JP2005528937A - Method and apparatus for electromagnetic correction of brain activity - Google Patents

Abstract

The present invention relates to a method of electromagnetically correcting brain activity in order to achieve a predetermined behavioral target, and includes a behavior model expressing a correlation between behavior and transition of brain activity characteristics, and brain activity. Using a brain activity model that expresses quantitatively. The characteristics and transition of brain activity can be derived from the quantitative expression of brain activity. According to the method of the present invention, the nominal transition given by the achievement of the behavioral target and the corresponding brain activity characteristics is determined using a brain model, an extrinsic electric field and / or magnetic field is induced, and the electromagnetic brain activity is A brain activity index is calculated to achieve a nominal transition in a control or adjustment circuit that is measured and includes calculation of a brain model.

Description

  The present invention also relates to a method and apparatus for electromagnetic correction of brain activity, in particular electromagnetic control of brain activity in vivo, each controlled or regulated based on a model, and also as a result behavior correction.

  Definition:

  Each model-based controlled or regulated modification means a change or maintenance of brain activity based on behavioral targets, behavioral models, brain activity models.

  Adjusted means that the measurement, calculation and correction steps are interrelated in the feedback loop.

  Action targets set by the user include the type, intensity, and duration of a particular action.

  A behavior relates to any combination of space and time in any current or possible situation of a person and is potentially verifiable. Behaviors include, but are not limited to, perception, activity, motivation, attention, memory, learning, consciousness, emotion, cognition, mental state, ability, and personality traits.

  Behavior is based on brain activity. The correspondence between the two is quantified within the framework of the behavior model, that is, an experimentally confirmed relationship between a certain behavior and the dynamics of a certain brain activity index.

  The brain activity index is a characteristic of brain activity calculated from measurement data. Examples of brain activity indicators include the potential difference between the EEG electrode and the reference electrode, or the frequency band of the most prominent frequency of these data for all EEG electrodes (“α wave” or similar indicator [12]) or by MEG sensor There is a similarity measure of measured values (see [5]).

  A brain activity model is related to a physiologically based model and describes brain activity by the dynamics and / or interaction of brain composition. (This includes, for example, [24], [1], [23], [25], or [18]).

  Traditional models are not unique but general. This is because the unobservability is evaluated by the average value of the group.

  Unobservability is a general model parameter or variable, which can either be measured in substantial effort only in vivo or not at all (with essential spatial and temporal resolution). Unobservability has a decisive influence on brain mechanics. It varies between different individuals, and there are up to 100 factors per individual. General brain activity models do not comment on direct extrinsic effects on the dynamics of brain organization. In the best case, comment on indirect extrinsic effects from receptor stimulation, loss, and damage.

The following specific brain activity model under normal circumstances is utilized and includes exogenous input by electric and / or magnetic fields. These models were based on unobservability associated with specific users within the time interval and were determined in an appropriate manner. Typically these brain activity models include each type of brain-constituting nonlinear differential equation of this type.

  Where x is a brain activity index, points indicate time derivatives, t is time, f is a function, in particular the type of configuration ("local parameter"), the type of influence of other configurations on the configuration under consideration Range (“endogenous input”), and direct exogenous input (“exogenous input”) on each configuration. Intrinsic inputs include other ("trans-local") parameters, such as coupling strength or transmission delay between configurations, depending on the usage model.

  The lack or incompleteness of brain activity model specification relative to a particular user usually leads to extensive changes in the unknown outcome of the extrinsic electromagnetic input in this case, and thus random outcomes of successful treatment. It leads to the non-applicability of related methods outside the area of neurological, psychiatric, or psychological deficits that are neglected for possibility.

  Electromagnetic means electrical and / or magnetic observability, the type of measurement data (essentially the electromagnetic correlation of the user's brain activity), and the main modes of correction (eg variable magnetic fields generated outside the skull) Through the user's cranium without problems and generating an evoked potential within the cranium, thereby enabling the user's brain activity to be affected).

  In vivo relates to the application of this method to live users as opposed to cell culture, brain slices, computer simulations, and others. This is therefore particularly demanding in terms of the complexity of the actual brain processes that occur, as well as the measurement speed, calculation speed, correction speed and method safety.

  Related technology:

In the current state of the art, electromagnetic modification of each controlled or regulated brain activity based on in vivo models is not possible (the human brain has about 10 ¹¹ neurons and 10 ¹⁵ neurons between them. So far, no consideration or attempt has been made. The same applies to related behavioral modifications.

There are three almost non-overlapping approaches of scientific / technical issues in brain activity modification, each associated with brain activity. :
-A nonlinear dynamics / neural network / artificial intelligence approach in which certain features of the living human brain are mimicked with simplified replication to create a machine with similar features (called a "neurocomputer") This is indicated as “NA” below).
A control theory approach (hereinafter referred to as “CON”) in which the principles of mathematical control theory are applied to a very specific part of the brain-type computer component.
A medical approach (hereinafter referred to as “MED”) associated with practical / clinical application of (mainly) strong electromagnetic fields to humans or animals.

A possible component of NA is a neural oscillator, i.e. an entity that can switch between vibrating and non-vibrating behavior, and may include smaller entities (not further considered). NA is related to neural networks, ie, a collection of small (less than 10 ⁴ ) connected neural oscillators, phenomena that occur in these networks (eg memory or pattern recognition). After the fundamental principles of function have been drawn from the nervous system of living organisms, NA now proceeds toward the implementation of these principles in artificial systems. An example of this development is [2].

  After starting to apply linear control to a sufficiently simple mechanical system, CON goes in the opposite direction. The application of linear control to each non-linear stochastic system (such as the human brain) is not only inappropriate but also potentially dangerous. This is because in linear control, the control force is proportional to the desired change. Therefore, it can be very large.

  Control methods can be classified into model-based methods and database methods. The quality of the model-based method is related to the suitability of the model with respect to the problem under consideration. Traditionally, no brain activity model or behavior model has been applied to humans or animals.

The quality of the database method depends on the simplicity of the system being controlled (because the first step of the database method tends to restore the phase space, a search is performed later and iterations are calculated-it is a higher dimension This is not technically feasible in the case of phase space, except for time-delayed feedback, which involves waiting for time to reach a stabilized target state). Each controlled or regulated electromagnetic correction of current control method of brain activity in vivo is not applicable. Because there are some of the following effects. :
-High and unattainable high demands on storage space and processor speed;
-A potentially infinitely long waiting time to reach the target orbit,
-A system with a stochastic factor starting intermittently from the target trajectory;
-The limitation of the control range of the system, only collective behavior is measured,
-The control range of the system is limited and the control power becomes unclear;
-The unknown amplitude of the control force in the polyphase system.

MED is absolutely based on a microscopic physiological model with individual neurons as a component of brain activity and is referred to below as a single neuron. As a result, in vivo non-invasive, unobservable variables such as ion concentration on both sides of the cell membrane, the number of dendrites, etc. play an important role in calculating the number of firing neurons per unit time (firing rate). Each of the current control methods of brain activity in a single neuron paradigm in vivo (with 10 ¹¹ neurons and 10 ¹⁵ connections between these neurons) does not automatically, individually, controlled or regulated electromagnetic correction Is appropriate. Therefore, this paradigm-based MED approach is hampered by a random, unexplained clinical outcome (see [3] for an overview of transcranial magnetic stimulators). Another MED method affects brain activity, intracranial or extracranial electric shock. All MED approaches target the depolarization of the neural membrane and thereby further target action potential generation (so-called stimulation).

  In CON terminology, the electromagnetic depolarization of a single neuron by the MED method can be described as “slaving” (assuming sensory slaving by rhythmic optical flashes, acoustic signals, etc. is not considered). Slaving is the result of the application of a high-load artificial signal to the system, and during this application incorporates an exogenous signal pattern instead of natural behavior. In NA, it is known that communication under the weak coupling of a neural network depends on matching the frequency at which the neural component vibrates. Thus, local slaving with mismatched or weakly matched frequencies removes such slashed neurons from normal communication, thereby creating a virtual lesion. The current MED-method considers virtual lesions as the main effect of TMS, an unexplained effect in a single neural paradigm.

  If low intensity is available to fire one neuron, or if the slaving area can be narrower and limited, the MED-method is considered to have improved “stimulation”. Repeated slaving at the same medically relevant site can lead to physiological changes. The period is longer than the application period and in some cases is positive (eg, statistically significant improvements in certain types of clinical depression under repeated TMS, see eg [3]).

  When applied to humans, there are only two partial aspects to the current method of individualization: finding the appropriate site of a single neuron that is depolarized first, secondly the slave field Find a reference value for intensity (quoted in percent of “movement threshold”). A more recent MED method [4] uses TMS to obtain information about the presence or absence of physiological connections between different regions of a particular individual's brain, and (under certain restrictions) the transmission rate between different brain regions. Is used for diagnostic purposes.

  Current MED methods are preferably implemented for medical purposes. This is because the disadvantages of slaving (including the absence of a seizure) are at least partially compensated by medical effects. Current MED methods include only passive safety measurements such as avoiding high frequency slaving, user exclusion of seizure propensity, and field switching off when seizures occur.

  Summary: The current techniques and methods do not achieve electromagnetic modification of brain activity that is controlled or regulated, respectively, based on in vivo models, and these techniques and methods do not move towards realization. : MED is incorporated into a single neuron paradigm, NA is moving towards artificial system construction, and CON is experiencing in-vivo-inapplicability to complex polyphase systems.

  The object of the present invention is the creation of methods and devices and the resulting behavior modification for electromagnetic modification of in vivo brain activity, each controlled or regulated.

  This object is solved by a method having the features set forth in claim 1 and a device having the features set forth in claim 44. Preferred embodiments are disclosed in the respective supplementary claims.

  In the method according to the invention for electromagnetic correction of brain activity, a brain activity model (explaining the influence of exogenous electric and / or magnetic fields on brain activity) is used, and a behavior model (between brain activity and behavior) is used. Will be used). As a result, it is possible to influence human behavior by means of exogenous input in a controlled manner for the first time in history.

  In particular, the invention relates to the use of brain activity models, and individual decisions, if necessary, within the individual and / or time-dependent observability, and individual decisions, if necessary, within the individual from extracranial signals to control. And / or associated with a time-dependent transfer operator, resulting in a reliable and controlled intervention as part of an open or closed control loop, and then achieving a brain activity target in a reliable manner. The use of behavioral models ensures the achievement of brain activity targets that correspond to the achievement of the user's personal behavioral targets. The intervention is achieved by a personalized behavioral target, the behavioral model measures the correspondence between behavior and the dynamics of brain activity indicators so that a reliable intervention is feasible, and an appropriate personalized brain activity model Based on the knowledge of quantifying the dynamics of brain activity indicators and showing appropriate control parameters.

  In the following, the intervention is illustrated by example with the aid of figures.

apparatus:
The device 1 according to the invention consists of:
One or two head units 2 connected to the intermediate unit 3 and one or more of them connect to the base unit 4 (FIG. 6).

  The head unit 2 includes a measuring system having a device for measuring electromagnetic data (“sensor”), a device for data preprocessing and data transfer to an intermediate unit, and a digital control instruction for generating an electromagnetic field and by means of a transmitter signal or A regulation system is included having a device (“transmitter”) 5 that performs the regulation instructions (arriving from the intermediate unit). The intermediate unit includes a computer with a connection between the software and the base unit that implements the method for controlling brain activity described below (if calibration is not required). The basic unit 4 includes a computer, firstly a database of model data and user data, and secondly another method which is explained in more detail below. The intermediate unit and the base unit are located on the same or different computers.

  The implementation of the method includes one head unit, one intermediate unit, and one base unit. The head unit includes, for example, an EEG cap with an extracranial sensor, a connection to an amplifier, an amplifier, a connection to an AD converter, an AD converter, a measuring system with a connection to an intermediate unit, and an extracranial induction coil as a transmitter Control systems with controllable power supplies, connections, DA converters, connections to intermediate units for these transmitters. The intermediate unit includes, for example, a personal computer with a connection between the screen and keyboard, software, and the base unit. The basic unit includes, for example, a powerful calculation unit.

  Suitable sensors are for example EEG sensors or MEG sensors. The MEG sensor includes, for example, a suitable control arithmetic unit for detecting a magnetic field and one SQUID sensor element with a cooling device. The EEG sensor has, for example, two electrodes for measuring a potential difference.

  Preferred embodiments of the sensor include a partial or complete electrostatic and / or magnetic shield of the sensor in relation to the surrounding environment, but without interfering with functionality.

  A preferred embodiment of the measurement system within the head unit may include a plurality of sensors located within or outside the cranium. This plurality of sensors is called a sensor grid.

  Preferred implementations of the extracranial sensor grid include fitting such that the sensor resumes its previous relative position relative to each user's skull after intermittently taking the sensor grid several times. This is accomplished, for example, by adapting the sensor grid to the inside of the helmet, but the inside fits the user's skull shape.

  Another preferred embodiment of the sensor grid is a buried electrode.

  A preferred embodiment of the extracranial transmission material 5 includes an induction coil 6 that is a paramagnetic core, a diamagnetic core, or a ferromagnetic core 7 as shown in the cross-sectional view of FIG. Represents the direction of flow. The transmitting substance 5 is basically cylindrical, and the side and top surfaces are covered with a shield 8. The coil 6 and the core 7 are directly adjacent to the side surface of the transmitting material 5 and have no shield. During operation of the device, this surface is oriented towards the skull to transmit the extrinsic magnetic field. A fitting 9 is located on the back of the transmission material 5 so that the transmission material 5 can fit into the helmet.

  The extracranial transmission material 5 can include protection from deformation, for example by embedding the transmission component in a suitable resin or by embedding the transmission component in a suitable insulator.

  The extracranial transmission material 5 may include a cooling device.

  Another preferred embodiment of the intracranial transmitter is an implanted electrode.

  The preferred embodiment of the present invention in connection with the control components of the head unit includes a plurality of transmission materials disposed in and / or outside the skull. This plurality is called the transmitter grid.

  A preferred embodiment of the present invention associated with an extracranial transmitter grid is associated with the user's skull, such that the transmitter restores its previous relative position after intermittently taking the transmitter grid several times. Includes fitting. This is achieved, for example, by adapting the transmitter grid to the inside of the helmet, the inside adapting to the user's skull shape.

  In another embodiment, a camera is used to fix the position of the sensor grid and / or transmitter grid relative to the user's skull, resulting in a spatial position of the user's head, the relative position of the sensor grid and transmitter grid. Captured by the camera and the data is processed into 3D data.

  In another embodiment, the sensor and / or transmitter is an implanted electrode, capable of EEG measurements, and current can be conducted to the brain. Leads and / or interfaces connecting these electrodes and intermediate units, other leads and / or other measuring devices and / or associated computers and / or electrode power supplies and / or computers can be implanted and consequently Full mobile mode operation is possible.

  FIG. 4 shows a schematic diagram of a preferred embodiment of the present invention relating to the extracranial measurement system and the head unit of the control system, together with a cylindrical overhead suspension 11 having a connection cable therein, and a jaw rest 12. A helmet 10 that fits the skull shape of each user is included. The sensor grid and transmitter grid on the inside of the helmet are mounted so that both grips overlap. That is, there are a sufficiently large number of transmitters in the vicinity of each sensor, and vice versa. The planar projection of the mechanical fitting of this embodiment is displayed in FIG. 3 (the drill hole 13 for the sensor is indicated by a circle and the drill hole 14 for the transmitter 5 is indicated by a square). In this embodiment, the user is described as sitting in an armchair with a necklace below the helmet 10.

  In another embodiment of the invention, the sensor grid is in the skull and the helmet has an extracranial transmitter mass.

  In another embodiment of the invention, the transmitter grid is in the skull and the helmet has an extracranial sensor grid.

  In yet another embodiment of the invention, both the sensor grid and the transmitter grid are intracranial.

  In a preferred embodiment of the present invention, the sensor density is adjustable along with the sensor placement of the extracranial sensor grid. In another preferred embodiment of the invention, the adjustment is performed automatically, controlled by an intermediate unit.

  In a preferred embodiment of the present invention, the transmitter density is adjustable with the transmitter arrangement and / or the direction of the transmitter of the extracranial transmitter grid relative to the user's skull. A planar projection of the mechanical fitting of this embodiment of the invention is displayed in FIG. Here, the drill hole 13 for the sensor is indicated by a circle, and the drill hole 14 for the transmitting substance is indicated by a square. Here, it is possible to attach the transmission substance to the drill hole 14 of the fitting and / or to tilt the transmission substance 5 in accordance with the fitting. With other coil arrangements, it is possible to emulate the spatial arrangement and direction of all conventional coil arrangements, especially the magnetic field direction.

  Preparation, testing and maintenance of extracranial devices are performed by engineers who are not required to acquire medical education.

  In a preferred embodiment of the present invention, the device includes conventional protection against power failures and / or voltage flicker.

  In the intermediate unit, in order to optimize the density and location of sensors and transmitters, and to eliminate anthropogenic effects, i.e. to eliminate measurement biases generated by extrinsic magnetic fields, eye movements, muscle spasms, etc. Automatic techniques are feasible.

  Method:

  The method according to the present invention includes the steps of the technique designated “S” and detailed below. Input data and output data are represented as “D”. Each step is described as follows: The basic principle of the described method of the present invention is shown in flowchart 20, where the rectangle is the method step, the rhombus is the decision step, and the parallelogram is the input or output data of the method step. Represent.

S500:
In this step, the variables and dynamics necessary to achieve the behavioral target (D2000) in a reliable manner are clarified with the aid of the behavioral model (D3000).
The behavior model models the correspondence between the behavior and dynamics of a certain BAI. In the following, “BAI” represents one or more “brain activity indicators”. The BAI is calculated from data measured according to a specific rule (D2200).
Examples of BAI are the voltage difference (electrode and reference electrode) measured by the sensor, or the power spectrum of the window signal derived from the data measured by this sensor, or the quotient of β by this sensor divided by αEEG activity (β = 13-30 Hz, α = 8-12 Hz), see [11].
An example of a behavioral model is the reduced Davidson model (see, for example, [11]), so that positive emotions correspond to a higher quotient of β in the left frontal cortex divided by αEEG activity (β = 13-30 Hz). , α = 8-12 Hz).
The behavioral model (D3000) allows the transformation of the behavioral target (D2000) into the target mechanics of the BAI (D2100), including in principle the desired duration of behavior, and thereby allows quantification.
The type and number of BAI minimizes device requirements (D4000). For example, to calculate the higher quotient (β = 13-30 Hz, α = 8-12 Hz) obtained by dividing β in the left frontal cortex by αEEG activity, at least one EEG-electrode and one reference electrode are required. In order to achieve the target BAI in a controlled manner, at least one transmitter is required.

S800:
The BAI used is calculated from the variables of the general brain activity model (D1000) used.
Example: A BAI based on Fourier decomposition of a limited brain region is calculated from the dynamics of a Wilson-Cowan-model neural oscillator (see eg [1]).
The BAI used results in a minimization of calibration requirements (D1100) in relation to the general model used. If the BAI calculation is not robust, the demand for calibration will be higher. Local calibration can be extended with non-local calibration if the BAI used is based on non-solid variables bound at different sites. In either case, the type of calibration (eg “Nan”, “Local”, “Translocal” D1110 in FIG. 21) and its range (eg list of sensors and transmitters) are listed (E100).
Example: For robust, non-linear similarity measures (see eg [5]), calibration is not absolutely necessary. For the β / α quotient mentioned above, a local calibration of the Wilson-Cowan-model is required. For complex cognitive abilities (eg music creation, see [19]), translocal calibration is generally required.

S2000:
Calibration involves the determination of a single, possibly parameter-dependent value (assessed by the population mean of a general model of endogenous input) and a single, possibly time-dependent effect of an extrinsic electromagnetic field on brain activity ("Electromagnetic" is used as "electric and / or magnetic" and is abbreviated as "EM" below). The scope and quality of the calibration affects the reliability of the correction of one or more brain activities.
Calibration leads to a specific brain model (D1200), and exogenous inputs are quantified as control variables in the model (D1300). Calibration is detailed in FIG. In the simplest case, local (see FIGS. 22-23) and translocal (see FIG. 24) calibrations are distinguishable from each other.
The primary goal of local calibration is to determine for each sensor, each specific user, what brain activity occurs within a specific period within the detection range of each sensor (Figure 22 shows the process flow). ).
The second goal of local calibration is for each brain activity within the detection range of each sensor, how this brain activity can be corrected by exogenous EM fields generated by at least one transmitter. (The process flow is shown in FIG. 23).
The goal of translocal calibration (S2400) is to determine the type and extent of impact of other brain component activities on the activity of the brain component being considered.

  The calibration is described below using FIGS. 21-24:

  FIG. 21 shows the part of the method. Based on the data D1100, D1110, and D1120 described above, whether or not to calibrate is determined (the determination step E100 described above).

  If calibration is not required, the method continues with step S3000 described above. If calibration is required, the method continues with step S2100, but includes local calibration (detailed below using FIGS. 22-23).

  Since local calibration calibrates brain activity locally, the same brain activity can be detected and calibrated by several sensors, each sensor unit. Therefore, identifying the same brain activity as identical so that brain activity calibrated several times is considered only once, or multiple calibrations of the same brain activity are evaluated in cooperation (step S2300). Useful. Each data is integrated into data sets D1200 and D1300.

  If translocal calibration is essential at D1110, this is checked at step E120, and if necessary, translocal calibration is performed at step S2400 (detailed below using FIG. 24). Each data is integrated into data sets D1200 and D1300.

  Here, the local calibration will be described with reference to FIGS. 22-23. :

  Here, the brain activity is measured by each sensor (S2110). The actual BAI mechanics is calculated from the resulting time series (D2205) (S2120), resulting in the actual BAI mechanics (D2210).

  Usually, the brain activity model (D1000) is given by a differential equation and includes exogenous inputs under certain assumptions about the influence (D1049).

  Model-mode is the solution of these differential equations with respect to a set of unobservability, ie parameters, extrinsic input and intrinsic input.

  These model-modes are calculated in step S2115 and stored in the database (D1051). This database contains a mapping from a set of unobservability to model-modes. The model-mode BAI mechanics are calculated by the data stored in the database D1051 and / or by using the data directly calculated in step S2115 (S2120). This results in BAI dynamics (D1030) being theoretically possible.

  Since BAI usually simplifies model-modes, it is possible that some model-modes give similar (indistinguishable within error boundaries) BAI-model-modes. It is possible that several sets of unobservability can produce the same model-mode. This relationship between the set of BAI-model-mode, model-mode, and unobservability is known from gauge theory, respectively electrodynamics (corresponding to BAI is eg “energy density in spatial volume”, model-mode The response to is a number of electromagnetic fields, and this energy density is generated, and the response to the unobservable set is a different potential and can be the basis of the electric / magnetic field). For example, in the probability mode, the time series of the Pth moment is stored. Model mode and possibly BAI mode are stored in the database and / or calculated quickly. Only a finite number of unobservability is stored (lattice), as are model-mode and BAI-mode. If necessary, refinement and / or reinforcement of the grid is possible (eg by setting parameters for intrinsic and / or extrinsic inputs and / or adding new coefficients).

  It is possible for the mode to be weakened and / or deformed and / or phase-shifted from the carrier to each sensor (representing the neural population as the “carrier” of the mode, so that activity produces each mode). The image of the mode (which reaches each sensor) will be referred to as the image mode and is similar to the associated BAI-mode.

  A priori assumption (D1040, FIG. 22) whether and how the BAI-model-mode image is computed and whether it can be identified as a BAI-mechanical element of fact and how (S2130) is determined. In the simplest case, the image of the BAI-model-mode is proportional to the BAI-model-mode itself and is set equal to the quotient of the BAI-model-mode and the reciprocal of the damping constant (apart from the case where the amplitude disappears). Can be. This easiest case is assumed as follows. In the simplest case, the (BAI-) model-mode is local, thus accounting for localized brain activity.

  Decomposition by S2130 (see FIG. 22) includes BAI-model-mode detection (D1030) and model-mode identification within empirical BAI-dynamics, generating these BAI-model-modes. There is also the identification of a set of unobservability, which may be the basis for these model-modes. Only a set of unobservable extrinsic inputs equal to zero is allowed here. For the purpose of local calibration, the intrinsic input is always assumed to be the easiest case, the second easiest case is a sine wave, and the third easiest case is vibration.

  For the practice of this method it is useful to limit further analysis of the relevant BAI-model-mode. A BAI-model-mode detected within a virtual BAI-dynamics is called association if it can be detected within the virtual BAI-dynamics on noise, measurement errors, and other similar predefined thresholds .

  The decomposition consists of a list of associated BAI-model-modes (D2300) within each detection range for each sensor, a list of potentially generated model-modes (D2350) for each element of D2300, and a potential for each model-mode of D2350. Leads to a list of parameters and the set of intrinsic inputs (D2400).

As shown in FIG. 23, each sensor S _i, each transmitter T _k, the next step of the method is performed in use. :

  Starting from a set of parameters for endogenous input (D2400), the test (S2170, test signal transmission by transmitter) extracts those unobservability sets (D2410) and Potentially underlying the mode. This includes the determination of how the transmitter signal is converted into an extrinsic input, ie part of the model equation, with the aid of the test signal. When this is achieved, the influence of the extrinsic signal on each model-mode (similar to BAI-model-mode) is quantified and available for actual correction of brain activity (S3000).

  The test signal (D2500) is derived from the intrinsic and extrinsic inputs possible in the simplest case (constant, including an elementary loop of vibration, ie the measured signal, and time delay feedback of more transmitter signals) Arise.

  Test signals are introduced into the general brain model as follows: Within the model equation, “input” is replaced with “endogenous plus extrinsic input”.

  Exogenous input is defined as a result of application of a mobile operator to a transmitter signal.

は、外因性で物理的に測定可能なシグナルを特定の、即ち校正脳モデルの方程式内の制御変数と連結する。各モデルの基礎を成すパラメータセットと内因性入力、そして伝達物質シグナルのパターンと振幅次第で、異なる値の制御変数は各モードを変更し、物理的に測定可能である。 Transfer operator

Connects the extrinsic and physically measurable signal with a control variable in a specific, ie calibration brain model equation. Depending on the parameter set and endogenous inputs underlying each model, and the pattern and amplitude of the transmitter signal, different values of control variables can be measured physically, changing each mode.

  The test signal has the task of distinguishing between different parameters, intrinsic inputs and, if necessary, the set of BAI-mode mobile operators under consideration. The test signal (similar to other transmitter signals) is characterized by an external reference value, and a (usually time-dependent) pattern at the transmitter location, and a series of amplitude multipliers (possibly containing only one element) .

  Each transmitter signal (and thus each test signal) is displayed here as a hybrid symbol, that is, as a series of amplitude multipliers and as a variable in the brain activity model that indicates an external reference value that specifies the generation of the signal. Is done.

For example, an extracranial magnetic field having a magnetic field strength of 0.01 Tesla can be generated by a coil having a current strength I ₀ as follows. For the selected voltage unit, the amplitude of the voltage is normalized to 1 induced naturally by changes in the magnetic field. The pattern of the induced voltage is sin (2 ^* π ^* t ^* 5/1000) (the voltage t as a variable of the brain activity model is millisecond time), and the others are constant current conditions I ₀ , 2 ^* I ₀ , 4 ^* I ₀ , 8 ^* I ₀ is displayed as (0.01T, sin (2 ^* π ^* t ^* 5/1000), (1,2,4,8)).

  Furthermore, if z (t), system behavior is feedback with a time delay of 10 milliseconds, and the coil current is normalized to a voltage amplitude of 1 with the same strength as before: (0.01T, z (t-10), (1, 2, 4, 8)) is displayed.

The duration of the partial signal (I ₀ , 2 ^* I _{0 etc.} ) depends on the desired effect (eg transient vs. limit cycle) as well as the possible pauses of the two partial signals.

  The transfer from the transmitter signal to the signal in each mode of the carrier is affected by invariant and hypothetical effects (D1050).

  The possible amplitude order is affected by technical constraints and / or health constraints (D4100, D4200).

  In the simplest case, for example, there are two different parameters and a set of endogenous inputs, resulting in two different model-modes, but similar BAI-model-modes. That is, it is not possible to determine which set of unobservability is the basis for each brain activity by BAI observation. Thus, under speculative assumptions about the mobile operator (eg, a combination of phase deviation and signal decay), both sets of unobservability become general brain models so that a model-mode is created by the test signal. By inserting, a series of amplitude multipliers are numerically calculated in the test signal pattern (S2160) and the resulting BAI-model-mode is different for both sets.

  This numerical calculation usually leads to a set of possible patterns of test signals, each with a set of possible amplitude multipliers. An optimal test signal is selected according to predefined criteria (eg, the minimum vibration of all brain activity that occurs except the brain activity being tested).

Starting with the transmission of a signal with the first amplitude in a series (S2170), the minimum is the sensor under consideration (S _i ), the maximum is the measurement of electromagnetic brain activity (S2110) on all sensors in the sensor grid (see FIG. 5) ) And results in a time series (D2520) in which BAI dynamics are calculated (S2120).

  Subsequently, the de facto BAI mechanics is decomposed into BAI-model-mode by the extrinsic input (S2135). Only the set of parameters and extrinsic inputs from D2400 (see Fig. 22) are allowed for this decomposition, and as a result the setting "exogenous input equal to zero" is probably from different extrinsic inputs, invariant and hypothetical effects. The resulting pattern is exchanged by the intensity resulting from (usually non-linear) changes associated with different amplitudes in a series of amplitudes.

  This decomposition results in an approximately best set of some unobservability (E210), and another test signal is calculated, D2410 (see figure) instead of the previous parameter and endogenous input set from D2400. 23), the loop following S2160 is executed again at S2135. If multiple iterations of this loop do not produce a result, the mediator direction needs to be changed and another mediator is selected.

In summary, after the local calibration is complete, you will know:
-Which model-mode is within the detection range of the sensor,
-Which parameter set and which endogenous input generate each model-mode,
-Methods that affect each model-mode.
In addition to local calibration, the same carrier in model-mode can be identified in step S2300.

によって特徴付けられるセンサーＳ_k,に属するモードである場合、またＡ_mn（ｔ）がパラメータＰ_mn,内因性入力Ｉ_mn（ｔ）、移動作用素

（χ伝達物質の列挙）によって特徴付けられるセンサーＳ_m,に属するモードである場合、次の記述が有効である：もし各Ｐ_klがそれぞれＰ_mn,に等しい、また各Ｉ_kl（ｔ）がＩ_mn（ｔ）に等しい場合、両モードは同一である。

は

に等しくなく、および／または

が

に等しくないように伝達物質Ｔ_χ,があるなら、これらの同一のモードは同じキャリアを有しない。 A _kl is the parameter P _kl, endogenous input I _kl (t), transfer operator

When a sensor S _k, belonging mode characterized by also A _mn (t) parameter P _mn, endogenous input I _mn (t), the mobile operator

If the mode belongs to the sensor S _m , characterized by (enumeration of χ transmitters), the following description is valid: If each P _kl is equal to P _mn , and each I _kl (t) is When equal to I _mn (t), both modes are the same.

Is

And / or

But

These identical modes do not have the same carrier if there is a transmitter T _χ , not equal to.

  In the subsequent translocal calibration, method step S2300 is not essential, but leads to possible minimization of the function matrix (see D2410 FIG. 24).

  Translocal calibration (S2400) determines the type and range of sensory input effects on other modes and / or modes (indirect extrinsic effects, ie visual, auditory, tactile, other sensory stimuli The display is mediated by the sensory system, resulting in delay and modeling as part of the endogenous input). A tool for this purpose is a function matrix (D2410): if a total of n different model modes are found within the detection range of at least one sensor, this function matrix is an n × n matrix and mode i If the influence of mode j on mode j can be determined, the (i, j) cell indicates the component. The component quantifies this effect in the equation of the brain model of mode j so that “endogenous input” is replaced by “possibly another intrinsic input in addition to the time delay function of the i th mode”.

  A method for filling the required portion of the function matrix is shown in FIG. By learning the function matrix or parts of it, the events of each brain model, including several components of the model (synchronization, phase locking, and more) by traditional mathematical methods, neural networks, nonlinear dynamics, etc. Enable calculation.

  The transfer operator obtained from local calibration (D1300) and further unavailability allow BAI prediction (S2410) in all modes, taking into account translocal effects only in a collective manner. When changing the mode to active (S2420), the possible difference between the actual BAI and the predicted BAI may become apparent. This difference is captured in step S2430. These differences are basically based on translocal effects. In step S2440, such a function matrix is calculated and can account for the above differences (when included in the brain model equations).

  In E400, it was tested whether the mapping from the above differences to the function matrix was unique. If not, in step S2221, the test signal is modified to limit the set of potentially appropriate function matrices. This iteration stops after uniquely determining the desired elements of the function matrix.

  Following calibration, preferably EM control or regulation of brain activity (S3000) is a target for achievement and support of BAI target dynamics by a brain activity model-based feedback loop. This is detailed in FIG.

  Here, the transmitter signal is first calculated, but it is theoretically appropriate to convert the actual BAI dynamics (D1010) to the target BAI dynamics (S3100). In a preferred embodiment, these transmitter signals are composed of elementary signals, and impacts can be calculated within a specific brain model, each of which can be learned from calibration. Because of this combination, various effects (effect of one transmitter signal on several brain activities), spatial composition (signal of several transmitters), and composition in time (series of signals) Is appropriate. After the calculated signal (for health constraints (D4200), device constraints (D4100), possible ripple limits (D4300)) feasibility test, the utility function selects the best transmitter signal, It is transmitted (S3400). The theoretical impact of signals within a specific brain model is known and allows prediction of BAI dynamics (D4040). If (in continuous EM measurement and continuous de facto-BAI calculation) the de facto and target-BAI-dynamics comparison (S3500) results in a significant difference (E300), the calibration is performed in a further calibration step (S2000) To improve, S3000 is interrupted as specified. In this case, the method then proceeds to step S3100 to recalculate the transmitter signal.

  With some examples, the main steps of the method are explained. An example will be described step by step using a flowchart (FIGS. 20-25). :

  Example 1:

  FIG.

D2000 (behavioral target):
Active emotional amplification during application time

D3000 (behavior model):
The subsequent correspondence is called the “reduced Davidson model” (see, for example, [11]). A positive emotion corresponds to a higher quotient of β to α activity in the left frontal cortex (β = 13-30 Hz, α = 8-12 Hz).

（前提：整数周波数、フーリエ・ウィンドウ、例えば５００ミリ秒、i−ヘルツ−モードのフーリエ係数のｆ_iモジュラスだけ。１から５０Ｈｚの周波数を扱うためにはNyquistによると、少なくとも１００データポイントが必要である）。ＢＡＩ（１００）はデータポイント１−１００のパワー・スペクトルであり、ＢＡＩ（１０１）はデータポイント２−１０１のパワー・スペクトルである、等。Ｐｏｓ（１００）はＢＡＩ（１００）から導出され、Ｐｏｓ（１０１）はＢＡＩ（１０１）からである、等。この導出されたＢＡＩは増加する、例えば時間ｔから（あるいは各データポイントから）、Ｐｏｓ（t,影響）＞２^*Ｐｏｓ（ｔ,影響なし）。この要求はすべての標的ＢＡＩ力学を決定する。 S500 (BAI specification and calculation, target BAI-dynamics specification):
The Davidson model is based on the power spectrum of the EEG-signal. The appropriate BAI is therefore a series of square coefficients of the Fourier coefficients and is calculated by a fast Fourier transform. An example is a frequency between 1 and 50 Hz. Derived BAI is

(Assumption: integer frequency, Fourier window, e.g. 500 milliseconds, only the f _i modulus of the Fourier coefficient of the _i -hertz mode. To handle frequencies from 1 to 50 Hz, Nyquist requires at least 100 data points. is there). BAI (100) is the power spectrum of data point 1-100, BAI (101) is the power spectrum of data point 2-101, and so on. Pos (100) is derived from BAI (100), Pos (101) is from BAI (101), etc. This derived BAI increases, eg, from time t (or from each data point), Pos (t, influence)> 2 ^* Pos (t, no influence). This requirement determines all target BAI dynamics.

D4000 (minimization device requirement):
Surface-One electrode on the left frontal for EEG, eg one reference electrode on the ear (see [16]), one extracranial coil on the left frontal (as transmitter). 100 data points every 0.5 seconds (see S500), the minimum sampling rate is 200 / second.

D2100 (Target BAI-dynamics):
Pos (t, affected)> 2 ^* Pos (t, unaffected)

D2200 (BAI calculation rule):
Using the Fast Fourier Transform, the square coefficient (f _i ) ² of the Fourier coefficient of each i-th mode (i is between 1 and 50 Hz) is calculated.

はシグモイド関数である。 D1000 (general brain activity model):
Simplified Wilson-Cowan-model (from [1], refractory time or transmission delay set to zero, etc.) and brain activity is based on neural oscillator activity. Neural oscillators oscillate or do not oscillate depending on the input and also depending on unobservable physiological parameters. A neural oscillator is a group of interconnected excitatory or inhibitory single neurons (in [1], a group of 10 ⁵ single neurons. As long as each neuron group follows the following equation, (Multiples of this number are also allowed), modeled by a system of two nonlinear differential equations. :

Is a sigmoid function.

また,

ではρ_x,exo＝

（伝達物質シグナル）はｉ−番目の伝達物質から考慮中の神経振動子へ伝達される外因性入力を示す。そして「ｅｎｄｏ」は「内因性」を意味する。

は移動作用素であり、因子の中でも特に各伝達物質から考慮する神経振動子の距離、伝達物質の軸に関連して個々のニューロンの方向、他の生理学的パラメータ、さらに伝達されるシグナルのタイプに依存する。移動作用素は観察不可能性である。上記の観察不可能性は、異なる人々の最大１００までの要因によって異なり、in vivoの１つの脳の異なる脳領域内でさえも相当異なり、変化し得る（ＥＭ振動子と比較するとゆっくり）ことが知られている。得られた複数の理論的および事実上のモードは、ほとんど任意の数のモードを含む。したがってローカル校正無しの制御されたローカル修正は、通常は不可能である。比較的一定の内因性入力という前提下では、トランスローカル校正は必要ではない。 “X” and “y” are electromagnetic variables and can be observed as αx + βy. As a result, both α and β are derived from the interval [−1, 1] and are called mixing angles. “X” indicates a set of individual excitatory neurons and “y” indicates a set of individual inhibitory neurons (“set” = such neurons in a neural oscillator).
“A”, “b”, “c”, “d” are slow unobservability.
“A” models the influence of a set of individual excitatory neurons on itself,
“D” models the effect of an individual set of inhibitory neurons on itself,
“B” models the effect of an individual set of inhibitory neurons on an individual set of excitatory neurons,
“C” models the effect of a set of individual excitatory neurons on a set of individual inhibitory neurons,
“Ρ _x ” models the endogenous input from outside the neural oscillator to a set of individual excitatory neurons,
“Ρ _y ” models the intrinsic input from outside the neural oscillator to a set of individual inhibitory neurons,
Within the model, input variables are not necessarily considered slow (this aspect is specifically considered when describing translocal calibration). All times are displayed in milliseconds, and all other units are dimensionless. The “t” on the left side of the Wilson-Cowan equation models the membrane time constant, another unobservability, in milliseconds.
A simplified Wilson-Cowan-model is augmented to include extrinsic effects: direct electromagnetic extrinsic effects are modeled as part of the input:

Also,

Then ρ _{x, exo} =

(Transmitter signal) indicates the exogenous input transmitted from the i-th transmitter to the neural oscillator under consideration. “Endo” means “endogenous”.

Is a mobile operator, among other factors, the distance of the neural oscillators to consider from each transmitter, the orientation of individual neurons relative to the transmitter axis, other physiological parameters, and the type of signal transmitted Dependent. Mobile operators are unobservable. The above unobservability depends on up to 100 factors of different people, can vary considerably even within different brain regions of one brain in vivo and can vary (slowly compared to EM transducers). Are known. The resulting plurality of theoretical and de facto modes includes almost any number of modes. Thus, controlled local correction without local calibration is usually not possible. Translocal calibration is not necessary under the assumption of a relatively constant endogenous input.

S800 (minimized calibration requirement specification):
[tau], a, b, c, d, [rho] _x , [rho] _y are transfer sensors of one sensor-transmitter-pair. Only local calibration is required.

D1100 (minimized calibration request):
Calibrate τ, a, b, c, d, ρ _x , ρ _y , transfer operator locally for one sensor-transmitter-pair.

E100 (Do you need calibration?):
Local calibration is necessary.

D4100 (device restriction):
Depending on the device used, for example, by restricting to a conventional adhesive electrode digitized at a sampling rate of 200 / sec on the input side, for example a single transmitter with a maximum magnetic field of 0.5 Tesla on the output side. It becomes impossible to influence several modes independently.

D4200 (health constraints on exogenous electromagnetic fields):
For example, the recommendation [14] is included.

S2000 (calibration):
This will be described in detail when the flowchart of FIGS.

D1200 (specific brain activity model):
D1000 and D1300 including empirically determined τ, a, b, c, d, ρ _{x, endo} , ρ _{y, endo} (for each relevant brain activity of each user within the consideration period).

D1300 (quantification of the influence of exogenous electromagnetic fields on brain activity):
In the simplified Wilson-Cowan equation that models this brain activity, the extrinsic electromagnetic field to each relevant brain activity of each user within the period under consideration, represented by the control variables ρ _{x, exo} and ρ _{y, exo} Impact determined empirically.

S3000 (control or regulation of brain activity by BAI control variables and target correction by measurement / calculation):
This will be described in detail when each flowchart is handled (FIG. 25).

  FIG.

D1100 (minimized calibration request):
Same as above.

D1110 (calibration type):
local.

D1120 (calibration range):
Sensor S1, transmitter T1.

E100 (calibration?):
Yes.

S2100 (local calibration):
This will be described in detail when handling the flowcharts of FIGS. In this example, there is only one sensor-transmitter-pair that needs to be calibrated.

S2300 (identification of the same brain activity):
Omitted because multiple sensors and transmitters are not available.

E120 (Trans-local calibration?):
No.

D1200 (specific brain activity model):
Same as above.

D1300 (quantification of the influence of exogenous electromagnetic fields on brain activity):
Same as above.

S3000:
See FIG.

  FIG.

S2110 (EM brain activity measurement by S1):
The potential difference is measured by the left frontal electrode and the reference electrode and digitized at a value of 100 per second. The generated time series (D2205) is stored.

D2205 (empirical time series of S1):
Time series of potential difference from S2110.

S2120 (calculation of de facto BAI dynamics from the time series of S1):
A frequency between 1 and 50 Hertz was obtained at a sampling rate of 100 / sec. A power spectrum is calculated from a data window of a predetermined length (100 data points) by FFT (Fast Fourier Transform). From there Pos is calculated based on the formula of S500 (S2130 already starts with BAI (100)).

D2210 (virtual BAI mechanics of S1):
The de facto dynamics of the power spectrum are BAI (100), BAI (101), BAI (102), etc. The actual dynamics of Pos are Pos (100), Pos (101), Pos (102), and the like.

D1000 (general brain model):
Same as above.

D1049 (impact assumption):
D1050 will be described in detail with reference to FIG.

D1051 (database):
The infinite multiplicity of possible parameter combinations in the general model decreases finitely a priori. First, by computing the limits of the parameter range where there is an intrinsic input leading to the unfixed EM output, and second, by digitizing a superset n times larger than this range (definition of parameter lattice, n natural number). Each parameter set is included in this grid. In the simplest case, each endogenous input is considered constant. In the most simplified Wilson-Cowan-equation, random variables (uniformly distributed in the interval [0,1]) are selected as initial values for the activity variables x and y.
Here, it demonstrates with an example. Initially the extrinsic input is set to zero and the resulting power spectrum results of the different parameter values have a constant intrinsic input (x-axis is Hertz frequency, y-axis is each Fourier Indicates the square factor of the coefficient).

FIG. 7 shows the BAI mode under the variation of the film time constant t.
(I) Generated with τ = 5, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Ii) τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iii) τ = 15, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iv) τ = 20, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
Under constant input, these modes are stationary, so the BAI displayed over time is also BAI dynamics. Transients between the initial value and the subsequent vibration are always excluded, although only for a few milliseconds.

FIG. 8 shows the BAI of the mode under the variation of the self-excitability parameter a:
(I) It is generated with τ = 10, a = 14, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Ii) Generated with τ = 10, a = 18, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iii) τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iv) Generated with τ = 10, a = 26, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
For the two small self-excitability parameters, the lack of vibration is noteworthy.

FIG. 9 shows the BAI of the mode under the variation of the cross inhibition parameter b:
(I) τ = 10, a = 22, b = 15, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Ii) τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iii) τ = 10, a = 22, b = 25, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iv) Generated with τ = 10, a = 22, b = 30, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.

FIG. 10 shows the BAI of the mode under the variation of the cross excitability parameter c:
(I) Generated with τ = 10, a = 22, b = 20, c = 10, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Ii) τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iii) τ = 10, a = 22, b = 20, c = 18, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iv) Generated with τ = 10, a = 22, b = 20, c = 22, d = 3, ρ _x = 1.5 and ρ _y = 0.
The lack of vibration in the case of strong cross-excitability of inhibitory neurons in the assembly is clearly conspicuous (case iv).

FIG. 11 shows the BAI of the mode under the variation of the cross inhibition parameter d:
(I) Generated with τ = 10, a = 22, b = 20, c = 14, d = 0, ρ _x = 1.5 and ρ _y = 0.
(Ii) τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5 and ρ _y = 0.
(Iii) τ = 10, a = 22, b = 20, c = 14, d = 6, ρ _x = 1.5 and ρ _y = 0.
(Iv) τ = 10, a = 22, b = 20, c = 14, d = 9, ρ _x = 1.5 and ρ _y = 0.
The lack of vibration in the case of strong self-suppression is clearly striking (case iv).

S2115 (for each set of unobservability: associated model-mode calculation and / or retrieval of mode from database):
Inserting unobservability into the simplified Wilson-Cowan equation and solving it numerically.

S2120 (calculation of BAI mechanics from model-mode and / or retrieval from database):
Application of BAI-calculation to the result of S2115

D1030 (BAI model-mode):
S2120 result partially described under D1051.

D1040 (BAI measurement-unchanged and-premise):
For the purpose of model-mode detection in de facto mode, in the simplest case, the BAI is selected so that it is partially or completely invariant with respect to the measurement method, or such invariance can be assumed Is done. In a given example (power spectrum) this premise is: The attenuation of the power spectrum between the model-mode carrier and the sensor is frequency independent. This means that the model-mode coefficient equation is unchanged on the way to the sensor. Further assume that the signal is piecewise stationary (on average, the period of rest is longer than the size of the Fourier window). And (S. [1]), due to the difference in the potentials emanating from the excitatory and inhibitory neurons related to each brain activity, the EEG signal is the same amount as xy, that is, the mixing angle is α = 1, β = −1.

S2130 (S1: Decomposition of model-based BAI mode from BAI mechanics to BAI model-mode by related tests):
In principle, model-based BAI mode is per default EM variable and is described in the brain model (eg, potential difference). In the given example (quasi-stationary neural oscillator), the power spectrum is used instead. A priori modification of default settings according to the brain model used is left to the person skilled in the art: confirm the model-mode, its BAI-model-mode (in the simplest case, divided by the decay constant) (reference, The power spectrum is best described (for example with respect to L1). Remove this mode divided by the attenuation constant from the power spectrum. Identifying the model-mode, its BAI-model-mode best describes the residual power spectrum (relative to the reference). For example, the predefined threshold is until passed from above.

For example, if the de facto power spectrum module noise has a steady state maximum value of 10 Hz (coefficient 10), 20 Hz (coefficient 5), and a smaller maximum value in multiples of 10 Hz, Both BAI model-modes shown in FIG. 12 explain the virtual power spectrum equally well (with respect to the maximum of the vertices, not the lower region). The first (E _1,1 ) has an attenuation of about 10 / 2.2 and the second (E _1,2 ) has an attenuation of about 10 / 6.8. Removal, eg, the first BAI-model-mode multiplied by 0.22, results in a zero power spectrum. Analogously, the second BAI-model mode is multiplied by 0.68. This means that in the given example, only model-modes E _1,1 and E _1,2 are both confirmed for sensor S ₁ . In the case of an endogenous input that does not change, it is impossible to distinguish between these modes. Therefore, the distinction is performed by the test signal.

D2300 (list of relevant BAI model-modes within detection range of each sensor Si):
Here is a list consisting only of E _1,1 and E _1,2 for sensor S ₁ .

D2350 (list of possible model-modes for each element of D2300):
This list includes model-modes M _1,1,1 (for E _1,1 ) and M _1,2,1 (for E _1,2 ), the generation of which will be described next.

D2400 (list of possible parameter sets for endogenous input for each model-mode of D2350):
This list includes, among other parameter sets for endogenous inputs, each model-mode of the parameter set for intrinsic inputs of D2350, which is generated based on the simplified Wilson-Cowan equation. For example, for D 2400, the set of parameters for M _1,1,1 is τ = 10, a = 22, b = 20, c = 14, d = 3, intrinsic inputs are ρ _x = 1.5, ρ _y = 0, and for M _1,2,1 , the parameter set is τ = 10, a = 26, b = 20, c = 18, d = 3, the intrinsic input is ρ _x = 1.36, ρ _y = −0 Including .14.

  FIG.

D2300, D2350, D2400, D1000, D1051, D4100, D4200:
Same as above.

移動作用素の形状については、試験可能な前提が作られる。例えば音響刺激の特別な場合は、

（［１８］を参照）。また同じ前提が作られる。さらにこの前提は、概して有効性の範囲内で、振幅だけが異なる２つの伝達物質シグナルは（例えば、振幅関係１：ｒ，ｒ実数＞０）、その振幅も１：ｒの関係を有する外因性入力にトランスレーションする。さらに、ρ_x,endo，ρ_x,endoはローカル校正中静止している。 D1050 (invariant and assumed effects):
Control variables in the general brain model equations are created by applying a priori unknown transfer operators to the transmitted signal (to each model-mode used and to each transmitter T1). In a given example:

A testable assumption is made about the shape of the mobile operator. For example, in the special case of acoustic stimulation,

(See [18]). The same assumption is made. Furthermore, this premise is that, within the scope of effectiveness, two transmitter signals that differ only in amplitude (eg, amplitude relationship 1: r, r real> 0), the amplitude is also extrinsic with a relationship of 1: r. Translate to input. Further, ρ _{x, endo} and ρ _{x, endo} are stationary during local calibration.

S2160, D2500:
In principle, all signals are suitable as test signals and are based on device constraints (D4100) and health constraints (D4200).
The purpose of the test signal is to eliminate candidates that generate parameters and endogenous inputs from D4200 until the mapping from the first observed de facto BAI mechanics is clear. Second, to determine the translation from the test signal to the exogenous input.
From the set of all possible test signals, the test signals that meet these requirements can be determined numerically (eg, by order-based selection of aggressive calculations in addition to functions).
It is a reasonable simplification to determine from the test signal the relationship with the measured brain activity and with the used brain activity model and the signal (in a consistent manner) that can be used for correction (S3000). . For linear oscillators, frequency and matched and mismatched frequencies are distinguished. In certain instances, the fundamental frequency (which is the lowest frequency in the power spectrum and is not generated by noise) is used analogously. : According to FIG. 12, the fundamental frequency of two similar BAI model-modes is 10 Hz. Thus, the mismatch test signal can be a frequency of 3 Hz, 7 Hz, 9 Hz, 11 Hz or other sinusoidal signals. The matching test signal can be a frequency 2 Hz, 4 Hz, 5 Hz, 6 Hz, 8 Hz, 10 Hz or other sinusoidal signal (1 mode, 1 carrier, phase deviation is not relevant for a given example). Furthermore, the elementary loop is in principle used as a test signal, ie a measured signal, for example a time delay feedback of the shape.
exo (t) = x (t-10) -y (t-10)
In addition to the possibilities shown for signal patterns, different amplitude multipliers are available starting from small amplitudes. Example: Test signal l = (0.05 Tesla, sin (2 ^* Pi ^* t ^* 3/1000), (1, 2, 3, 4)), time in t milliseconds.

S2170 (send test signal):
To avoid the effect of time weighting, the sequence of signals determined in D2500 was performed with a pause between two transmissions in the order of several membrane time constants.

S2110, D2510:
Same as above. The measurement and transmission of the test signal are performed simultaneously, and are eliminated from the signal by calculation if, for example, the missing and imperfection of the sensor is artificially generated. Normally, the calculation of the artificial strength created by the transmitter in the sensor takes into account the distance between the two and the composition of the substance between them. Before starting calibration, these artificial measurements can be made directly by transmitting ultra-low-amplitude signals that are sufficiently distinguishable from natural activities (low levels that can be assumed to have little potential to affect neural activity). amplitude). Artifacts are possible without transmitter signals (eg, due to muscle activity being measured) and are removed by conventional means.

S2120:
Same as above.

S2135 (test signal and associated BAI model-mode decomposition):
In contrast to S2130, here no decomposition occurs for any BAI model-mode, but decomposition occurs in relation to the BAI model-mode of the test signal.
Here a D2400 parameter set and intrinsic input is possible, and extrinsic input can now lead to other solutions (test signal model-mode) and therefore also to other test signal BAI model-modes.
The attenuation of S2130 is maintained on the assumption that the position of the carrier in each mode does not change under extrinsic input.
The BAI model-mode to be considered is shaped in FIG. 13 with multipliers 2, 4, 6 and 8, for example under mode I frequency 7 Hz sinusoidal extrinsic sign input (see FIG. 12 (i)) . :
(I) ρ _exo = 2 ^* sin (2pt ^* 7/1000)
(Ii) ρ _exo = 4 ^* sin (2pt ^* 7/1000)
(Iii) ρ _exo = 6 ^* sin (2pt ^* 7/1000)
(Iv) ρ _exo = 8 ^* sin (2pt ^* 7/1000)
There are already some observations on the quantification of signal effects, the results. : The power spectrum in which the 7 Hz and 14 Hz components are reversed is normalized when the multiplier is doubled from 2 to 4. When the multiplier is increased from 2 to 4, the square coefficient of the 7 Hz component indicates the following approximate ratio: : 1/4, 4 to 6: 3/2, 6 to 8: 4/3.

With the same extrinsic input, in other BAI model-modes (mode II, see FIG. 12 (ii)), the power spectrum shown by the multipliers 2, 4, 6, and 8 in FIG. 14 is generated. :
(I) ρ _exo = 2 ^* sin (2pt ^* 7/1000)
(Ii) ρ _exo = 4 ^* sin (2pt ^* 7/1000)
(Iii) ρ _exo = 6 ^* sin (2pt ^* 7/1000)
(Iv) ρ _exo = 8 ^* sin (2pt ^* 7/1000)
Even at small multipliers, the test signal distinguishes between mode I and mode II so that the comparison of FIG. 13 (i) and FIG. 14 (i) becomes clear.

The effect can be calibrated by multiplying the signal strength I ₁ : for example, in mode II, the ratio of the first vertex divided by the second vertex by doubling from signal strength I ₁ to I ₂ Decreases from 6 to 2.5. By comparing this result with the BAI model-mode database, it is shown that I ₁ corresponds to a multiplier of 2 (FIG. 14 (i)) and I ₂ corresponds to a multiplier of 4 (FIG. 14 (ii)).

  Previous examples that were ignored (for reasons of simplicity of display)-due to frequent complexity in practice-may not be able to perform measurements and transmissions simultaneously. In the case of a short measurement interruption compared to the oscillation period, the interrupted measurement signal can usually be reconstructed using conventional signal theory methods. Next, a predetermined uninterrupted measurement signal is considered. Transmission interruption is indicated in the case of an extrinsic input elementary loop.

および

、そして
ρ_x,exo＝ρ_y,exo＝μ^*(ｘ(ｔ−１０)−ｙ(ｔ−１０))、移動作用素がシグナルをひずませないおよび／またはさらに遅延させないと仮定するなら、mは伝達物質シグナルの連続的な振幅倍加によって決定される乗数である。他のすべての記号についてすでに説明した。 As an example of an elementary loop, the measured signal x (t) -y (t) has an equation with a time delay t, ie t = 10 milliseconds, and feeds back to the system. :

and

And ρ _{x, exo} = ρ _{y, exo} = μ ^* (x (t−10) −y (t−10)), assuming that the mobile operator does not distort and / or further delay the signal: m is a multiplier determined by continuous amplitude doubling of the transmitter signal. All other symbols have already been explained.

If the sensor grid and the transmitter are not operating at the same time (eg, associated with a sensor being blocked during transmission), the switch time is selected as the delay time. In the previous example, switching between measurement and transmission every 10 milliseconds. In the above equation,
ρ _{x, exo} = ρ _{y, exo} = μ ^* (xi (t))
Then, xi (t): = if (t mod 20> 10, x (t−10) −y (t−10), 0) in the generic code.

FIG. 15 shows mode I, with μ = 4 (i) and μ = 8 (ii) for this extrinsic input.
FIG. 16 shows mode II, with μ = 4 (i) and μ = 8 (ii) for this extrinsic input.
FIG. 17 shows mode I, with μ = 12 (iii) and μ = 16 (iv) for this exogenous input.
FIG. 18 shows mode II, with μ = 12 (iii) and μ = 16 (iv) for this exogenous input.

  Here, the test signal also distinguishes between mode I and mode II, for example by frequency shift (FIG. 15 (ii) vs. FIG. 16 (ii)) and / or the ratio of the maximum value of continuous vertices (FIG. 17 (iii)). FIG. 18 (iii)). The quotient of the maximum value of the continuous vertices of the mode for different intrinsic inputs makes it possible to show analogously here for a sinusoidal input for the calibration of the influence of the extrinsic input.

  In this simple example, there is no phase consideration. Another arrangement of test signals that are phase-shifted is used for fine-tuning calibration.

For further display of Example 1, it is assumed that time delay feedback, test signals, determined that the observed brain activity is generated by Mode I (unobservability τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _{x, endo} = 1.5 and ρ _{y, endo} = 0), the time delay feedback signal of the transmitter T1 enters at 40 ^* [Magnetic field strength Tesla] ^* xi (t). Further assume that the mode from carrier to sensor S ₁ is attenuated by 10 / 2.2.

  Note 1: In general, it is not always possible to assume that the appropriate mode is generated by the parameters and inputs of the selected grid. That is, local refinement and / or interpolation is required.

  Note 2: One or more jumps in the quasi-stationary endogenous input occurred during transmission, suggesting that the results are inconsistent within the signal sequence. In this case, the mode of time-dependent input needs to be considered for calibration.

E210 (set of some closest parameters):
Not relevant in this example. Uniqueness is a defined modulo identity. If each A mode is indistinguishable from each B mode (with respect to a defined tolerance) under all inputs, the two sets of parameters, A and B, are called identical.

FIG.
In this example, only local calibration is required, so it is omitted.

  FIG.

D2100, D1200, D1300, D1050, D1051:
Same as above.

Ｓ５００に基づく。 D1010 (virtual BAI-dynamics):
Using data from continuous measurements of brain activity, Pos is continuously calculated (continuous = within an appropriate time period).

Based on S500.

S2210, D2120:
Same as above.

S4100, D4200:
Same as above.

D4300:
Not possible with just one transmitter. This will be explained in Example 3.

S3100 (calculation of one or more transmitter signals):
The normally infinite number of possible transmitter signals decreased similar to the calculation of the test signal. In the simplest case, a transmitter signal is used, but it is already used as a test signal. The transmitter signal is tested for feasibility (D4200, D4100, D4300) and fits for a defined utility function (eg, as low frequency as possible, as low as possible, magnetic field strength, etc.). The best possible transmitter signal is automatically selected. Example: The action target is achieved by the feedback signal transmitted at T1 of 0.4 Tesla mentioned above, but Pos is more than doubled (compare Figure 12 (i) and Figure 17 (iv)) ).

S3400 (transmission):
S3100 transduction of signals that are selected (s) based on (by transmitter T ₁ in this case).

D4040 (BAI-dynamic prediction):
The exogenous input model-mode determines relevant brain activity over time and generates a time series. From the future part of this time series, the theoretical BAI mechanics are calculated as before and serve as the predicted BAI mechanics. In a given example, only one model mode is relevant. Therefore, the prediction of BAI dynamics is calculated by dividing the BAI shown in FIG. 13 (i) by the damping constant 10 / 2.2.

S3500 (comparison):
On the other hand, the actual BAI mechanics of S2120 is subtracted from the predicted BAI mechanics of D4040. The absolute number of this difference is calculated at each frequency between 1 Hz and 50 Hz, with the maximum value being the result of the comparison. On the other hand, whether a behavioral target (target BAI mechanics) has been achieved is calculated analogously by comparing the actual BAI mechanics with the target BAI mechanics.

E300 (acceptable difference?):
If the assumption and the calibration result are correct, it is expected that the predicted BAI dynamics and the actual BAI dynamics are slightly different (below a predetermined threshold), and in this case, transmission can be continued (S3400). If there is a significant difference, the brain model equations need to be solved numerically with other endogenous inputs to test whether the jump of piecewise continuous fixed intrinsic inputs explains the difference (S3600).
If the behavioral target is not achieved within the calculated time limit, the transition to S3800 is made or a “behavioral target achieved” message is signaled.

S2000 (calibration):
As explained before.

  Concluding remarks on this example: Brain activity targets are achieved in a reliable manner with calibration and subsequent controlled modifications. Therefore, in the case of the reduced Davidson model application, behavioral targets are also achieved.

  Example 2 (explanation of steps is only an excerpt):

  FIG.

D2000 (behavioral target):
Amplification of positive emotions during application time and reduction of passive emotions

D3000 (behavior model):
Davidson model (see eg [11]): Positive emotions correspond to a higher quotient of β for α activity in the left frontal cortex (β = 13-30 Hz, α = 8-12 Hz), passive Emotion corresponds to a higher quotient of β for α activity in the right frontal cortex.

、右前頭センサーＳ₂に関して

である。（ｆｉｊセンサーｊのｉ−Hertzモードの係数）。Ｐｏｓは増加し、例えばＰｏｓ（t，影響あり）＞２^*Ｐｏｓ（ｔ，影響なし）。Ｎｅｇは減少し、例えばＮｅｇ（ｔ，影響あり）＞０.５^*Ｎｅｇ（ｔ，影響なし）。これらの要求は標的ＢＡＩ力学を決定する。実例１であるような計算。 S500 (identification of BAI, identification of target BAI dynamics):
Similar to Example 1. Derived BAI is for one left frontal sensor S ₁

Regarding the right frontal sensor S ₂

It is. (The coefficient of the i-Hertz mode of the fij sensor j). Pos increases, for example Pos (t, affected)> 2 ^* Pos (t, unaffected). Neg decreases, for example Neg (t, affected)> 0.5 ^* Neg (t, unaffected). These requirements determine the target BAI dynamics. Calculation as in Example 1.

D4000 (minimization device requirement):
In the surface -EEG, a first electrode S ₁ to the left frontal, (see [16]) one reference electrode in the left ear, first electrode S ₂ to the right frontal, one reference electrode in the right ear, two extracranial coils There are (transmitter T ₁ and T2), for example, T ₁ is directly adjacent to S _1, the connection is S ₁ and the central electrode Cz virtual line of the head surface (see [16]). T ₂ is analogously between S ₂ and Cz, and the distance from S ₁ to T ₁ is equal to the distance from S ₂ to T ₂ . Everything else is as in Example 1.

D2100 (Target BAI-dynamics):
Pos (t, affected)> 2 ^* Pos (t, not affected) and Neg (t, affected) <0.5 ^* Neg (t, not affected).

である。 D2200 (BAI calculation rule):
As in Example 1, in addition to the second right front sensor

It is.

D1000 (general brain activity model):
Simplified Wilson-Cowan model as in Example 1.

S800 (specification of calibration minimization requirement):
τ, a, b, c, d, ρ x, ρ y, the four sensors - pairs _{.S 1 -T 1, S 1 -T} 2, S 2 -T 1, S 2 -T 2 - mediators It is a transfer operator. Only local calibration is required. This is because it is assumed that both frontal brain regions are independent (simplification). Each endogenous input is assumed to be constant.

D1100 (minimized calibration request):
Four sensor-transmitter pairs: S ₁ -T ₁ , S ₁ -T ₂ , S ₂ -T ₁ , S ₂ -T ₂ τ, a, b, c, d, ρ _x , ρ _y , migration Calibrate the operator locally.

E100 (Do you need calibration?):
Local calibration is necessary.

D4100 (device restriction):
Similar to Example 1, on the output side, for example, there are two coils with a maximum magnetic field of 0.5 Tesla.

D4200 (health constraints on exogenous electromagnetic fields):
As in Example 1.

S2000 (calibration):
Explain later.

D1200 (specific brain activity model):
As in Example 1.

D1300 (quantification of the influence of exogenous electromagnetic fields on brain activity):
As in Example 1.

S3000 (control or regulation of brain activity by BAI control variables and target correction by measurement / calculation):
This will be described in detail when handling each flowchart of FIG.

  FIG.

D1100 (minimized calibration request):
Same as above.

D1110 (calibration type):
local.

D1120 (calibration range):
Sensors S1, S2, transmitter substances T1, T2.

E100 (calibration?):
Yes.

S2100 (local calibration):
In analogous to example 1, four sensors - mediators - pair _{.S 1 -T 1, S 1 -T} 2, S 2 -T 1, S 2 -T 2. Since it is a considerable simplification when performing local calibration four times, the mode parameters and intrinsic inputs that are detectable by S ₁ are already detectable when calibrating pair S ₁ -T ₁ locally. Therefore, it can be regarded as given in S ₁ -T ₂ at calibration (especially during decomposition among other processes). It is similar to the mode in the detection range of S _2. In conjunction with the sensor, it is recommended to start a local calibration of the transmitter with the smallest spatial distance for this sensor.
The possibility of the result of the local calibration, the sensor S ₁ is for example:
40 ^* [Magnetic field strength at T ₁ position Tesla] ^* Example ₁ mode I (xi ₁ ) with T ₁ effect calibrated with xi ₁ (t), and T ₂ effect = 0. (T) is the time delay feedback of the signal measured by S ₁ in the reconstructed form).
For the sensor S _2:
40 ^* [Magnetic field strength at T ₂ position Tesla] ^* Mode _{2 of} Example ₁ (xi ₂ ) with T ₂ effects calibrated with xi ₂ (t), and T ₂ effects = 0. (t) is the time delay feedback signal measured by S ₂ in the reconstructed form).

S2300 (identification of the same brain activity):
The above modes are not identical.

E120 (Trans-local calibration?):
No.

D1200 (specific brain activity model):
Same as above.

D1300 (quantification of the influence of exogenous electromagnetic fields on brain activity):
Same as above.

  Figures 22-23:

  As before, the same as that disclosed in S2100.

  FIG.

  In this example, only local calibration is required, so it is omitted.

  FIG.

It is similar to Example 1 and will be described next with the exception of S3100. : For example, the combined signal of 0.4 Tesla T ₁ , xi ₁ (t) and 0.4 Tesla T ₂ , xi ₂ (t) increases Pos by 200% or more of the original value in mode I (example) 1), induce a Neg decrease of less than 50% of the original value in mode II (compare FIG. 12 (ii) with FIG. 18 (iv)). Brain activity targets are achieved as a result. Similarly, behavioral targets are achieved in the case of the effectiveness of the Davidson model.

  Example 3:

  Example 3 is not described in its entirety as in the previous example (eg, all modes, all sensor-transmitter-pair local calibrations, etc.), but mainly highlights significant differences from the previous example.

  FIG.

D2000 (behavioral target):
Increased concentration of attention to mental activity during the application period.

がある（最大＝５０）。 D3000 (behavior model):
For example, based on [20] and [21], Fmθ (frontal midline theta), ie the action of 6-7 Hz in the vicinity of the Fz-electrode (see [16]), matches the concentration of attention to mental activity. This fact is then called the Ishihara-Yoshii-model.
Neurons near the skull within the detection range of the Fz-sensor are directly connected to other regions, but in particular, subcortical drive is detected. The power spectrum was calculated previously. For each sensor instead of Pos and Neg,

There is (maximum = 50).

D4000 (minimization device requirement):
It is meaningful to increase the Fz electrode by at least one adjacent electrode in order to determine the phenomena that may spread. The detection range of the Fz electrode (eg, Cz electrode) is directly physiologically coupled to the detection range of the surface EEG with two measurement electrodes and two reference electrodes (as indicated by [21]). Two coils (transmitters) are located in the immediate vicinity of each electrode. : T ₁ is the front part of Fz, and T ₂ is the rear part of Cz. Everything else is the same as Example 1.

D2100 (Target BAI-dynamics):
For example, Att ₁ (with influence) = Att ₁ (no influence, all driving), Att ₂ (with influence) <= Att ₂ (no influence), where “1” is “Fz and reference electrode”, “2” is “Cz and reference electrode” and “=” indicate “the difference between the right side and the left side is less than 20%”.

。 D2200 (BAI calculation rule):
In addition to the power spectrum of Example 1, for both sensors,

.

に分かれるが、「ｈ_jl」はｊ番目の神経振動子のｌ番目の影響の強さであり、各Δはこの影響が有効になるまでの時間遅延である。 D1000 (general brain activity model):
The simplified Wilson-Cowan model as in Example 1, but now paying attention to the translocal effect, there is no assumption that the endogenous input is constant. The intrinsic input to the excitatory subpopulation to the jth neural oscillator is the easiest case for translocal calibration:

“H _jl ” is the strength of the l-th influence of the j-th neural oscillator, and each Δ is a time delay until this effect becomes effective.

S800 (specification of calibration minimization requirement):
Function matrix in addition to Example 2. Local and translocal calibration is required.

D1100 (minimized calibration request):
Function matrix in addition to Example 2.

E100 (Do you need calibration?):
Yes.

D4100 (device restriction):
As in Example 2.

D4200 (health constraints on exogenous electromagnetic fields):
As in Example 2.

S2000 (calibration):
This will be described when the flowchart of FIGS. It is assumed that the intrinsic input during calibration is steady, not the entire application.

D1200 (specific brain activity model):
As in Example 1.

D1300 (quantification of the influence of exogenous electromagnetic fields on brain activity):
As in Example 1.

S3000 (control or regulation of brain activity by BAI control variables and target correction by measurement / calculation):
This will be described in detail when handling each flowchart of FIG.

  FIG.

D1100 (minimized calibration request):
Same as above.

D1110 (calibration type):
Translocal (including local).

D1120 (calibration range):
As in Example 2.

S2100 (local calibration):
This will be described in detail when handling the flowcharts of FIGS.

S2300 (identification of the same brain activity):
Similar to Example 2.

E120 (Trans-local calibration?):
Yes.

S2400 (trans-local calibration):
This will be described in detail when considering FIG.

S1200, D1300:
Function matrix in addition to Example 2.

  FIG.

Similar to Example 1. This time the BAI model-mode within the detection range of S1 is concerned, but the main parameters are τ = 10, a = 16, b = 20, c = 8, d = 3, and the intrinsic input is ρ _x = 1.5 + m ^* sin (2 ^* π ^* t ^* 7/1000) and ρ _y = 1.5 + m ^* sin (2 ^* π ^* t ^* 7/1000), and m = 0 in FIG. 19 (i). FIG. 19 (ii) shows m = 2 and FIG. 19 (iii) shows m = 4 (in this example, 7 Hz subcortical activity has the same impact on the excitatory subset as on the inhibitory subset. ). This mode is called mode III. A test signal close to 7 Hz sinusoidal activity is probably appropriate, for example, for a 7 Hz sine signal and switches off every 10 milliseconds for 5 milliseconds for measurement. Such a test signal transmitted at 0.5 Tesla by the transmitter T ₁ has an effect on mode III with a multiplier of 16. The transmitter T ₂ has an effect with a multiplier of 0.

Here, the local calibration includes the phase deviation of the test signal. The BAI minimum of the activation signal provides the phase deviation theta necessary for mode switch-off. For example, the sine signal is transmitted on the phase deviation theta at 0.5 Tesla by the transmitter T ₁ . The result is shown in FIG. 19 (i).

Only the S ₂ mode is the S ₁ mode of Example 1. Both T ₁ -signals have no impact on the S ₂ mode. 0.1 Tesla sign signal from the T ₂ does not have an impact to the theta mode.

  FIG.

  Similar to Example 2.

  FIG.

D1300 (moving operator):
If i and k are different, these transfer operators are equal to zero for both signals shown in FIG.

S2410 (BAI prediction):
Same as above.

Regarding the S ₁ -mode: S2420 (correction):
Illustrative example: Transmission of a 0.5 Tesla 7 Hz sine signal by T ₁ phase shifted by theta causes the S ₁ -mode to be switched off.

Regarding the S ₁ -mode: S2430 (S ₂ -determining the gap): Since the signal has no direct impact on the S ₂ mode, the BAI prediction for S ₂ is “invariant continuation”. This prediction is consistent with the BAI on facts about S _2.

Regarding S2-mode: S2420 (correction):
Illustrative: Although shifts to the fundamental frequency of the frequency of the S ₂ mode, a 7Hz incompatible. For example, by time delayed feedback of the signal measured by S ₂ , transmitted by T ₂ and of opposite polarity.

Regarding the S ₂ -mode: S2430 (S ₁ -determining the gap):
Since the T ₂ signal has no direct impact on the S ₁ mode, the BAI prediction for S ₁ is “continue unchanged”. This prediction is consistent with the BAI on facts about S _2. As a result, for example, changes in the signal component are shown in FIGS. 19 (iii) to 19 (ii).

S2440 (calculation of function matrix accounts for gaps):
Since there is no gap in the case of change of S _1, cell (1,2) is equal to zero (mode 1 = display S ₁ mode, Mode 2 = Display S ₁ mode). There is no inherent gap in cell (2,1).

E400 (unique?):
Clarify if one or more function matrices result from S2440.
If there is no interest in the general correspondence between brain activities for each application, but a particular correspondence, for example between brain activities i and j, cells (i, j) and (j, i) in the function matrix It is clear that satisfying
In a given example, it is only important whether the mode 1 BAI-save modification has spread, ie, a mode 2 change. For this purpose, consideration of cell (1,2) is sufficient-but the value has zero.

S2221 (signal change):
For example, frequency shift generation, amplitude change, stationarity change, noise addition, Lyapunov exponent change, and many more (not necessary in this example).

D2410 (function matrix):
Each result consists of one function matrix, for example, in the following form: :

  FIG.

Similar to Example 2. In addition to this, the type of propagation limit (D4300) has been determined by translocal calibration. : "Change due to T ₁ does not affect other brain activity than those that have been captured by the S _1" is fulfilled. In a given example, Fmθ stabilizes in the vicinity of the natural spectrum when the drive decays to zero (see FIG. 19 (i)) (the discontinuous sign by T ₁ without disturbing the brain activity detected by S ₂ . “Full drive” by signal FIG. 19, (ii) and (iii)). The results (measured by S ₁ ) are shown in FIG. 19 (iv). Without disrupting Att ₂ , the brain activity target of reconstructed Att ₁ is achieved as a result, and the same applies to behavioral targets in the case of Ishihara-Yoshii-model application.

  Further explanation and note:

脳活動について、ｘ脳活動指標、ｔ時間。「外因性入力」は、脳活動モデルの方程式において人為的に（頭蓋外で誘発された電磁界により）生成された入力変数の値に言及する（「外因性入力」と反対に、通常は感覚または神経チャネルを介して、例えば光による刺激、視床ペースメーカー作用、交響曲鑑賞、さらに多くにより伝達される）。外因性入力＝

（伝達物質シグナル）、

移動作用素。関数fは使用脳活動モデルに依存する。また観察不可能性パラメータのセットで、ｘ、ｔ、観察不可能性内因性および外因性入力によるｘの時間微分を決定する。一般的な脳活動モデルは、点状のｘの代わりにパス、および／また確率微分方程式を使用するが、雑音を含む。
異なる脳活動モデルにより異なる校正結果となり、異なる修正の可能性がある。適切な脳モデルの場合、脳活動標的が信頼性の高い方法で達成されるであろう、適切な行動モデルの場合、行動標的が信頼性の高い方法で達成されるであろう。脳活動モデルと行動モデルは統合形式でも表示可能である。 Ad D1000: A brain activity model is, in the simplest case, a set of difference equations and / or differential equations, for example of type

For brain activity, x brain activity index, t time. “Exogenous input” refers to the value of an input variable generated artificially (by an extra-cranial electromagnetic field) in the brain activity model equation (as opposed to “exogenous input”, usually sensory Or transmitted through nerve channels, eg by light stimulation, thalamic pacemaker action, symphonic appreciation, and more). Exogenous input =

(Transmitter signal),

A moving operator. The function f depends on the brain activity model used. It also determines the time derivative of x with x, t, unobservable intrinsic and extrinsic inputs with a set of unobservable parameters. Common brain activity models use paths and / or stochastic differential equations instead of point-like x, but contain noise.
Different brain activity models result in different calibration results and different correction possibilities. In the case of a suitable brain model, the brain activity target will be achieved in a reliable manner, and in the case of a suitable behavior model, the behavioral target will be achieved in a reliable manner. The brain activity model and the behavior model can also be displayed in an integrated format.

Ad S2000:
The non-linearity of the system allows the determination of unobservability, including mobile operators, by appropriate variation of transmitter signals (“test signals”) with as little magnetic field as possible.
The dynamic calibration shown here is a parameter estimation (generally these make parameter distribution assumptions that are considered random variables and then observe the expansion along with the trajectory. Also from these observations towards the "true" parameter Existing mathematical methods for estimating the latter, see assumptions of convergence of the calculated parameter values (see [7]), and adaptive control / parameter estimation known from the art (generally these are linear systems Only useful if the primary control function differs from that incorporated in the system to minimize the difference between the ideal target value and the observed value (see [8]).

  Dynamic calibration is the “passive observation” (conventional non-invasive observation) in the mathematical approach described above, by using “active measurement” (measurement of the system in transmitting and interacting with the signal) And different. The results of active measurements (eg stochastic resonance) are often non-linear deterministic and / or non-linear stochastic systems (such as the human brain) and are often very sensitive to the response of linear systems to normal Dirac or snake side inputs, etc. Different.

There are significant differences between adaptive control / parameter estimation and dynamic configuration. :
The first dynamic calibration is usually a preceding control (and a process separated from the control).
In the second dynamic calibration, the test function used is emphasized, but the multiplicity (and the ability to distinguish between different sets of resulting unobservability) is consistent with the complexity of the system considered ( In contrast to simple test functions such as the snakeside function, system responses often need to be observed over a long period of time, but not in-vivo).
Third The type of approximation of the main parameter set is completely different (in dynamic calibration, it is a continuous elimination in various truncation parameter subsets as opposed to target / de facto comparison of control parameters).

  In addition to determining unobservability, polymorphism that translates characteristics of magnetic fields artificially generated by dynamic calibration (eg, magnetic field strength and / or frequency distribution of extracranial magnetic fields) into extrinsic inputs Other unknown mobile operators in the system and / or unknown complex system are also determined.

Ad D2410:
The function matrix for the observed m brain activity is the m ^* m-matrix, and the (i, j) that is allowed if no functional connection can be determined from activity j to activity i is zero.

  The functional connection between brain activities requires a physiological connection of the major brain regions, but the latter does not necessarily lead to functional linkage.

A function matrix is accepted if it is a solution that is comparable to observations for the overall observed brain activity (after truncation of parameters). Here, in the equations for each brain activity, the endogenous input is exchanged for example, residual input + sum of all upstream activities multiplied by the strength of each effect multiplied by z _j (t-delay) (change of A changes B In this case, “upstream regarding B” corresponds to: “downstream regarding B regarding A”).

A preferred implementation of the method for local correction includes the following elements (assuming a group of adjacent Wilson-Cowan oscillators can be modeled by a pair of Wilson-Cowan equations). :
(A) Data analysis by Fourier analysis is used to determine the possibility of measured signals belonging to the equivalence class of solutions of the Wilson-Cowan equation, or wavelet analysis (basic, fundamentally, x (t) and y ( Appropriate linear combination of t) (see [13]), phase space embedding, combined with statistical methods. A suitable test signal is a time-delayed feedback signal (using the measured signal) and / or a linear combination of the Wilson-Cowan equation μ ^* (x (t−τ) −y (t−τ)). .
(B) Local calibration,
(C) modify as described,
(D) Automatic preventive control in addition to the continuous seizure warning as in [9], for example.

  Similar to this is the preferred embodiment for translocal calibration in addition to the translocal phenomenon described above.

  When calibration is not required, or when the limit cycle oscillator is weakly coupled, and for a short period of time, a weakly coupled limit cycle oscillator can be converted to a group of phase oscillators instead of a limit cycle oscillator If the assumption is valid (solution of Wilson-Cowan equation), a coupled phase oscillator (see [23]) can be used. Intervention is based on continuous measurement of the signal in addition to phase reset and subsequent external slaving so that the cluster converges to the target. This embodiment is obviously applicable to a localized brain region so that unwanted interference does not occur due to reset.

Phase space embedding is illustrated for one input channel. :
FIG. 26 shows EEG data of one channel as a unit time (1/128 seconds) on the X-axis and a potential difference (between the measurement electrode and the reference electrode at each point in time) on the Y-axis. In the simplest case, one or more trajectories are determined based on [29]. In general, embedding by delay coordinates is performed (see [27]). The delay time can be determined as the first zero cross of the autocorrelation function (see [29]). : Data window padding for t = 130 and τ = 86 (FIG. 27). The x-axis shows a window of length 32α ^* x (t−86) + β ^* y (t−86), ending at t, and the y-axis shows α ^* x (t) + β ^* y (t). In FIG. 27, α is given 1, and β is given -1. Multiple occupancy of the same point is not displayed, and the number of points in FIG. Empirical phase space occupancy with all methods (on finite parameter grid) of Wilson-Cowan equations at each mixing angle (on truncation interval) by known methods (Bayes, Minimax, and many others) and after remeasurement Is determined. 28 (same region as in FIG. 27) and the same mixing angle is τ = 10, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5, for some ρ _y : diamond ρ The solution of _y = -1.5 and square y = 0 is shown. FIG. 29 shows the importance of the mixing angle. : As previously noted, where τ = 1, a = 22, b = 20, c = 14, d = 3, ρ _x = 1.5, ρ _y = 0, and α = 1, β = Obtain solutions for -1 (diamond), and β = -0.7 (square) and β = -0.3 (intersection).

In a preferred embodiment of the method, instead of the Wilson-Cowan equation, the stochastic Wilson-Cowan equation is used, ie ρ _x = ρ _{x, det} + noise, ρ _y = ρ _{y, det} + noise. “Det” indicates a deterministic part of the input, and “noise” indicates a probability input. The same method applies as before. In addition to this, the noise level is determined. Here, the use of a noise test signal and the measurement of the signal-to-noise ratio are also meaningful.

  In another preferred embodiment of the method, different time constants t are used for x and y and / or different multipliers m in the Wilson-Cowan equation. These calibrations are performed as already described for the other parameters.

、同様に

ε、δ、η、ν、新規の観察不可能性であり、決定する必要がある。
これに類似で、この実施様態は確率Wilson−Cowan−ニューロンについては、時間遅延入力有り、または無しで実行可能である。 In another preferred embodiment of the method, the other neurons of the cluster are coupled via mean field coupling to each Wilson-Cowan-neuron of the cluster in the Wilson-Cowan equation, i.e. all i = 1. . N

As well

ε, δ, η, ν are new observability and need to be determined.
Analogously to this, this embodiment can be performed for stochastic Wilson-Cowan-neurons with or without time delay inputs.

  In yet another preferred embodiment of the method, the shape of the sigmoid function of the neurons in the cluster is calibrated by using a noise test function, because depending on the sigmoid function, the cluster amplifies or attenuates signal noise with different strengths, respectively. (For simple networks, see eg [28]).

  In another preferred embodiment of the method, the test function and this test function of opposite polarity are used alternately, ie first the signal xi (t) and then -xi (t).

  In another preferred embodiment, calibration stops after n> 0 test signal (“fast calibration”), and one set of parameters, intrinsic and extrinsic inputs that match the measurement data is set as a pre-calibration result. . Another measurement data obtained as a result of performing the correction (S3000) is used for the adaptation of the continuous calibration.

  In another preferred embodiment, the measurement premise and measurement invariance (D1049, D1050) are confirmed.

  In a preferred embodiment, the EM method is combined with sensory input (auditory, visual, etc.). Example: The instruction to the user in Example 3 (see eg [20]) to start mental work at the beginning of the calibration greatly simplifies the task of finding the appropriate Fmθ-mode. Another example is the auditory pacemaker action for translocal calibration of auditory brain processes.

  In the preferred embodiment, the effectiveness of the method stabilizes over the duration of the individual application session by the use of an appropriate repetition rate but must be determined by external confirmation for each user. This result of a suitable repetition rate is known, for example, for TMS applications (magnetic field 1-2 Tesla transcranial magnetic stimulator) see eg [3].

  In a preferred embodiment, calibration is combined with sensor optimization: previously non-specific brain activity can be measured in particular by each (newly directed if necessary) sensor. Activation of active sensor and / or change of sensor position and / or change of sensor direction. For sensor optimization, conventional optimization methods are adapted.

In a preferred embodiment, calibration couples with sensor optimization by coupling multiple existing sensors to the virtual sensor so that brain activity is particularly measurable by each virtual sensor. Example: S ₁ , S ₂ , S ₃ , S ₄ are f (S ₁ , S ₂ , S ₃ , S ₄ ) = 0.3 ^* S ₁ + 0.5 ^* S ₂ + 0.02 ^* S ₃ ^* S ₃ It is connected to Svirtual by + 0.5 ^* sin (S ₁ + S ₄ ). The appropriate f is calculated for each subunit of the sensor by conventional optimization methods.

  In a preferred embodiment, calibration combines with transmitter optimization: previously identified and / or unspecified brain activity can be influenced in particular by each (newly directed if necessary) transmitter Is activation of inactive transmitter and / or change of transmitter position and / or change of transmitter direction. For the optimization of transmitter, conventional optimization methods are adapted.

  In a preferred embodiment, calibration combines with transmitter optimization by combining a plurality of existing transmitters with the virtual transmitter so that brain activity can be sufficiently affected, particularly with each virtual transmitter. The optimal (virtual mediator) is calculated from the mobile operator by using conventional optimization methods.

  In a preferred implementation of S2130, a decomposition of parallel n sensors (n is an integer between 1 and the total number of sensors used) is performed. Therefore, the time series to be resolved is a 1-sensor-time series n-tuple.

  In a preferred implementation of S2130, decomposition is performed on 1-sensor-time-series n-tuples by using a pursuit method (see, for example, battle pursuit method [13]).

  In another preferred embodiment of S2130, a 1-sensor by Karhunen-Loeve-method (see eg [17], [18]) and / or independent component analysis (see “ICA” eg see [22]) -Decomposition is performed on time-series n-tuples.

  In another preferred embodiment of S2130, decomposition is performed after embedding in the metaphase space, resulting in separation of stationary and non-stationary components (see, eg, [15]).

  In another preferred embodiment of S2130, the decomposition is performed in parallel with the plurality of methods. As a result, only the modes identified by many of these methods (under defined weights) are accepted for further processing (eg, for relevance testing).

  In a preferred embodiment of D2000, alteration or maintenance of perception and / or perception ability and / or perceptibility is set as a behavioral target. An illustration of this is a modification of the ability to distinguish between specific stimuli or types of stimuli.

  In preferred implementations of D2000, changes or maintenance of behavior and / or behavioral capabilities and / or behavioral possibilities are set as behavioral targets. The illustration here improves the reaction rate.

  In a preferred embodiment of D2000, activation and / or alteration or maintenance of activation ability is set as a behavioral target.

  In another preferred embodiment of D2000, alteration or maintenance of motivation and / or dynamic functional power and / or motivability is set as a behavioral target.

  In another preferred embodiment of D2000, changing or maintaining attention and / or attention ability is set as a behavioral target.

  In another preferred embodiment of D2000, changes or maintenance of memory and / or stored content and / or retrieval of stored content is set as a behavioral target.

  In another preferred embodiment of D2000, the change or maintenance of learning and / or learning ability and / or learnability is set as a behavioral target.

  In another preferred embodiment of D2000, changing or maintaining consciousness is set as a behavioral target.

  In another preferred embodiment of D2000, alteration or maintenance of emotion and / or emotional ability and / or emotional potential is set as a behavioral target.

  In another preferred embodiment of D2000, desire and / or aversion change or maintenance is set as a behavioral target.

  In another preferred embodiment of D2000, the change or maintenance of recognition and / or cognitive ability and / or recognizability is set as a behavioral target. (Here "recognition" is defined as "execution of psychological action").

  In another preferred embodiment of D2000, the behavioral sequence is set as a behavioral target.

  In another preferred embodiment of D2000, behavioral correlation is set as a behavioral target.

  In another preferred embodiment of D2000, multiple compatible behavioral targets bind.

  In another preferred embodiment of D2000, a plurality of not necessarily conformable action targets are combined hierarchically, i.e. a priority list of action targets is generated. In the case of procedural steps that are inconsistent with each other, the steps associated with the lower priority behavioral target are performed later or not at all.

  In another preferred embodiment of D2000, the list of behavioral targets is increased by the highest priority target without seizures. This includes applying [9] to the user without each chart.

  In another preferred embodiment of D3000, behavioral models are taken from a vast amount of literature (eg, for the examples shown in [11], [20], [21], the same / similar subject literature is Dozens).

  In another preferred embodiment of D3000, a plurality of behavioral models are used in parallel, resulting in a step compatible with a large number of these behavioral models (under defined weights), each of which is controlled or adjusted, respectively. Only (S3000) is executed.

  In a preferred embodiment, there is an external confirmation (usually a psychological test) of changing or maintaining an individual user's behavior model during and after EM application. As a result, usage behavior models (generally statistically confirmed) are tested for the degree of effectiveness for each user.

  In another preferred embodiment of D1000, multiple generic brain activity models are used in parallel, resulting in a large number of these behavioral models (under defined weights), each of which is controlled or adjusted, respectively. Only the adaptable step (S3000) is performed.

  In another preferred embodiment of the method, in the steps relating to a single sensor and / or a single transmitter, these replace a set of sensors and / or transmitters. The purpose is non-local space-time mode determination and controlled modification, for example in the Jirsa-Haken-model (see eg [18]).

  In another preferred embodiment of D4000, the minimum device requirements are exceeded to obtain additional information, but are potentially useful for further application of behavioral targets that may have changed (eg, 10 / Always use 20-surface EEG, see eg [16]). The advantage of this embodiment is the possibility of standardizing the parts of the device and the possibility of using commercially available hardware. Another preferred embodiment of the D4000 is minimal to obtain additional possibilities for control or adjustment (eg, each non-reference electrode of a 10/20 surface EEG to have one coil in front of the electrode). Exceeds device request. The advantage of this embodiment is the possibility of standardization of device parts.

  In another preferred embodiment of S3000, the extrinsic electromagnetic field that results in an increase in control variables is enhanced by sensory input and / or biochemical activators, and the procedural steps described as a result remain unchanged in principle.

  In another preferred embodiment of the method, a test signal and a transmitter signal with a stochastic component are used (especially with a brain activity model with an explicit stochastic component, eg [25], [18] supplement, probability (For example, see [26] for the FitzHugh-Nagumo-model, along with targets that use resonance and stochastic coherence).

  In another preferred embodiment of the method, a test signal and a transmitter signal with a chaotic component are used (especially together with a brain activity model with an explicit chaotic component, eg between the oscillators as in the enhancement of [1] Strong local coupling).

  In another preferred embodiment of the method, the inactive mode is directly activated by the transmitter signal, i.e. by the direct effect of the signal on the mode.

  In another preferred embodiment of the method, the mode is indirectly influenced by the transmitter signal, ie a direct effect on the mode that induces an effect on the indirectly affected mode (see function matrix). by.

  In another preferred embodiment of the method, the bifurcation point is controlled, i.e. brain activity is qualitatively changed (as in Example 3, a neural oscillator that is inactive without driving is switched on. ).

  In yet another preferred embodiment of the method, the inactive mode is activated.

  In another preferred embodiment, the amplitude is amplified by inactivation mode activation and / or mode synchronization, and is similarly reduced by mode switch-off and / or mode desynchronization.

  In another preferred embodiment, amplitude amplification is used for age-related flattening (for EEG see eg [12]).

  In another preferred embodiment of the method, the non-stationary signal is broken down into a constant and non-stationary part (see, eg, [15]), resulting in modulation of the test and transmitter signals along with the non-constant part.

  In another preferred embodiment of the method, the user is identified based on calibration data, but is stable for a long time. Even conventional spectrally resolved EEG-data for individuals has components that are stable over time (experimentally valid for up to 5 years, eg [10]).

Related publications
[1] "Excitatory and inhibitory interactions in localized populations of model neurons", Wilson HR, Cowan JD, Biophysical Journal 1972; 12: 1-22
[2] WO 00/29970 A1
[3] "Therapeutic application of repetitive transcranial magnetic stimulation: a review", Wassermann EM, Lisanby SH, Clin Neurophysiol 2001; 112: 1367-1377
[4] WO 98/18384 A1
[5] "Anticipation of epileptic seizures from standard EEG recordings", Le Van Quyen M, Martinerie J, Navarro V, Boon P, D'Have M, Adam C, Renault B, Varela F, BaulacM, The Lancet 2001 Jan 20; 357: 183-8
[6] WO 01/21067 A1
[7] "Numerical Solution of Stochastic Differential Equations", Kloeden PE, Platen E, Springer, Berlin 1992, chapter 6.4
[8] "Regelungstechnik", F ▲ o ▼ llinger O, H ▲ u ▼ thig Buch Verlag, 8th rev. Ed., Heidelberg 1994, chapter 1.6
[9] DE 102 15 115.6
[10] "Intraindividual Specificity and Stability of Human EEG: Comparing a Linear vs a Nonlinear Approach", D ▲ u ▼ nki RM, Schmid GB, Stassen HH, Method Inform Med 2000; 39: 78-82
[11] "An EEG Biofeedback Protocol for Affective Disorders", Rosenfeld JP, Clin Electroencephalography 2000; 31 (1): 7-12
[12] "Leitfaden f ▲ u ▼ r die EEG-Praxis", Ebe M, Homma I, Gustav Fischer, Stuttgart 1992, chapter 6.2.2
[13] "A Wavelet Tour of Signal Processing", Mallat S, Academic Press, San Diego 1999,
[14] "Bioelectricity", Reilly JP, Springer, New York 1998, chapter 11
[15] "Stationarity and nonstationarity in time series analysis", Manuca R, Savit R, Physica D 1996; 99: 134-161
[16] "Leitfaden f ▲ u ▼ r die EEG-Praxis", Ebe M, Homma I, Gustav Fischer, Stuttgart 1992, chapter 3.6.1, and 3.6.2
[17] "Nonlinear Spatio-Temporal dynamics and Chaos in Semiconductors", Sch ▲ o ▼ ll E, Cambridge University Press, Cambridge 2001, Kapitel 7.3.4.
[18] "A derivation of a macroscopic field theory of the brain from the quasi-microscopic neural dynamics", Jirsa VK, Haken H, Physica D1997; 99: 503-526
[19] "Approaches to verbal, visual and musical creativity by EEG coherence analysis", Petsche H, Int J Psychophysiol 1996; 24: 145-159
[20] "Frontal midline theta rhythm and mental activity", Inanaga K, Psychiatry and Clin Neurosciences 1998; 52: 555-566
[21] "Frontal midline theta rhythms reflect alternative activation of prefrontal cortex and anterior cingulate cortex in humans", Asada H, Fukuda Y, Tsunoda S, Yamaguchi M, Tonoike M, Neuroscience Letters 1999; 274: 29-32
[22] "Imaging Brain Dynamics Using Independent Component Analysis", Jung TP, Makeig S, McKeown, Bell AJ, Lee TW, Sejnowski TJ, Proceedings of the IEEE July 2001; 89 (7): 1107-1122
[23] "Desynchronizing double-pulse phase resetting and application to deep brain stimulation", Tass PA, Biol Cybern 85,2001: 343-354
[24] "Mathematical aspects of Hodgkin-Huxley neural theory", Cronin J, Cambridge University Press, Cambridge 1987, chapters 2 and 3
[25] "Impacts of noise on a field theoretical model of the human brain", Frank TD, Daffertshofer A, Beek PJ, Haken H, Physica D 1999; 127: 233-249
[26] "Coherence and stochastic resonance in a two-state system", Lindner B, Schimansky-Geier L, Physical Review E June 2000; 61 (6): 6103-6110
[27] "Analysis of Observed Chaotic Data", Abarbanel HDI, Springer, New York 1996, Kapitel 3.3
[28] "lntroduction to Neural Dynamics and Signal Transmission Delay", Wu J, DeGruyter, Berlin 2001, chapter 3.4
[29] "Periodic Orbits: A New Language for Neuronal Dynamics", So P, Francis JT, Netoff TI, Gluckman BJ, Schiff SJ, Biophysical Journal 74, June 1998: 2776-2785

Transmitter shown in cross section The transmitter of Figure 1 shown from below Planar projection of a sensor and a drill hole for the transmitter reflecting the position in the helmet Helmet with overhead suspension and chin Another planar projection of sensor and transmitter drill holes based on position in helmet Schematic of the interaction of the head unit, intermediate unit, and base unit BAI-model-mode under τ variation (extraction) BAI-model-mode (extraction) under fluctuation of a BAI-model-mode (extraction) under fluctuation of b BAI-model-mode (extraction) under fluctuation of c BAI-model-mode (extraction) under fluctuation of d Two different models-similar BAI of mode I and II-model-mode (sample) BAI-model-mode of sine-shaped exogenous input with different intensities (example) BAI-model-mode of sine-shaped exogenous input II of different strengths (example) BAI-model-mode (example) under low-strength extrinsic elementary loop I BAI-Model-Mode (Exemplary) under low-intensity extrinsic elementary loop II BAI-Model-Mode (Exemplary) under medium-strength extrinsic elementary loop I BAI-Model-Mode (Exemplary) of Medium Strength Exogenous Elementary Loop II Theta-BAI-Model-Mode (example) with different endogenous inputs and one exogenous input Overall method flowchart Flow chart of method S2000 (calibration) Flowchart of method S21000 (local calibration, first part) Flowchart of method S21000 (local calibration, second part) Flowchart of method S2300 (trans-local calibration) Flowchart of method S3000 (control and / or regulation of brain activity, respectively) Diagram showing data from one EEG-channel Diagram showing the phase interval representation of the part of the data from FIG. Diagram showing the phase interval representation of neural action under various inputs modeled by the Wilson-Cowan model Diagram showing a phase interval representation of neural action at the same input under different mixing angles modeled by the Wilson-Cowan model

Claims (54)

An electromagnetic correction method for brain activity to achieve a behavioral target, using a behavior model that explains the correspondence between behavior and the dynamics of brain activity indicators, and a brain activity model that quantitatively explains brain activity It is possible to use and get brain activity indicators and their dynamics:
-Decisions made by behavioral models that target the dynamics of specific brain activity indicators corresponding to the achievement of behavioral targets;
-Methods involving induction of extrinsic electric and / or magnetic fields to achieve target dynamics as part of open or closed loop control, including measurement of brain models, measurement of electromagnetic brain activity, and calculation of brain activity indicators .   A claim that uses a general brain activity model and is calibrated to a specific brain activity model by determining unobservability for a specific user, and the influence of the extrinsic electric field and / or magnetic field on the brain activity is determined for this user Item 2. The method according to Item 1.   The method according to claim 1 or 2, wherein an electromagnetic variable describing the brain activity or a variable derived from such an electromagnetic variable is used as the brain activity index.   The method according to claim 3, wherein the electromagnetic variables include potential and / or current and / or magnetic field measured extracranial and / or intracranial.   The brain activity index includes the frequency of the electromagnetic variable in the Fourier base, the wavelet base, or the Karhunen-Loeve base, and / or the time series expression coefficient of the electromagnetic variable, and / or the probability such as stationarity, moment, correlation, etc. 5. A method according to claim 3 or 4, wherein the indices and / or non-linear indices including stability, Lyapunov exponent, limit cycle, minimum dimension of embedded space, etc. are included.   6. Target dynamics according to one of claims 1 to 5, wherein the target dynamics are calculated from the behavioral target by a behavioral model, such as, for example, a Davidson model, an Ishihara-Yoshii model, or a variant of the chaotic seizure model according to [9]. The method described.   The method according to one of claims 2 to 6, wherein the unobservability is calibrated with respect to a particular user.   The method according to one of claims 1 to 7, wherein the influence of an extrinsic electric field and / or magnetic field on the brain activity of a particular user is determined.   The method according to one of claims 2 to 8, wherein the calculation of the influence of the extrinsic electric field and / or magnetic field on the brain activity of a particular user is combined with the determination of unobservability.   The first calibration includes decomposing (S2130) de facto brain activity index dynamics into BAI-modes, the BAI-modes corresponding to the brain activity index dynamics calculated from the brain activity model. 10. The method according to one of 2-9.   11. The method of claim 10, wherein based on these modes, a set of parameters and intrinsic inputs and / or extrinsic inputs are determined and are the basis for each mode according to a brain activity model. The test signal is calculated (S2160)
The test signal is an extrinsic electric field and / or magnetic field applied to the user,
As a result, the user's brain activity is measured,
Brain activity measured during or after application of the test signal is evaluated along with previous brain activity,
The method according to one of claims 1 to 11, wherein the result of this evaluation determines the set of influences of the test signal on each of the modes in addition to the parameters, and is compatible with the measurement data.   13. The test signal is modified and reapplied until the set of parameters, intrinsic inputs and test signal effects on the mode being considered is uniquely determined, whereby the mode is calibrated. the method of.   14. The method according to one of claims 12 to 13, wherein the calibration of brain activity is performed by calibration of modes and interaction between modes, the former or both taking into account extrinsic inputs.   The method according to one of claims 1 to 14, wherein fast calibration selects a set of parameters, intrinsic inputs, extrinsic inputs that are compatible with the measurement data after transmission of one test signal or several test signals.   A calibrated brain activity model is used to calculate which control variable mechanics are appropriate for the conversion of brain activity indicators to the actual mechanics of the target mechanics, so that the control variables are able to capture the extrinsic inputs in the brain activity model. 16. The method of claim 15, wherein the method is shown and related to an exogenous magnetic field and / or electric field.   The method of claim 16, wherein the calculated mechanics of the control variable is simplified by application of a signal with known effects.   18. The local mode is calibrated for several detection ranges and the interaction is determined by successive modifications of the individual local modes (S2420, S2430, S2440, S2221). the method of.   19. A method according to one of claims 1 to 18, in addition, magnetic and / or electric field sensory inputs and / or biochemical activators are used.   20. A method according to any one of the preceding claims, wherein brain activity is corrected or stabilized over the period of application with an appropriate repetition rate.   21. A method according to one of the preceding claims, wherein several sensors are interconnected for the measurement of electromagnetic brain activity.   The method according to one of claims 1 to 21, wherein a transmitter substance is used for the induction of the electromagnetic field and the position and direction can be changed.   23. A method according to claim 1, wherein several transmitters are used for electromagnetic field induction, some of which are interconnected.   The method according to one of claims 10 to 23, wherein a decomposition of the n-element time series into a space-time mode is performed.   The method according to one of claims 10 to 24, wherein the disassembly is performed by a follow-up method.   26. A method according to any one of claims 10 to 25, wherein a decomposition is performed on the transformed output function of the brain model, so that each transformation indicates a change in signal between neurons and each sensor involved in brain activity.   27. A method according to any one of claims 10 to 26, wherein the decomposition is performed after embedding in the metaphase interval so that the stationary and non-stationary components can be separated.   28. One of the claims 10 to 27, wherein the decomposition is performed in different ways, with the result that the set of modes is weighted according to each method, only these modes are further used and the total weight exceeds a certain threshold. The method described in one. 29. A method according to one of claims 1 to 28, wherein the behavioral targets include one or more subsequent behavioral targets:
Change or maintain perception and / or perception ability and / or perceptibility,
Change or maintain behavior and / or behavioral capabilities and / or behavioral potential,
Changing or maintaining the reaction rate,
Change or maintain activation and / or activation ability,
Change or maintain motivation and / or motive power and / or motive potential,
Change or maintain attention and / or attention ability;
Change or maintain memory and / or memory content and / or memory retrieval;
Change or maintain learning and / or learning ability and / or learnability,
Change or maintain consciousness,
Change or maintain emotions and / or emotional abilities and / or emotional potential,
Change or maintain desire and / or disgust,
Change or maintain recognition and / or recognition ability and / or recognition,
Changing or maintaining the sequence of actions,
Change or maintain behavioral correlation.   30. The method of claim 29, wherein a plurality of not necessarily compatible behavioral targets are combined hierarchically.   32. The method of claim 30, wherein seizure-free behavioral targets are given top priority in this hierarchy.   The plurality of weighted behavior models are used in parallel, whereby only their respective controlled or adjusted steps (S3000) are performed and are compatible with many of these behavior models. 32. The method according to one of claims 31 to 31.   The method according to one of claims 1 to 32, wherein an external confirmation of individual user behavior modification / maintenance of behavior is performed.   A plurality of weighted brain activity models are used in parallel, whereby only these respective controlled or regulated steps (S3000) are performed and are compatible with many of these behavioral models. The method according to one of 1 to 32.   35. A method as claimed in any one of claims 2 to 34, wherein a calibration request for an individual sensor replaces a calibration request for a set of sensors.   36. A method according to claim 1, wherein the induction and measurement of the electric field is carried out alternately.   36. The method according to one of claims 1 to 35, wherein the induction and measurement of the electric field is carried out simultaneously.   38. The method according to one of claims 14 to 37, wherein the test signal has a linear or non-linear and / or chaos and / or probability and / or non-stationary component.   The method according to one of claims 17 to 38, wherein the user is identified based on the calibration data and is stable for a long time.   40. A method as claimed in one of claims 17 to 39, wherein the calibration step and the correction step are carried out sequentially or in parallel for the sensor and transmitter group.   41. A method according to one of claims 17 to 40, wherein the branch point is controlled, and as a result a particularly inactive mode of the effect is activated.   42. A method according to any one of claims 18 to 41, wherein an amplitude amplification of age-related activity reduction is performed.   43. A method according to one of claims 1 to 42, wherein the method is performed automatically. A device for achieving a given behavioral target, using a behavioral model that explains the correspondence between behavior and the dynamics of brain activity indicators, and a brain activity model that quantitatively explains brain activity, In particular, it is possible to obtain a brain activity index and the dynamics of said index for the execution of the method according to one of claims 1 to 43, and consists of: :
-Means for determining and using behavioral models that target the dynamics of specific brain activity indicators corresponding to the achievement of behavioral targets;
-Means for inducing an extrinsic electric field and / or magnetic field with at least one transmitter and applying them to the user;
-One measuring device for measuring electromagnetic brain activity with at least one sensor;
-Means for calculating brain activity indicators to achieve target dynamics as part of open or closed loop control.   45. The apparatus of claim 44, wherein the measuring device is characterized by having a plurality of sensors that generally constitute a sensor grid.   46. An apparatus according to claim 44 or 45, wherein the device for inducing a magnetic field is characterized by having a plurality of transmission materials that generally constitute a transmission material grid.   47. Apparatus according to claim 44 to 46, characterized by a plan of at least one computer in which a software module for performing the method according to one of claims 1 to 43 is stored.   48. Apparatus according to claims 44 to 47, characterized by an electrostatic and / or magnetic shielding scheme for all sensors and all transmitters.   49. A device according to claims 44 to 48, characterized in that the measuring device is mechanically separable from other parts of the device so that the user can carry the measuring device.   50. Apparatus according to claims 44 to 49, characterized by the location of the sensor and transmitter substance outside the skull.   51. A device according to claim 44 to 50, characterized in that the sensor and the transmitter are localized inside the helmet that conforms to the shape of the skull of each user.   52. Apparatus according to claims 44 to 51, characterized by superposition of a sensor grid and a transmitter grid such that there is a transmitter in the vicinity of each sensor and a sensor in the vicinity of each transmitter.   To allow another transmitter to adapt to it, so that the transmitter density of the transmitter grid can vary locally and / or to change the angle of the individual transmitter in relation to the user's skull 53. Apparatus according to claim 44 to 52, characterized in that the fitting is planned in the transmitter grid.   The device can be disassembled into one head unit, one intermediate unit and one basic unit consisting of sensors and transmitters, so that the intermediate unit consists of a computer and software module that performs steps independent of calibration, and a basic 54. Apparatus according to claims 44 to 53, characterized in that the unit consists of a computer and a software module performing the steps depending on the calibration.

Images