WO2014072564A1

WO2014072564A1 - Neuromodulation-based cognitive training system and method

Info

Publication number: WO2014072564A1
Application number: PCT/ES2013/070781
Authority: WO
Inventors: Carlos ESCOLANO MARCO; Mónica AGUILAR HERRERO; María LOPEZ VALDÉS; Javier MINGUEZ ZAFRA
Original assignee: Bit&Brain Technologies, S.L.
Priority date: 2012-11-12
Filing date: 2013-11-11
Publication date: 2014-05-15
Also published as: ES2467515B1; ES2467515A1

Abstract

The invention relates to a cognitive training method and system that can be used to train a specific cognitive ability selected by a user (1) from a group of cognitive abilities, by performing a calibration customised for the user (1) and for the specific cognitive ability. The calibration comprises using sensors (2) to measure reference electroencephalograms (3) during the execution of a task associated with the selected cognitive ability, and calculating characteristics of interest and a neuronal activity reference level for said characteristics, the parameters being customised for the particular user and the particular cognitive ability.

Description

SYSTEM AND METHOD OF COGNITIVE TRAINING THROUGH

NEUROMODULATION

Object of the invention

The present invention relates to the field of cognitive skills training, and more specifically to a training method and system through the modulation of brain activity measured by electroencephalograms. Background of the invention

The field of skills training or cognitive skills using electronic media has undergone a strong expansion in recent years. Thanks to this expansion, its focus has ceased to be limited to medical treatments for patients with neurological disorders, and has reached healthy users who wish to stimulate their brain activity to improve their cognitive abilities.

Within this framework, a multitude of electronic devices and programs were developed that provide the user with a series of tasks or computerized games related to particular cognitive abilities such as attention or memory. However, although training with these devices allows users to improve their response to the proposed tasks and games, there is no evidence that their use provides general improvements when facing different problems, even when they are closely related to Same cognitive ability.

Against this approach, different neuromodulation systems have been proposed. Neuromodulation is a technique framed within the field of biofeedback, that is, systems that provide the user with real-time information on their physiological functioning. In particular, neuromodulation provides the user with functional information regarding their brain and their central nervous system. To do this, it obtains information about its brain activity from an electroencephalogram (EEG) or an MRI, analyzes said information, and transmits the result of said analysis to the user through an interface. The user can therefore use the information shown in the interface to train and optimize specific characteristics of their brain activity, managing to modulate them to operate at certain levels. The advantages of this type of systems have been demonstrated both for medical patients with neurological or psychological disorders (for example epilepsy, depression, attention deficit hyperactivity disorder, or addiction), as well as for healthy people who wish to improve their cognitive abilities.

In the case of the electroencephalogram, the measurement of brain activity is carried out through sensors placed on the surface of the scalp, with the possibility of varying the sensor technology, the position of the sensors on the scalp, and their clamping system. . Although the technique of electroencephalophy raffia does not allow an accurate functional interpretation of neuronal activity, it is possible to relate it to certain aspects of brain function, associating parts of the brain and characteristics of brain activity with different cognitive abilities (memory, attention , concentration, etc.).

The neuromodulation techniques therefore start from the signals measured through the sensors, and extract from them signals characteristic of interest (traditionally the activity of cerebral rhythms) for each particular system. The selection of the characteristics of interest can be done statistically, using prior information from a group of reference users, which implies a loss in the accuracy of the selection given the large variations of the characteristics between users and even between different times of use of the device for the same user. Alternatively, it is possible to select the characteristics of interest individually by user from reference measurements taken for each user who uses the system in a calibration phase prior to the training phase. Although this option is more expensive in time, greater precision in the selection is achieved and it allows individualizing the training to each user and to each particular moment of employment. Next, it is necessary to determine objective levels of work of the characteristics of interest. In some traditional systems, the parameterization of said target work levels is done manually, or is predefined by the system. This supposes an important limitation, since the parameters that characterize the levels of work are dependent on the subject, on the characteristics of interest and on the moment of use of the technology. To deal with this limitation, said parameterization can be carried out based on reference measurements taken for each user who uses the system during the calibration phase, obtaining an individualized parameterization for each user and at each particular moment of use.

The selection of the characteristics of interest and the parameterization of their target work levels have variations depending on the type of activity used as a reference during the calibration phase. While some systems, such as the one proposed in "Increasing individual upper alpha power by neurofeedback improves cognitive performance in human subjects" (Simón Hanslmayr et al, Applied Psychophysiology and Biofeedback, Vol. 30, No. 1, 2005) use as a calibration reference an electroencephalogram taken while the user is at rest, other alternatives, such as the one presented in "Neurofeedback training of the upper alpha frequency band in EEG improves cognitive performance" (B. Zoefel, Neurolmage 54, pages 1427-1431, 201 1), use as reference an electroencephalogram taken during the execution of a cognitive task proposed to the user. In either case, a single reference measure is used for calibration, so that the selection of the characteristics of interest can become poor and not accurately represent the user's brain activity, therefore decreasing or canceling the expected cognitive improvements Since the parameterization of the target work levels is based on the selected characteristics, an inappropriate selection of them is also reflected in the form of errors in the parameterization of the work level, and therefore, in the overall efficiency of the system .

Once the selection of the characteristics of interest and the parameterization of its target work levels has been completed, training begins with neuromodulation, that is, a phase in which the continuous measurement of level of work of these characteristics of interest, and is compared with the levels of objective work. An interface, for example graphic or sound, indicates to the user the relationship between the levels of work measured during the training phase, and the levels of objective work, thus guiding them towards the correct modulation of them. In particular, US 2004/0210156 A1 presents a neuromodulation system in which the level of work is presented to the user in real time by means of a color code. This level of work corresponds to a direct measurement of basic parameters such as the amplitude or frequency in the user's electroencephalogram, so that the system can present individualization problems for different users, as it does not characterize their brain activity properly.

One of the most relevant elements for the proper functioning of a neuromodulation device is the correction of artifacts. Artifacts are elements that appear superimposed on the measured signal, and that do not correspond to the brain's electrical activity but rather come from non-neural sources. In the case of the electroencephalogram, the most common artifacts come from one part of the eye or muscle movements, and the other from sensor failures and interference with other electronic devices. The elimination of these devices is, therefore, essential for the correct analysis and interpretation of the measured signals, both to identify the characteristics of interest and the levels of objective work, and to provide correct information in the feedback to the user. One of the most common artifact removal techniques is the application to the electroencephalogram signal of an activity threshold that, if exceeded, cancels the analysis of the part of the contaminated signal. However, this is a serious temporary limitation, since a large part of the information is discarded. In particular, the most common artifacts are those generated by eye movements (such as blinking), which can contaminate between 20% and 60% of a typical neuromodulation session. This limitation can be overcome by using other types of signal processing methods, such as a filter based on independent component analysis (ICA). ICA is a blind separation technique from sources that allow to break down a signal into additive components assuming their statistical independence. This technique applied to the electroencephalogram signal allows to discard those components that come from non-neural sources, and reconstruct the original signal eliminating the influence of said non-neural components.

US 2012/0100514 A1 and US 2012/0184870 A1 present two neuromodulation techniques that include artifact correction by ICA. These techniques make use of ICA to perform offline processing of the signal prior to training to determine the characteristics of interest, or after training to analyze the acquired signal, but do not use the filter in real time during training. They therefore have limitations in terms of the proportion of the training signal they allow to use. Additionally, the inventions and techniques mentioned throughout this section provide a general cognitive training, so they do not provide the user with the freedom to train their brain activity characteristics associated with particular cognitive abilities of their choice, such as , memory or attention.

There is therefore a need in the state of the art for a neuromodulation technique that is capable of providing individualized training to each user, with a data processing that optimizes the technique for the user, time of use of the technology, and the chosen cognitive ability. Finally, it should be noted that none of the inventions mentioned above allow to determine and quantify the improvements generated by the training.

Description of the invention

The present invention solves the problems described above by means of a cognitive training method and system through neuromodulation that allows the user to select the cognitive ability to train (for example: memory, attention, concentration, etc.), and that performs a treatment of signal individualized on electroencephalograms measured by sensors placed on the user's scalp, so that said treatment is completely individualized for the particular user, and time of use of the technology.

In a first aspect of the present invention a cognitive training system is presented by neuromodulation comprising:

- Sensors to measure electroencephalograms of the user during the time of use of the system.

- An interface for communication with the user, which allows users to receive control commands, as well as transmit information, for example regarding the use of the system. At the beginning of the session, the interface receives a command from said user indicating a training of at least one skill, preferably selected from a set of cognitive skills available for training shown to the user through the interface, thus allowing configurable training and particularized to the cognitive ability on which the user wishes to focus. Subsequently, the interface presents stimuli that represent the level of work of the characteristics of interest with respect to certain levels of objective work determined during a calibration phase.

- Signal processing methods adapted to analyze the electroencephalograms measured by the sensors, filter the artifacts, determine the characteristics of interest (related to the cognitive ability to train) and their current and objective levels of work.

In particular, once the user command is received, the sensors are adapted to measure electroencephalograms during the following phases:

- A calibration phase, in which the signal processing parameters of the system are determined from a set of calibration electroencephalograms. In particular, during this phase, the processing methods determine the characteristics of interest and their objective work levels, that is, determine a reference level with which subsequent electroencephalograms are compared.

The task of selecting the characteristics associated with cognitive ability is performed from, preferably, at least one electroencephalogram measured at rest and another during the performance of an active task or game related to the cognitive ability to train. The active task or game is designed so that through signal processing tools certain characteristics can be observed that mediate said ability, called characteristics of interest. More in detail, the execution of the active task allows to observe the change from rest to activation of these characteristics, obtained in the resting electroencephalogram and the electroencephalogram measured during the execution of the active task, in order to select them in an optimized way. It is thus achieved that the selection of the characteristics is individualized for the particular user, and that it is dependent on the cognitive ability and the moment of use of the technology.

Preferably, and in an equivalent manner, the parameterization of the target work levels on the characteristics of interest also uses one or more of the calibration electroencephalograms for the same reasons.

In the preferred case in which the user's command chooses the ability from among a set of available skills, each skill in the set has at least one associated task, so that all subsequent data processing, from which they are determined The characteristics of interest and the levels of work are fully individualized for the selected skill and the specific moment of operation. Preferably, the processing methods are adapted to calculate an artifact correction filter from one or more of the electroencephalograms measured during the calibration phase. Said artifact correction filter is, therefore, adapted to each user, and applying said filter in real time on the phase electroencephalograms Training allows you to eliminate the part of the signal not coming from neural sources and take advantage of a greater amount of information for training. The filter can also be applied to one of several of the calibration electroencephalograms to clean the signal prior to the determination of the characteristics of interest and the target work levels. Preferably, the artifact correction filter is generated by independent component analysis.

Preferably, the data processing to determine the characteristics of interest and their target work levels of the calibration encephalograms is performed on a subset of sensors. Said subset of sensors can be a predetermined subset, dependent on the selected cognitive ability, or be selected in this phase from among all available sensors.

- A training phase, in which a succession of electroencephalograms is recorded. Real-time processing methods eliminate electroencephalogram artifacts using the artifact correction filter to determine the current level of work of the characteristics of interest. The current work level is compared to the target work levels, and the result of such comparison is presented to the user in any modality of sensory stimulation through the interface. With this information, the user can learn to modulate the work levels to take them to more optimal levels, which will manifest in improvements in the chosen cognitive ability. Preferably, the data processing to calculate the work levels of the training encephalograms is performed on a subset of sensors, determined in the calibration phase.

- Preferably, the system elements are adapted to carry out a validation phase after the training phase, in which at least one electroencephalogram is measured which is compared by the processing methods with one or more of the electroencephalograms of the phase of calibration to quantify the improvement in cognitive ability achieved through the use of the system. More preferably, the validation phase It comprises measuring at least one electroencephalogram during a resting state, and an electroencephalogram during an accomplishment of the same task used during the calibration phase. Also preferably, the artifact correction filter is applied on said calibration and validation electroencephalograms. The quantification of the improvement is therefore carried out by comparing the variation of the electroencephalograms between the resting state and the active task, before and after training. In a second aspect of the present invention a method of cognitive training by neuromodulation is presented comprising the following steps:

- Receive a command from the user through which you select a training of at least one cognitive ability, preferably chosen from a set of cognitive skills available for your training shown to the user through the interface.

- Show through the interface at least one calibration task associated with the at least one skill whose training has been selected.

Preferably a task is performed at rest, and an active task or game related to the selected cognitive ability. These calibration tasks allow to observe in the electroencephalograms the characteristics of interest related to the chosen skill.

- Measure the calibration electroencephalograms during the execution of said calibration tasks.

- Preferably, calculate an artifact correction filter on one or more of the calibration electroencephalograms. More preferably, the filter is calculated by independent component analysis, and is subsequently applied to the calibration electroencephalograms to remove the artifacts before performing subsequent steps. Select characteristics or parameters of interest from the calibration electroencephalograms, preferably taken both at rest and during the execution of the active task or game. These characteristics of interest are obtained by means of a signal treatment that determines the differences between the state of rest and the active task, related to the selected cognitive ability. Preferably, the signal treatment comprises a time-frequency treatment of the electroencephalograms, and an analysis to characterize the differences between the result of the treatment on said electroencephalograms, said analysis being statistical in nature.

Parameterize the work levels of the characteristics of interest in the calibration electroencephalograms, and determine the target work levels for those characteristics of interest. The levels of objective work are determined in a particular way for each cognitive ability from an average level of work. For example, a cognitive ability may require increasing the level of work with respect to the average level, so that the levels of objective work is the set of values that is above the average level of work. On the contrary, another skill may require decreasing the level of work with respect to the average level, so that the levels of objective work is the set of values that is below the average level of work. In a preferred option, higher and lower levels of work are also determined that allow establishing a bounded scale during the training stage. The signal processing to determine the average, upper, and lower level of work comprises any set of mathematical operations applied to the level of work of each characteristic, or a subset of them, measured on one or more of the calibration electroencephalograms.

Measure a set of training electroencephalograms during a training phase. Preferably, the real-time artifact correction filter is applied to said training electroencephalograms, said filter having been previously calculated in the phase. Calibration

Calculate in real time on the set of training electroencephalograms, previously filtered the devices according to the previous step, the level of work of the characteristics of interest. The signal processing to determine the level of work of the characteristics of interest comprises any set of mathematical operations applied to the level of work of each characteristic, or a subset of them. Note that this mathematical treatment may be the same applied to determine the average (or higher, or lower) level of work in the calibration phase, or different. Preferably, the data processing to calculate the work levels of the training encephalograms is performed on a subset of sensors. Said subset of sensors is chosen in the calibration phase from among all available sensors.

Calculate in real time the feedback values that relate the work level of the characteristics of interest of the training electroencephalograms with respect to the target work levels. The target work levels are previously calculated in the calibration phase.

Present to the user in any form of sensory stimulation a representation of the calculated feedback values, allowing said user to know the status of the characteristics of interest and thus learn to modulate the level of work of said characteristics to reach objective work levels.

Preferably, after completion of the training, the method comprises a validation phase in which one or more validation electroencephalograms are measured, and the final work level is compared with the initial work level, obtained in the calibration phase. More preferably, the validation phase comprises measuring electroencephalograms in two stages, one at rest and one in which the user repeats the same task of the calibration phase. Also preferably, the method comprises applying the correction filter of artifacts on said calibration and validation encephalograms. The quantification of the improvement is therefore carried out by comparing the variation of the electroencephalograms between the resting state and the active task, before and after training.

Preferably, the step of selecting the characteristics of interest is carried out with a signal treatment that allows comparing electroencephalograms in which the user is at rest, with electroencephalograms in which the user is performing an active task or game. The combination of these tasks allows to observe the change from rest to activation of the brain (in which the cognitive ability intervenes) reflected in the characteristics of the signal in order to allow an optimized selection of these characteristics. This strategy implies a qualitative improvement both in the selection of the characteristics of interest and in the parameterization of their target work levels, given that both strategies are individualized by user and time of use of the device, and can be extended to any cognitive ability.

In a third aspect of the present invention there is presented a computer program comprising computer program code means adapted to implement the described method, when running on a computer, a digital signal processor, an application-specific integrated circuit , a microprocessor, a microcontroller or any other form of programmable hardware. The system, method, and computer program described, therefore provide a training tool through neuromodulation that adapts perfectly to each user and training situation, and that allows the training of particular cognitive skills, without requiring the presence of a professional doctor. This and other advantages of the invention will be apparent in light of the detailed description thereof.

Description of the figures

In order to help a better understanding of the characteristics of the In accordance with a preferred example of practical implementation thereof, and to complement this description, the following figures are attached as an integral part thereof, the character of which is illustrative and not limiting:

Figure 1 shows a diagram of the elements of the invention according to a preferred embodiment thereof, as well as the information transmitted between them and with a user of the system. Figure 2 presents the phases of an example training session in which the method and system of the invention is applied, as well as the different electroencephalograms measured during said phases.

Figure 3 shows a diagram of the steps performed by the processing means during the training phase according to a particular implementation of the method of the invention.

Preferred Embodiment of the Invention In this text, the term "comprises" and its derivations (such as "comprising", etc.) should not be construed in an exclusive sense, that is, these terms should not be construed as excluding the possibility that that is described and defined may include more elements, stages, etc. Note that the invention does not require the presence of medical personnel, being able to be fully controlled by the user, or directed by other people, without such persons requiring any type of medical training.

Note also that, throughout the text, the nomenclature "calibration electroencephalogram", "training electroencephalogram", and

"validation electroencephalogram" refers to the phase of use of the invention during which they are measured, in order to facilitate the understanding of the text. All these electroencephalograms are therefore measured in the same way, and represent the same type of information, measured during different moments of a typical session of use of the invention.

A diagram of a particular embodiment of the system of the invention is shown in Figure 1, which in turn implements a particular embodiment of the method of the invention. The figure also shows a user (1), who performs cognitive training using the invention. The system comprises sensors (2) that measure electroencephalograms (EEG) (3) of the user (1), and send the resulting electrical signal to processing means (4). During the training of cognitive skills, the processing means (4), following the steps described below, generate a feedback value

(5) that is transmitted to the user (1) through a stimulus (7) generated by an interface (6). In turn, the interface (6) receives commands (8) from the user (1).

The sensors (2) can be any sensor known in the state of the art capable of providing a signal relative to the user's electroencephalogram (3). In a particular implementation, the use of a raffia electroencephalog cap with 19 sensors, adapted to be arranged on the user's scalp according to the international 10/20 system, is contemplated. The sensors are distributed over the prefrontal, frontal, central, parietal and occipital areas. Specifically, positions FP1, FP2, F7, F3, Fz, F4, F8, T7, C3, Cz, C4, T8, P7, P3, Pz, P4, P8, 01, and 02 are used. Two sensors are also used. additional. The first additional sensor is placed in the prefrontal zone (FPz position) to record the signal corresponding to the user's potential ground. The second additional sensor is placed in an earlobe to record a signal against which the 19 sensors are measured, called the reference potential in the electronics field. The signal acquired by each sensor is filtered, amplified, and finally digitized.

As a result, the electroencephalogram of the user filtered in the frequency range of 0.5 Hz to 60 Hz is obtained, amplified and digitized at a sampling frequency of 256 samples per second. Obviously, other sensor arrangements, as well as other filtering ranges and sampling frequencies can be used within the scope of the invention as claimed.

The processing means (4) can be implemented in any programmable hardware device, whether in a computer, a digital signal processor, an application-specific integrated circuit, a microprocessor, a microcontroller, or any other. Consequently, the system can be both fixed and portable, and can be implemented, for example, in a telephone or other portable devices adapted to receive the signal from the sensors (2). Also, the interface (6) can be any type of interface capable of transmitting information to the user and receiving instructions from it. The transmission of information can therefore be visual, auditory, somatosensory or any combination of the foregoing. The reception of commands can be done through buttons, touch screens, voice recognition, or any other known means of interaction with a user.

In particular, one of the options to transmit to the user (1) the feedback values (5), is to establish a scale that assigns different colors to different feedback values (5), and display the corresponding color on an interface screen (6).

Prior to each training session, the system receives from the user (1) through the interface (6) a command (8) by which, the user (1) selects the cognitive ability that he wishes to train during said session. The cognitive ability is selected from a set of options provided by the system, each of the cognitive skills having one or more predetermined tasks or games, which allow characterizing the user's neuronal response to a problem that activates the characteristics involved in that cognitive ability Note that these tasks or games are used exclusively during the calibration phase explained below, and not during the training phase, in which the neuromodulation takes place. This allows "a la carte" learning, in which it is the user who decides what cognitive skills he wants to develop, instead of performing a generic and inefficient training. Figure 2 shows the different phases of an example of a session of use of the invention, once the cognitive ability to train during said session has been selected. The figure also shows the electroencephalograms (3) that are measured during each of said phases, and which allow the steps of the method performed by the processing means (4) to be executed. The session begins with a calibration phase (9) that allows to calculate the parameters required by the system. Next, the bulk of the session is occupied by the training phase (10), in which the user receives in real time stimuli (7) that represent the feedback values (5) and can therefore learn to modulate their activity cerebral. Finally, the process concludes with a validation phase (1 1) in which the user's progress (1) is evaluated.

The calibration phase (9) comprises the execution of two well differentiated tasks. First, a rest task (12), in which the user does not perform any activity related to the cognitive ability to train, but must remain for a few minutes in a state of rest with eyes closed. During this resting state, a resting calibration electroencephalogram (14 ') is measured. Next, it executes an active task (13), in which the user performs an active task in which the cognitive ability previously selected by the user intervenes, following the instructions provided through the interface (6). During this active task (13), an activation calibration electroencephalogram is measured (14). For example, in the case of the cognitive ability called working memory, the active task may consist of the continuous observation of an element in which the gradual changes of color in a visual interface must be counted (6), which allows to appreciate the difference between both electroencephalograms. In the particular case of working memory, this difference is mainly observed as a desynchronization of alpha brain rhythms in the parieto-occipital area of the scalp, considered key factors in working memory.

The activation calibration electroencephalogram (14) and the resting calibration electroencephalogram (14 ') are used to calibrate an artifact correction filter (18) that eliminates from the signal activity caused by non-neural sources such as eye movements and muscle of the user, as well as sensing errors and interference. Once the artifact correction filter (18) has been calibrated, the filter itself is applied to the calibration electroencephalograms before analysis. Note that in this way, a particular artifact correction is achieved for the specific user, for the selected cognitive ability and time of use of the technology. This correction of Artifacts are also used in real time to clear the signal during the training phase. In particular, a blind source separation technique is applied, such as independent component analysis (ICA). From the calibration electroencephalograms (14, 14 '), filtered by the artifact correction filter (18), the differences between the resting task (12) and the active task (13) are determined. From these differences, the characteristics or parameters of interest associated with the selected cognitive ability are determined, as well as the levels of objective work for said characteristics of interest. In particular, neuroscience studies relate increases in activity (or potency) in the upper part of the alpha band in parieto-occipital areas with improvements in working memory. Therefore, the characteristics of interest and their work levels are measured on the sensors placed in the parieto-occipital area of the scalp, more specifically in the sensors P3, Pz, P4, 01 and 02, called training sensors, on which are looking for the differences between the states of rest and activation.

In detail, for each training sensor, the power spectrum of the resting electroencephalograms and the active task, previously filtered from artifacts, is calculated and the frequency band related to maximum desynchronization is identified (between the resting electroencephalogram and the activated electroencephalogram) in a frequency range [f¡ _nf , f _sup ] = [5, 15] Hz. Next, for each training sensor the frequency band of interest is determined as the range [f _m , f _n ], where f _{m is} the maximum desynchronization frequency, and f _n the first value with synchronization less than a threshold, where f _n is greater than f _m , and f _m and f _{n are} contained within the interval [f¡ _nf , f _sup ]. In this particular implementation, the power in the frequency band [f _m , f _n ] for each training sensor are the characteristics of interest. The level of work of a characteristic of interest is its instantaneous power value. The target work levels are determined as the set of values that exceed an average work level, which is used as a reference during training. Said average work level is determined as the power in the frequency band [f _m , f _n ], calculated for the electroencephalogram of the active task and averaged for the training sensors. They are also calculated lower and upper limits of the work level so that said interval covers 95% of power distribution values in said band of interest in the electroencephalogram of the active task. Note that in the case of another cognitive ability, it is contemplated to use any other characteristic of interest that differentiates the activation calibration electroencephalograms (14) and the resting calibration electroencephalogram (14 '). Said characteristic of interest can be any parameter resulting from the comparison of the electroencephalograms in the two calibration states, whether in the time domain, frequency, or a combination of both domains.

Likewise, said parameters can be static, statistical, or consider temporal evolutions of the signal under analysis. Preferably, said parameters are individualized for each sensor used in the measurement of the electroencephalogram, although it is also contemplated to use an average measurement thereof, within the scope of the present invention.

The sensors on which the determination of the characteristics of interest is made, as well as the subsequent measurement of the training electroencephalograms (15), can be a predetermined subset, depending on the selected cognitive ability, or be selected as part of the own process of determining the characteristics of interest such as those sensors in which the greatest differences appear between the activation calibration electroencephalograms (14) and the resting calibration electroencephalogram (14 ').

Through this calibration process, it is achieved that the invention is individualized for each user and for each cognitive ability, as well as for each moment in which the training is carried out, including any variation produced by the state of the sensors or the sensor itself. Username.

Once the calibration phase (9) is completed, the training phase (10) is passed. The training phase (10) comprises a set of executions or trials, and the number and duration of said tests can be configurable by the user. A typical example would consist of 5 trials of 5 minutes each.

The scheme of the steps and elements of the training phase (10) implemented in the processing means (4) is shown in Figure 3. The objective of the training phase is to provide feedback values

(5) in real time from the training electroencephalograms (15), which represent the relationship between the current work level of the characteristics of interest of the training electroencephalograms (15) and the target work levels. For this part of some calibration parameters (17), obtained during the calibration phase (9). Said calibration parameters (17) comprise the parameters calculated for the artifact correction filter (18), the selected characteristics of interest (power in a given frequency band for each training sensor), and the target work levels (power of these characteristics averaged for the training sensors in the activation calibration electroencephalogram) that serve as a reference for the calculation of the feedback values (5). These levels of objective work also include an upper and lower limit to establish a bounded scale during training. During the training phase (10), the training electroencephalograms

(15) first pass through the artifact correction filter (18) to eliminate the artifacts from the signal. Next, the filtered signal reaches calculation means (19) that determine the level of work of the characteristics of interest of the electroencephalogram for each of the training sensors. These work levels are averaged for training sensors, and the resulting value serves as input to a comparator (20) that determines the feedback value (5) as the difference between that value and the average work level calculated during calibration . The joint use of the average work level, with the upper and lower limits, allows to obtain a standardized feedback value in the interval [-1, 1], so that positive values indicate target work levels, being more optimal the higher it is the value, while negative values indicate that the user is not at levels of objective work for their brain activity. The feedback values (5) are provided to the user (1) in real time, which allows him to learn to adequately modulate his brain activity, which leads to an improvement in the selected cognitive ability. These feedback values (5) are presented to the user (1) through stimuli (7) generated by the interface (6). In this specific case, the stimuli consist of a square drawn on the interface (6) that changes color according to the feedback value (5). In detail, feedback values in the interval [0, 1] (positive feedback) correspond to a color scale from gray to red, gradually increasing its saturation. Feedback values in the interval [0, -1] (negative feedback) correspond to a color scale from gray to blue, gradually increasing its saturation. Therefore, the objective of the user (1) is to ensure that the stimulus (7) is shown in red, and better saturation. Optionally, the system can provide the user (1), through the interface (6), with instructions on how to achieve this modulation.

Finally, the process ends with a validation phase (1 1), in which validation electroencephalograms (16, 16 ') are measured to determine the improvement of the cognitive ability experienced by the user. Preferably, a first validation electroencephalogram (16) is measured during a second iteration of the resting state (12 '); and a second validation electroencephalogram (16 ') during a second iteration of the same active task (13') used during the calibration phase (9). In this phase the processes of selection of the characteristics of interest and parameterization of their target work levels performed in the calibration phase (9) are repeated, and compared with the results obtained in said calibration phase (9), which allows quantifying the evolution and progress caused by training in the work levels of the characteristics of interest. The artifact correction filter is also applied to the validation electroencephalograms (16, 16 ').

In view of this description and figures, the person skilled in the art may understand that the invention has been described according to some preferred embodiments thereof, but that multiple variations can be introduced in said preferred embodiments, without departing from the object of the invention such and as claimed.

Claims

CLAIMS A user's cognitive training system (1) through neuromodulation comprising:

- sensors (2) adapted to measure at least one calibration electroencephalogram (14) during a calibration phase (9), and a plurality of training electroencephalograms (15) during a training phase (10), the training phase being (10) after the calibration phase (9);

- processing means (4) adapted to locate at least one characteristic of interest in the at least one calibration electroencephalogram (14) that characterizes a brain activity to be trained, determining a reference level from information of the at least one characteristic of interest measured for at least one calibration electroencephalogram (14), and calculate a relationship (5) between the reference level and the at least one characteristic of interest in the plurality of training electroencephalograms (15);

- and an interface (6) adapted to receive commands (8) from the user (1) and transmit information (7) to the user (1), said information (7) comprising the calculated ratio (5); characterized in that the processing means (4) are also adapted for, during the calibration phase (9):

- receive through the interface (6) a command (8) from the user (1) selecting a training of at least one cognitive ability;

- show through the interface (6) a task (13) to be executed by the user associated with the cognitive ability whose training is selected by the command (8); - and locate the at least one characteristic of interest associated with the cognitive ability whose training is selected by the command (8), and determine the reference level, from at least one calibration electroencephalogram (14) measured during the performance of said task (13).

2. System according to claim 1 characterized in that the command (8) of the user (1) that the processing means (4) are adapted to receive, is a command that selects a cognitive ability from a set of available cognitive skills; because each cognitive ability of the set has at least one associated task (13); and because the processing means (4) are adapted to locate the at least one characteristic of interest for each skill of the set through the task (13) associated with each skill of the set.

3. System according to any of the preceding claims characterized in that the processing means (4) are adapted to, during the calibration phase (9), locate the at least one characteristic of interest, and determine the reference level, from differences between at least a first calibration electroencephalogram (14) measured during the task (13) associated with the cognitive ability whose training is selected by the command (8), and at least a second calibration electroencephalogram (14 ' ) measured during a resting state (12).

4. System according to claim 3 characterized in that the at least one characteristic of interest that the processing means are adapted to locate is at least one static, dynamic, and / or statistical parameter that differentiates the first calibration electroencephalogram (14) and the Second calibration electroencephalogram (14 ') in a temporal, frequency, and / or time-frequency domain.

5. System according to any of claims 3 and 4 characterized in that the characteristics of interest that the processing means are adapted to locate are at least one power measured between initial and final frequencies of a frequency range.

6. System according to claim 5 characterized in that the frequency range in which the power is measured is a higher alpha band.

System according to any one of claims 3 to 6, characterized in that the at least one characteristic of interest is located individually for each sensor used in the measurement of the calibration electroencephalograms (14, 14 ').

System according to claim 7 characterized in that the at least one characteristic of interest in the training electroencephalograms (15) is measured for a subset of sensors, said subset of sensors being selected during the calibration phase (9).

9. System according to any of the preceding claims characterized in that the processing means (4) are adapted to:

- during the calibration phase (9), generating an artifact correction filter (18), said artifact correction filter (18) being generated based on at least one calibration electroencephalogram (14) measured during the calibration phase ;

- during the training phase (10), apply the generated artifact correction filter (18) to the plurality of training electroencephalograms (15).

10. System according to claim 9 characterized in that the processing means (4) are adapted to generate the artifact correction filter (18) by independent component analysis.

eleven . System according to any of claims 9 and 10 characterized in that the processing means (4) are adapted to apply the filter of correction of artifacts (18) on the at least one calibration electroencephalogram (14) prior to the location of the at least one characteristic of interest.

12. System according to any of the preceding claims characterized in that the processing means (4) are adapted to measure through the sensors (1) at least one validation electroencephalogram (16) during a validation phase (1 1), being the validation phase (1 1) after the training phase (10); and to calculate an evolution resulting from the use of the system by comparing at least one calibration electroencephalogram (14) and at least one validation electroencephalogram (16).

13. System according to claim 12 and any of claims 9 to 1, characterized in that the processing means (4) are adapted to apply the artifact correction filter (18) on the at least one validation electroencephalogram (16) .

14. System according to any of claims 12 to 13 and any of claims 3 to 1 1 characterized in that the processing means are adapted to measure at least one first validation electroencephalogram (16) measured during a second embodiment in the validation phase of the task (13 ') associated with cognitive ability, and at least a second validation electroencephalogram (16') measured during a second resting state (12 '), and to calculate the resulting evolution by comparing the first electroencephalogram of validation (16) and the second validation electroencephalogram (16 ') with the first calibration electroencephalogram (14) and the second calibration electroencephalogram (14').

15. Method of cognitive training of a user (1) by neuromodulation comprising:

- measure at least one calibration electroencephalogram (14) during a calibration phase (9), and a plurality of training electroencephalograms (15) during a training phase (10), the training phase (10) being subsequent to the calibration phase (9);

- locate at least one characteristic of interest that characterizes a brain activity to be trained, from at least one calibration electroencephalogram (14);

- determine a reference level from information of the at least one characteristic of interest measured for at least one calibration electroencephalogram (14);

- calculate a relationship (5) between the reference level and the at least one characteristic of interest in the plurality of training electroencephalograms (15);

- receive commands (8) from the user (1);

- and transmit information (7) to the user (1), said information comprising the calculated ratio (5); characterized in that the method further comprises, during the calibration phase (9):

- receive a command (8) from the user (1) selecting a training of at least one cognitive ability;

- show through the interface (6) a task (13) to be executed by the user associated with the cognitive ability whose training is selected in the command (8); and because the steps of locating the at least one characteristic of interest, and determining the reference level, are performed from at least one calibration electroencephalogram (14) measured during the performance of said task (13).

16. Method according to claim 15 characterized in that the user command (8) (1) is a command that selects a cognitive ability from a set of available cognitive skills; because each cognitive ability of the set has at least one associated task (13); and because the step of locating the at least one characteristic of interest in a manner is performed for each skill of the set through the task (13) associated with each skill of the set.

17. Method according to any of claims 15 and 16 characterized in that the steps of locating the at least one characteristic of interest, and determining the reference level, are performed by determining differences between at least a first calibration electroencephalogram (14) measured during the completion of the task (13) associated with the cognitive ability whose training is selected by the command (8); and at least a second calibration electroencephalogram (14 ') measured during a resting state (12).

18. Method according to claim 17 characterized in that the at least one characteristic is at least one static, dynamic, and / or statistical parameter that differentiates the first calibration electroencephalogram (14) and the second calibration electroencephalogram (14 ') in a domain temporal, frequency, and / or time-frequency.

19. Method according to claim 18 characterized in that the characteristics of interest comprise a power measured between initial and final frequencies of a frequency range.

20. Method according to claim 19 characterized in that the frequency range in which the power is measured is a higher alpha band.

twenty-one . Method according to any of claims 17 to 20, characterized in that the at least one characteristic of interest is located individually for each sensor used in the measurement of the calibration electroencephalograms (14, 14 ').

22. Method according to claim 21, characterized in that the at least one characteristic of interest in the training electroencephalograms (15) is measured for a subset of sensors, said subset of sensors being selected during the calibration phase (9).

23. Method according to any of claims 15 to 21 characterized in that it further comprises:

- during the calibration phase (9), generating an artifact correction filter (18), said artifact correction filter (18) being generated based on at least one calibration electroencephalogram (14) measured during the performance of the task (13) associated with the selected cognitive ability;

24. Method according to claim 23, characterized in that the artifact correction filter (18) is generated by independent component analysis.

25. Method according to any of claims 23 and 24 characterized in that it further comprises applying the artifact correction filter (18) on the at least one calibration electroencephalogram (14) prior to the location of the at least one characteristic of interest.

26. Method according to any of claims 15 to 25 characterized in that it further comprises measuring at least one validation electroencephalogram (16) during a validation phase (1 1), the validation phase (1 1) being subsequent to the training phase (10); and calculate a evolution resulting from the use of the system by comparing at least one calibration electroencephalogram (14) and at least one validation electroencephalogram (16).

27. Method according to claim 26 and any of claims 23 a

25, characterized in that it further comprises applying the artifact correction filter (18) on the at least one validation encephalogram (16).

28. System according to any of claims 26 to 27 and any one of claims 17 to 25 characterized in that it further comprises measuring at least one first validation electroencephalogram (16) measured during a second task completion (13 ' ) associated with cognitive ability, and at least a second validation electroencephalogram (16 ') measured during a second resting state (12'), and to calculate the resulting evolution by comparing the first validation electroencephalogram (16) and the second validation electroencephalogram (16 ') with the first calibration electroencephalogram (14) and the second calibration electroencephalogram (14').

29. Computer program comprising computer program code means adapted to perform the steps of the method according to any of claims 15 to 28, when said program is executed on a computer, a digital signal processor, a application-specific integrated circuit, a microprocessor, a microcontroller or any other form of hardware.