the computer system 100 of the present invention, and the three burst parameters of the gamma wave band (burst duration (Duration), CV value of maximum amplitude in burst (CV of Peak amplitude), maximum amplitude interval Data were derived for a total of five burst parameters: the CV value (CV of IPI) and two burst parameters in the high-γ waveband (Spike rate, Burst rate). A principal component analysis was performed using the derived data.

By performing principal component analysis, the loading of the first principal component (PC1) and the second principal component (PC2) was obtained as follows.

FIG. 13A shows the results of plotting data for each compound on a two-dimensional plane having a first principal component axis and a second principal component axis, based on loadings obtained by principal component analysis.

As shown in FIG. 13A, 5 seizure-positive compounds were distributed in the 3rd and 4th quadrants, while 3 seizure-negative compounds were distributed in the 1st and 2nd quadrants. In particular, the three seizure-negative compounds are ranged one standard deviation (SD range) and two standard deviations (2SD range) around their mean, with all seizure-negative compounds included within the 2SD range. was Thus, the seizure-positive compound and the seizure-negative compound could be separated by performing the principal component analysis using the five burst parameters. Furthermore, this result suggested the possibility of being able to estimate whether or not the compound is a seizure-negative compound by determining whether or not the plotted data is within the 2SD range. In other words, it was suggested that a seizure-positive compound could be predicted using criteria derived from the results of analysis of electroencephalogram waveforms that did not lead to seizures.

Additional experiments were performed to substantiate this suggestion. Five additional compounds (Aminophyline, Picrotoxin, Tramadol, Venlafaxine, Bupropion) were administered to rats. Each compound was administered at the following doses.
・Aminophyline (100 mg/kg, ip)
・Picrotoxin (5 mg/kg, ip)
- Tramadol (50 mg/kg, ip)
・Venlafaxine (400 mg/kg, po)
- Bupropion (200 mg/kg, sc)
In order to examine whether it is possible to detect a potential convulsive risk, an experiment was conducted at a dose that does not lead to convulsions.

The acquired electroencephalograms were analyzed using the computer system 100 of the present invention, and principal component analysis was performed on each of the burst parameter data obtained, and plotted on the same plane.

FIG. 13B shows the results of additional experiments. Four of the five additional compounds plotted outside the 2SD range. From this result, four additional compounds were judged positive for seizures. In fact, these five types of seizure compounds are all known as seizure-positive compounds, and it can be said that seizure risk could be correctly determined for four types of compounds. The seizure risk could be correctly determined for a total of nine seizure-positive compounds, the first five seizure-positive compounds and the additional four seizure-positive compounds.

From this experiment, it can be seen that the compounds could be separated with generally high accuracy using the criteria derived from the results of analysis of electroencephalogram waveforms that did not lead to convulsions. In other words, it was suggested that it may be possible to establish a new evaluation method for detecting potential seizure risk using electroencephalogram waveforms that do not lead to seizures. This method is considered preferable in that it can reduce the load on the subject.

The invention is not limited to the embodiments described above. It is understood that the invention is to be construed in scope only by the claims. It is understood that a person skilled in the art can implement an equivalent range from the description of specific preferred embodiments of the present invention based on the description of the present invention and common technical knowledge.

INDUSTRIAL APPLICABILITY The present invention provides a burst analysis method that can be used to predict the state of a target, and is useful as a method of predicting the state of a target using burst analysis.

100 Computer System 110 Receiving Means 120 Processor 130 Memory 140 Outputting Means 200 Database Unit

Claims (13)

A method of performing burst analysis on the waveform data in order to quantitatively analyze the waveform data acquired from the brain,
dividing the waveform data into a plurality of frequency bands;
acquiring spike data for each frequency band by detecting spikes in waveform data for each of the plurality of frequency bands;
detecting bursts in the acquired spike data , wherein detecting the bursts comprises:
calculating an average spike interval (average ISI) from the acquired spike data;
identifying each of a plurality of consecutive spikes having an ISI less than or equal to the average ISI as a burst candidate;
calculating the Poisson surprise of each burst candidate while changing the tail spike of the burst candidate toward the head spike direction, and determining the spike with the maximum Poisson surprise as the tail of the burst;
calculating the Poisson surprise of each burst candidate while changing the leading spike of the burst candidate to the trailing spike direction, and determining the spike with the maximum Poisson surprise as the leading spike of the burst;
a step comprising
deriving data for burst parameters for each frequency band based on the detected bursts. The step of dividing the waveform data into a plurality of frequency bands includes a θ wave band of about 5 to about 8 Hz, an α wave band of about 8 to about 14 Hz, a β wave band of about 15 to about 30 Hz, and about 30 to about 50 Hz. and a high-gamma band from about 70 to about 150 Hz. detecting the spike includes detecting a spike having a predetermined duration;
3. The method according to claim 1, wherein said predetermined time duration is the reciprocal of a representative frequency of said divided frequency band. The burst parameters include firing frequency, burst frequency, burst interval, burst duration, number of spikes within burst, maximum amplitude within burst, maximum amplitude interval, CV value of burst interval, CV value of burst duration, CV value of number of spikes within burst. , the CV value of the maximum amplitude within a burst, or the CV value of the maximum amplitude interval. A method according to any one of claims 1 to 3, wherein said burst parameter comprises a CV value of maximum amplitude within a burst or a CV value of maximum amplitude interval. A method of predicting a state of a target, said method being executed in a computer system comprising a processor, said method comprising:
the processor receiving first waveform data obtained from the brain of a subject having a first condition;
the processor receiving second waveform data obtained from the brain of a subject having a second condition;
The processor performs multivariate analysis on burst parameter data combinations derived from the first waveform data and the second waveform data according to the method of any one of claims 1 to 5 . , identifying a prediction parameter set;
said processor predicting said state based on said prediction parameter set from target waveform data obtained from said target brain. further comprising the processor obtaining third waveform data obtained from the brain of the subject having a third condition ;
3. The step of identifying the predictive parameter set comprises performing a multivariate analysis of the first waveform data, the second waveform data, and the third waveform data to identify a predictive parameter set. 6. The method according to 6. The first state is a state having a predetermined symptom,
The second state is a state that does not have the predetermined symptom,
The third state is a state having a precursor of the predetermined symptom,
8. The method of claim 7 , wherein the predictive parameter set comprises a combination of parameters capable of separating the first, second and third states. 9. The method of claim 8 , wherein the predetermined symptom is a convulsive symptom. A computer system for performing burst analysis on waveform data acquired from the brain in order to perform quantitative analysis on the waveform data ,
dividing means for dividing the waveform data into a plurality of frequency bands;
Acquisition means for acquiring spike data in each frequency band by detecting a spike in waveform data in each frequency band of the plurality of frequency bands;
detection means for detecting bursts in the acquired spike data, the detection means comprising:
calculating an average spike interval (average ISI) from the acquired spike data;
identifying each of a plurality of consecutive spikes having an ISI less than or equal to the average ISI as a burst candidate;
calculating the Poisson surprise of each burst candidate while changing the tail spike of the burst candidate toward the head spike direction, and determining the spike with the maximum Poisson surprise as the tail of the burst;
calculating the Poisson surprise of each burst candidate while changing the leading spike of the burst candidate to the trailing spike direction, and determining the spike with the maximum Poisson surprise as the leading spike of the burst;
a detection means configured to perform
derivation means for deriving data for burst parameters for each frequency band based on said detected bursts. A program for performing burst analysis on waveform data acquired from the brain in order to perform quantitative analysis on the waveform data, the program being executed in a computer system comprising a processor, The program
dividing the waveform data into a plurality of frequency bands;
acquiring spike data for each frequency band by detecting spikes in waveform data for each of the plurality of frequency bands;
detecting bursts in the acquired spike data , wherein detecting the bursts comprises:
calculating an average spike interval (average ISI) from the acquired spike data;
identifying each of a plurality of consecutive spikes having an ISI less than or equal to the average ISI as a burst candidate;
calculating the Poisson surprise of each burst candidate while changing the tail spike of the burst candidate toward the head spike direction, and determining the spike with the maximum Poisson surprise as the tail of the burst;
calculating the Poisson surprise of each burst candidate while changing the leading spike of the burst candidate to the trailing spike direction, and determining the spike with the maximum Poisson surprise as the leading spike of the burst;
a step comprising
deriving data for burst parameters for each frequency band based on the detected bursts. A computer system for predicting the state of a target, comprising:
receiving means for receiving first waveform data obtained from the brain of a subject having a first state and second waveform data obtained from the brain of a subject having a second state;
Multivariate analysis is performed on a combination of burst parameter data derived from the first waveform data and the second waveform data according to the method according to any one of claims 1 to 5 , and a prediction parameter set identifying means for identifying
a prediction means for predicting said state based on said prediction parameter set from target waveform data obtained from said target's brain. A program for predicting the state of a target, said program being executed in a computer system comprising a processor, said program comprising:
obtaining first waveform data acquired from the brain of a subject having a first condition;
obtaining second waveform data obtained from the brain of a subject having a second condition;
Multivariate analysis is performed on a combination of burst parameter data derived from the first waveform data and the second waveform data according to the method according to any one of claims 1 to 5 , and a prediction parameter set a step of identifying
predicting the state based on the prediction parameter set from the target waveform data obtained from the brain of the target.

Images