ES2396811B1 - PROCEDURE AND SYSTEM FOR COMMUNICATION WITH SUBJECTS IN A STATUS OF DECREASED CONSCIOUSNESS. - Google Patents

Abstract

Procedure and system for communication with subjects in a state of diminished consciousness. # The present invention consists of a procedure or method of bidirectional and effective communication with people in a state of diminished consciousness (coma, vegetative state or similar) as well as in a system that It implements this procedure and allows communication to be achieved, generating the auditory signals that the subject receives in a state of diminished consciousness and processing the potentials evoked by his brain after processing said signals.

Description

PROCEDURE AND SYSTEM FOR COMMUNICATION WITH SUBJECTS IN A DECLINE CONSCIOUSNESS

TECHNICAL SECTOR

The field of application of this invention is that of man-machine interaction techniques and apparatus, applied in particular to subjects in a state of diminished consciousness (coma, vegetative state or the like).

The technologies and methods applied involve the fields of neuroscience, electronics and communications, smart computing, and rehabilitation.

STATE OF THE ART

Originally the Brain-Computer Interfaces, BCI (Brain-computer Interfaces), were systems designed for subjects with serious problems of a motor nature that made their communication difficult or very difficult. Since the subjects could not give orders to their muscles to carry out actions with which they could communicate (press a key, gesture, speak, etc ...) it was necessary to overcome this obstacle by placing some system capable of capturing the message that was he wanted to transmit before being sent to the muscles and divert it to a communications system to receive, interpret, and execute the corresponding command or message. In this way, a communication channel between the subject and the outside was established. Currently, its study and development has been generalized to a broader group of users and applications, including even recreational or biometric security applications.

For reasons of cost and convenience, BCIs are mainly based on Electroencephalography (EEG) analysis. EEG analysis has the advantages over other methods such as Functional Magnetic Resonance (RMF) or Electrocorticogram (ECoG), on the one hand, its low cost and on the other hand that EEG recording is performed in a non-invasive and harmless way for the subject. The EEG has a series of electrodes on the entire surface of the skull in order to transform the ionic currents generated during brain activity into electrical currents. These in turn will be amplified, filtered, sampled and conveniently converted to digital format to be processed by a computer. The computer will execute an algorithm that will analyze the data, along with other information such as the electrode it comes from and its timing with the rest of the electrodes and / or stimuli generated from the outside. With this information the computer will have to classify with the highest possible probability of success the message that the subject wants to communicate to the outside. Once the message is communicated, it is presented to the subject in real time so that he knows the effect caused and can internally readjust his brain activity through a process of trial and error and successive approaches. This technique, which is known as biofeedback, is the basis of mutual learning between system and subject and is typical of any communication system in general and of BCIs in particular.

As of today most of the applications are designed for subjects with some type of motor deficiency that makes it impossible for them to properly interact with the world around them. In this area, BCI systems enable a significant improvement in the quality of life of these subjects. Beyond this end, the BCI has been studied as an effective tool in the creative work of musicians, choreographers, etc.

Some of the typical applications of BCI systems are:

•

Communications: This is a application priority for sick

completely paralyzed. Various methods have been investigated, from

a simple binary application "yes / no" to virtual keyboards with which

can spell words. These applications have proven effective at

time of the transmission of information although the

number of characters

processed per minute is very low.

•

Control of Virtual Reality environments: Virtual reality I know uses in

BCI systems due to their degree of immersion, motivation and safe environment

and

bounded. Of this shape the subject interact single on a environment

reduced and the system will transfer the executed orders on that environment

to the real world.

• Neural Prosthetics: Another source of applications is in the restoration of movement for people with motor disabilities. Cortical signals have been used to control a robotic arm. Other studies have used signals drawn directly from the brain to move a hand.

5 virtual.

• Art: There are various projects such as "Neural Art", "Neural Music" or "cyberPRINT ', intended to use the new possibilities offered by BCI for artistic creation.

1O However, BCI systems currently have the following problems:

•

Bandwidth: The number of orders that can be executed per minute using a BCI system is very low.

•

Probability of error: The signals extracted from the brain are highly variable. Any unexpected stimulus causes brain activity to worsen

15 the analysis conditions. Also factors such as fatigue, medication, etc. unintentionally alter brain activity.

• Autonomy: The known BCI systems need the participation of assistants when placing and adjusting the electrodes. Also, most of these BCis systems cannot be started

20 independently by the user, but are externally initiated (if not, we would be facing the "King Midas" problem, where sometimes the user could be sending signals to the BCI system even without wanting to. An example of this phenomenon occurs when the subject is asleep or unconscious or is carrying out an activity

25 brain that has nothing to do with BCI control).

• Cognitive load: Most of the tests are carried out in laboratories in a controlled environment. Users can focus on testing without distraction. But the real world is not like that at all: emotional responses, interactions with other individuals, different

30 brain activities, etc. they influence altering the working conditions of the BCI.

OBJECT OF THE INVENTION

The first object of this invention is a procedure or method of bidirectional and effective communication with people in a state of decreased consciousness (coma, vegetative state or similar). Communication can be established with subjects who, even in a state of diminished consciousness, are able to receive external auditory stimuli, process them cognitively, and be able to voluntarily modify their brain activity.

The second object of this invention is a BCI system that implements the procedure and enables said communication to be achieved, generating the auditory signals that the subject receives in a state of diminished consciousness and processing the potentials evoked by his brain after processing said signals.

DESCRIPTION OF THE FIGURES

Figure 1.- Shows a user subject of the object of the invention and the device used for stimulation and biofeedback.

A: Indicates a subject in a state of diminished consciousness.

B: Device that presents both auditory stimulation and biofeedback. This device can consist of headphones, either external or inserted in the ear.

Figure 2.- Shows a diagram of the communication system with individuals in a state of diminished consciousness.

C: Subsystem 3, auditory stimulus reproducer.

D: System for fixing the electrodes to the scalp. Generally it will be an adjustable cap, suction cup, headband or headband with or without straps with a series of holes through which the electrodes are inserted, thus being fixed in the appropriate position. It can also be an array of sensors joined by elastics that adjust to the contour of the individual's head or through patches.

E: Electrodes. The electrodes collect ionic currents from brain activity and convert them into electrical currents.

F: Interface between the signal conditioning and registration subsystem and the subject. Generally a bundle of electrode hoses or a wireless interface.

G: Subsystem 2, for recording and conditioning the signal.

H: Interface between the signal conditioning and recording subsystem and the main subsystem.

1: Subsystem 1, Main.

J: Interface between Subsystem 1, Main, and Subsystem 3 reproductive of auditory stimuli.

Figure 3.- Flow diagram representing the process associated with the training phase of the system.

Figure 4.- Flow diagram representing the process associated with the system operation phase.

DETAILED DESCRIPTION OF THE INVENTION

The invention consists of a bidirectional communication procedure between subjects in a state of diminished consciousness and the outside world based on the analysis of patterns of brain activity generated by auditory stimulation.

In general, and after being consulted about a decision, preferably binary, the subject with diminished consciousness is stimulated in various ways through their auditory channels and different patterns of brain activity are measured, which translate into responses to that decision. Specifically, the patterns of brain activity from which the information necessary for the operation of the system procedure is extracted are those evoked by the stimulation of auditory stimuli, preferably PEA (Auditory Evoked Potential), P300x (Endogenous evoked potential that appears after 300 ms of the stimulus) and more preferably from the PEAEE (Steady State PEA). The procedure can also extract the necessary information from alpha, beta, mu, gamma bands and low frequency cortical potentials.

Throughout this specification, "decision" means any selection of options that the subject makes from among a set of available options. The simplest case would correspond to a binary answer (yes / no) selected from two possible options. In more complex cases, the decision could correspond to an order or instruction selected from a previously presented set.

For the procedure to work effectively, it is necessary to previously adapt the procedure to each subject. Not all subjects respond the same way to auditory stimuli, so in this phase, which we will call the training phase, the systems that implement this procedure will adapt or calibrate to present auditory stimuli with those characteristics for which the pattern of brain activity The subject has a better signal-to-noise ratio, so the system will have a greater ability to classify these patterns and, consequently, facilitate the identification of the decision that the subject wants to communicate.

Subsequently, the so-called operation phase will begin in which, after a question from the system, the subject's response will be associated with a previously determined pattern of brain activity during training.

The interaction between the subject and the system follows different steps depending on whether it is the training phase or the operation phase. Figures 3 and 4 show examples of execution flow charts of both phases, which will be described in detail later.

These phases are:

to.

Selection of set of auditory stimuli.

b.

Training phase, which includes the following steps:

i. A stimulus is selected and the subject is asked to discriminate it from among all the stimuli it will receive.

ii. The complete set of auditory stimuli is emitted.

iii. Simultaneously with step iii, certain brain patterns are recorded from which the parameters of interest for the stimulus classification are extracted and stored in the so-called feature vector.

iv.

We associate this vector of characteristics with the discriminated stimulus.

v.

Repeat steps iii to v for the selected stimulus to obtain a vector of average characteristics.

saw. Repeat step vi for each stimulus in the set.

vii. Elimination of the vectors of average characteristics in which the statistical distribution of the sample of its components corresponds to a noise.

c.

Selection of the subset of stimuli and their associated feature vectors not discarded in step b.vii

d.

Operation phase comprising the following steps:

i. Emission of the subset of stimuli selected in phase e)

ii. Simultaneously with step d.i, certain brain patterns are recorded from which the parameters of interest that will make up the vector of characteristics are extracted.

iii. The SNR ratio is calculated from the vector of characteristics obtained, during the emission of the stimulus subset, until it reaches a value statistically related to a probability of success determined in the classification.

iv.

The vector obtained is classified according to the stored vectors, thus identifying the associated stimulus

v.

The instruction associated with the identified stimulus is executed.

In the training phase and / or in the operation phase, the procedure may take an additional step of Biofeedback in which the decision made is communicated to the subject.

BCI system that implements the procedure: The execution of this procedure is carried out using a BCI system that comprises the following different electronic means (represented in Figures 1 and 2):

•

Subsystem 1, Main (1): Consists of an embedded or integrated control system with real-time response capacity, the purpose of which is to execute the main application and communicate with the rest of the subsystems. A series of functions are performed in this subsystem, among which are, and are not limited to:

o Generation of the necessary stimulation to the subject.

o Execution of the algorithms used for the extraction, processing and classification of data from the subject's brain activity.

o Presentation of data or useful indicators for the control, operation and maintenance of the system.

o Storage and / or transmission by appropriate technological means.

•

Signal Recording and Conditioning Subsystem 2 (G): Includes the means necessary to record and condition signals from the brain activity of the subject obtained by any method (for example, EEG, RMF, ECoG, NIRs, etc.). In particular, the subsystem comprises the following means: probes, amplifiers, filters and analog-to-digital converters.

•

Subsystem 3, Stimulus Player (C): Includes the means necessary to reproduce any of the defined auditory stimuli with the necessary precision and characteristics following the instructions of the Main Subsystem.

Description of the Training Phase:

As previously indicated, in order to achieve an effective application of the communication procedure in the subsequent phase of operation, it is necessary / recommended that the BCI system that implements this procedure be adapted or calibrated to the particular needs of the subject with whom the communication.

During the training or calibration phase, mutual learning between the subject and the system, as well as the latter's calibration, is carried out thanks to what is called "biofeedback". Biofeedback techniques allow a subject to become aware of internal physiological and biological changes of which he is not normally aware. The biofeedback applied to this invention consists in presenting the subject by means of auditory stimuli with their own brain activity, allowing them to search for the most appropriate strategies so that the differences between the objective brain activity and that achieved are minimal.

In this training or calibration phase, the response of the subject to evaluation tests carried out with different configurations of auditory stimuli is evaluated, understanding as the configuration of an auditory stimulus a specific combination of tone, timbre, repetitive presentation, repetition frequency, interval of presentation, volume or duration, as well as types of sound sources (for example natural speech and natural speech modulated by analogue or digital carriers).

Each configuration of the stimulus is capable of evoking a specific pattern of brain activity that is different from any other configuration.

During the training phase, the subject will be instructed, preferably verbally, to perform a certain mental task (for example, pay attention to a certain auditory stimulus and ignore others). As long as the subject, who is immersed in exceptional circumstances due to his situation, is able to collaborate, he will generate an objective pattern of brain activity specific and known for that mental task and the configuration of the stimulus used. At the end of the training phase, a statistical study will show if the registered brain activity and the expected brain activity are similar enough to consider that the patient has collaborated efficiently and therefore has completed the training phase.

That configuration of auditory stimuli whose response reaches the highest level of quality, understanding as such a better signal-to-noise ratio will be saved for later use during the operation phase. The signal is defined as the energy of that brain activity caused by auditory stimulation and the mental task necessary for its processing, while the noise will correspond to the energy of the rest of the brain activity.

After the training or calibration phase is complete, communication between the subject and the system can begin.

Description of the Operation Phase:

In this phase, each decision made by the subject will be associated with a trial (minimum action necessary for the system to detect brain activity associated with a decision), at the end of which the system decides which decision the subject wants. communicate, present it and offer a biofeedback to the subject about said decision. In this sense, biofeedback is a key element in learning by the subject, since it allows them to know to what degree they are able to use the system correctly, thus producing trial and error learning.

Once the auditory stimulus has been presented to the subject, the useful information for the functioning of the system, extracted from the patterns of brain activity, will form the so-called "characteristics vector", which will have, among others, the following components:

• Amplitudes, latencies, marks and time windows of the individual components of brain activity patterns. Coefficients obtained by the representation in the frequency domain of brain activity patterns, for example from the DFT (Discrete Fourier Transform) or the Wavelet transform and Estimated noise power.

Once the vector of characteristics has been extracted, a classification algorithm will be used so that the system decides what order or message the subject wants to communicate. Among the classification algorithms that can be used are:

• RNA (Artificial Neural Networks), Fuzzy Logic, SVM (Support Vector

5 Machine), "clustering" techniques, or statistical algorithms based on decision theory.

Once the auditory stimulus to which the subject paid attention was classified, the system

1 will transmit the code associated with said stimulus. For this there is a dictionary that assigns stimulus and codes and that the subject previously learned during the training phase. In the particular case that there is only a single stimulus, the subject must either attend to it or ignore it completely. In this case, the dictionary is very simple since the only two possibilities (attending or ignoring the stimulus) will be

15 associated with a binary type code (1, 0).

Finally, an auditory-type biofeedback is produced using the same electronic devices, so that the subject knows the code they have transmitted. In this way, the subject will be able to assess how he has to self-regulate internally to

20 pay attention to a single stimulus at the same time that the rest are ignored with the highest probability of success by the system when transmitting the desired code.

Among the applications that this procedure and the BCI system that implements it 25 offer are:

•

Two-way communication between the subject in a state of diminished consciousness and the outside world in general.

•

Communication between the subject and the rehabilitator to carry out rehabilitation techniques based on the use of the system.

30 • Communication between the subject and his family and friends, for example, the writing of the living will or last wills.

MODES OF REALIZATION Example 1.-Electronic system for synchronous communication with subjects in a state of diminished consciousness based on selective attention to two auditory stimuli consisting of modulated natural speech

Selective attention is an innate characteristic of the human being by which he is able to discriminate between different stimuli that are presented simultaneously simply by paying attention to the objective stimulus and ignoring the rest ("cocktail party" effect). This is a process of a cognitive nature that necessarily requires the will of the subject. Stimulation, under a series of restrictions, can generate a pattern of brain activity characteristic of each specific stimulus. That is, each auditory stimulus can be presented to the subject in such a way that the subject evokes a distinctive and independent pattern of brain activity for each stimulus. In this scenario, the selective attention that the subject would exert on a single stimulus of those present, would cause the amplification of the response energy evoked by the attended stimulus as well as the attenuation of the responses evoked by the rest of the ignored stimuli. This amplification of the evoked response that the subject provokes voluntarily through selective attention can be classified in a robust and reliable way, thus making communication possible.

The same process described in the previous paragraph can be particularized to the case of a single stimulus. In this case, it would not be selective attention but simply attention and the system would perform a binary classification on whether the subject was paying attention to the only stimulus or ignoring it in its entirety.

The system that will carry out the procedure described below comprises the following elements:

•

Subsystem 1: Embedded system (microprocessor based or similar with very high performance).

•

Subsystem 2: Consisting of EEG recording equipment (probes, amplifiers, filters, analog to digital converter system).

•

Subsystem 3: Device for the presentation of auditory stimuli and biofeedback, for example headphones.

•

Interface1-2: Based on wave transmission, whether guided or not.

•

Interface 1-3: Based on wave transmission, whether guided or not.

•

lnterfaz3-Subject: Based on sound waves

•

Subject-2 interface: Based on wave transmission, whether guided or not.

Procedure for the training and calibration phase: Figure 3 describes the algorithm for the training phase.

Firstly, it is verified that the subject has sufficient awareness to be able to complete the training and calibration procedure. This is done through a set of verbally given commands that ask the patient to modify their brain activity in a certain way (for example, by exerting selective attention or attention on one or more auditory stimuli respectively). Subsequently, a statistical study of correlation will be made between the brain activity that should have been generated and that actually generated by the subject, thus discriminating whether the subject was able to complete the orders given externally.

The explained process to discriminate if a subject is capable of completing externally given orders is carried out using a battery of tests. In each trial, it is excited by a different stimulus configuration capable of causing a specific brain activity pattern and distinct from any other stimulus configuration.

•

Different characteristics of tone, timbre, repetitive presentation, repetition frequency, volume, duration.

•

Different types of sound sources, for example natural speech and natural speech modulated by analog or digital carriers

o Natural speech stimulation will consist of one or more natural speech sound sources (depending on whether it is attention or selective attention respectively). Each source will be generated by registering an independent address from the same or different people.

o For the generation of modulated natural speech, the natural speech register is used. Said register will be multiplied, sample by sample, by an analog or digital carrier of a certain frequency and phase. An example would be AM modulations (amplitude modulated)

FM (frequency modulated) or both simultaneously, ASK (amplitude shift keying), FSK (frequency shift keying) or PSK (phase shift keying)

In any case, the stimulus or stimuli will always be auditory. Among all the 5 configurations, the one with the highest signal to noise ratio will be chosen to be used during the operation phase.

Described in more detail:

1o 1. The subject is indicated that they are going to proceed to present two aloutions, one for each ear

2. The cognitive task to be carried out is indicated: Pay attention to the left speech "(for example).

3. The subject is stimulated with a particular stimulus configuration:

•

Two independent addresses are presented by each of the ears. In the example, passages from chapters other than Don Quixote have been used.

•

Duration of 12 seconds each.

20 • The left address is modulated by a digital carrier at 50% load of 5Hz and offset of 0 °.

• The right speech is modulated in the same way but with a 180 ° offset.

4. During the test the brain activity of the subject is recorded with electrodes

25 located in positions Cz, Pz and Fpz of the International 10/20 system and reference the half-sum of the activity recorded in the left and right ear lobes.

5. At the end of the test, the DFT is computed on the EEG record, extracting the coefficients at 5Hz and the noise power at that frequency, estimated as

30 the average spectral power obtained in the lateral bands [4 .. 5) and (5 .. 6]. These values are stored in the characteristics vector.

6. Steps 1 to 5 are repeated 20 times. Of these, half the occasions the subject must attend to the left ear and in the other half attend to the right ear. The average characteristics vector of the 20 tests is obtained.

This vector of average characteristics corresponds to a particular stimulus configuration.

7.

Steps 1 to 6 are repeated for each of the 5 digital carrier frequencies (0.150r1Hz, (0.175r1Hz, (0.200r1Hz, (0.225r1Hz, (0.250r1Hz. Each of them corresponds to a different stimulus configuration.

8.

Therefore, the vector of average characteristics for each of the 5 stimulus configurations is obtained.

9.

It will be evaluated through a statistical study if the subject has been able to follow the process and the indications received. As the modulations of the left and right addresses are in contract (0 ° and 180 °), if the subject was not able to follow the procedure and did not attend to one of the sources, the effect is cancellative, being obtained at the spectral frequency of 5Hz a residual spectral power level similar to that of the adjacent bands (EEG noise). If the subject was able to follow the experiment, those tests that he attended to the left, a spectral power level at 5 Hz will be obtained with an exact lag of 1808 with respect to the records obtained when attention was paid to the right ear. By means of a simple hypothesis test or by correlation, it is possible to estimate whether the characteristic extracted throughout the tests corresponds to typical noise of the 5 Hz band or to 2 signals with 180 ° of lag. In the first case, it will be determined that the subject was unable to complete the training phase and therefore the operation phase will not proceed. In the second case, the stimulus configuration that obtained the best signal-to-noise ratio will be saved since this will be the configuration that will be used during the operation phase. The SNR signal to noise ratio is defined as SNR = I lc * conj (lc) -Dc * conj (Dc) 1 / N. Where

N: is the estimated noise power in the band around the digital carrier frequency. le: is the average DFT coefficient of the trials where the to the left address. conj (lc): is the conjugate of le. From: is the average DFT coefficient of the trials where the to the right address. conj (Dc): is the conjugate of De.

Procedure for the operation phase: Figure 4 describes the algorithm for the operation phase. For those subjects who have completed the training phase:

The subject voluntarily pays attention or selective attention (depending on whether it is a single stimulus or a set of them) to one of them. During the time that the subject exercises attention, the energy of the stimulus served is greater. The characteristics vector includes for each electrode the following components:

• Amplitudes and latencies, marks and time windows of the individual components of brain activity patterns. Coefficients obtained by the representation in the frequency domain of brain activity patterns, for example from the DFT (Discrete Fourier Transform) or the Wavelet transform. Estimated noise power.

To determine brain activity, the value of the PEAEE (Auditory Evoked Potential of Stable State) is collected, evoked by auditory stimulation.

Described in more detail:

one.

The stimulus configuration with the best signal to noise ratio (SNR) stored at the end of the training phase is loaded.

2.

A question is asked with a binary answer not known a priori, for example

"Are you in pain right now?"

3. The cognitive task to be performed is indicated: "If the answer is YES, listen to the speech from the left ear. If the answer is NO, listen to the speech from the right ear."

Four.

The subject is stimulated with a stimulation with the characteristics loaded in step 1.

5.

During the duration of the test, the brain activity of the subject is recorded with electrodes located in the Cz, Pz and Fpz positions of the International 10/20 system and reference the half-sum of the left and right ear lobes.

6.

The DFT is computed on the EEG register and the coefficient is extracted at the frequency of the digital carrier used as well as the noise power in the band of that frequency, estimated as the average spectral power

obtained in the lateral bands without taking into account the frequency of the carrier. Those values are saved in the feature vector.

7.

SNR is computed. If it does not exceed a certain value, for example 10 dB, the test is continued, without the stimulation ending.

8.

Once the SNR exceeds the established threshold (for example 1 O dB), the classification proceeds. For the classification, an optimal detector of minimum complexity will be used, based on a bank of correlation filters. The correlation will be made between the extracted characteristics vector and the average characteristics vector corresponding to the stimulus configuration used and stored during the training phase.

9.

The classifier makes the decision and proceeds to biofeedback the decision to the subject, for example "You have decided that you DO have pain at this time". At this time, the system would call a nurse to provide a pain reliever or, depending on how critical the action to be performed, could repeat another confirmation question with a binary answer, for example "You have decided that you DO have pain right now. Are you sure? ", repeating steps 2 to 9 again.

Example 2-Electronic system for synchronous communication with subjects in a state of diminished consciousness based on selective attention to two auditory stimuli of the AM-modulated tone type that generate PEAEE.

This execution example is exactly the same as the example with the following differences:

1. The stimulus setting will be

•

Pure tones of different frequencies are presented by each of the ears (100 Hz and 75 Hz for the left and right ear respectively).

•

Duration of 12 seconds each.

•

Each of the pure tones modulates a different carrier in AM (1Khz and 1.5 Khz respectively)

2. Since this stimulus configuration causes PEAEE, and due to the rectification effects of the cochlea (envelope detector or AM demodulator), the extraction of characteristics is done at the frequencies of the modulating tones (1 00 Hz and 75 Hz).

Example 3-Electronic system for communication with subjects in a state of diminished consciousness based on attention or selective attention to one or more auditory stimuli, respectively, of the "click" or tone type, both modulated or not.

10 This example of execution is exactly the same as example 1 with the following differences:

o The stimulation presented is not natural speech or modulated natural speech, but very short duration tones ("clicks") or pure tones,

15 both modulated or not by analog or digital carriers. In the particular assumption of pure tones in the 500 Hz to 5 Khz band modulated in AM or FM by carriers in the OHz band at 200 Hz, the evoked response would be composed of PEAEEs.

Claims (13)

1. Procedure for communication with subjects in a state of decreased consciousness by measuring parameters of brain activity generated after external auditory stimulation and classifying these parameters to determine the instruction that the subject wants to give, comprising the following phases:

to.

Selection of a set of auditory stimuli.

b.

Training phase, which includes the following steps:

i. A stimulus is selected and the subject is asked to discriminate it from

among all the stimuli you will receive.

ii. The complete set of auditory stimuli is emitted.

iii. Simultaneously with step iii, certain patterns are recorded

of the brain from which the parameters of interest are extracted

for the classification of the stimulus that are stored in the

called the feature vector.

iv. We associate this vector of characteristics with the discriminated stimulus.

v. Biofeedback is offered to the subject, communicating the

discrimination carried out.

saw. Repeat steps ii through v for the stimulus selected for

obtain a vector of average characteristics.

vii. Repeat step vi for each stimulus in the set.

viii. Elimination of the vectors of average characteristics in which

the statistical distribution of the sample of its components is

match a noise.

c.

Selection of the subset of stimuli and their feature vectors

partners not discarded in step b.vii

d.

Operation phase comprising the following steps:

ix. Emission of the subset of stimuli selected in phase c)

x. Simultaneously with step d.i, certain patterns are recorded

of the brain from which the parameters of interest are extracted

that will make up the vector of characteristics.

xi. The SNR ratio is calculated from the feature vector

obtained, during the emission of the subset of stimuli, up to

 ES 2 396 811 Al that reaches a value statistically related to a probability of success determined in the classification. xii. The vector obtained is classified according to the stored vectors, thus identifying the associated stimulus xiii. The instruction associated with the identified stimulus is executed. xiv. Biofeedback is offered to the subject, communicating the decision he has made.

2.

Procedure according to the previous claim, in which the parameters of brain activity measured in phases ii) and ix) are potentials evoked by the auditory stimulus.

3.

Procedure according to the previous claim in which the value of the PEAEE (Auditory Evoked Potential of Steady State) is measured, evoked by auditory stimulation.

Four.

Method according to any of the preceding claims, characterized in that two or more classification algorithms are executed and the one obtained by a simple majority is chosen as the classification of the characteristics vector.

5.

Procedure according to any of the preceding claims, in which the set of stimuli presented in the training phase consists of auditory stimuli modulated by a digital carrier at different frequencies.

6.

Procedure according to any of the preceding claims, in which only two auditory stimuli are presented.

7.

Procedure according to any of the preceding claims, characterized in that the auditory stimulation is of the natural speech type, preferably modulated by an analog or digital carrier.

8.

Procedure according to the previous claim, in which the auditory stimuli used are fragments of text.

9.

Procedure according to previous claim in which the auditory stimuli are presented with a load of 5Hz and there is a gap of 1800 between each ear

10.

Procedure according to claims 7 or 8 in which the configuration of the auditory stimuli used in the training phase is presented at the following digital carrier frequencies (0.150) -1 Hz, (0.175) -1 Hz, (0.200) 1 Hz, (0.225) -1 Hz, (0.250) -1 Hz.

ES 2 396 811 Al eleven . Procedure according to claims 7 or 8 in which the configurations of the stimuli presented in the training phase are:

•

Pure tones of different frequencies are presented by each of the ears (for example 100 Hz and 75 Hz for the left and right ear respectively).

•

Duration of 12 seconds each.

•

Each of the pure tones modulate in AM at 1 Khz and 1.5 Khz

12.

Method according to any of claims 1 to 6, characterized in that the auditory stimulation is of the "click" type, sound tones or sound tones modulated analogically or digitally.

13.

Electronic system for communication with subjects in a state of diminished consciousness based on the analysis of evoked brain activity and by external auditory stimulation and modulated by attention, comprising the following elements:

to. Main subsystem that allows you to run the application that performs the processing of a series of functions that includes the following stages: saw. Generation of the necessary stimulation to the subject. vii. Record of brain activity in general and of that evoked by stimulation in particular (PEA, PEAEE, exogenous and endogenous potentials Px, Nx where x is the time elapsed between the event that evoked the potential and the potential itself). Registration will be done by EEG.