Interface device for recovery of motor skills

Abstract

The present invention concerns a motor skills recovery interface device (1 ), comprising: one or more cortical sensors (3; 31, 32,..., 3n), each capable of detecting a respective cortical signal (SE1, SE2,..., SEn) coming from a respective portion of the cerebral cortex (C) of a patient (P); a decoding unit (2), operatively connected to said one or more cortical sensors (3; 31, 32,..., 3n), configured to process each of said cortical signals (SE1, SE2,..., SEn) transmitted by each of called one or more cortical sensors (3; 31, 32,..., 3n); one or more muscle electro¬ stimulators (4), each applicable to a muscle to be stimulated (M); wherein said decoding unit (2) is configured to process one or more actuation signals (SM1, SM2,..., SMn) of said one or more muscle electro-stimulators (4; 41, 42, 4n) on the basis of the cortical signal received by said one or more cortical sensors (31, 32,..., 3n). The present invention also concerns a method of operation of an interface device (1).

Description

[0001] INTERFACE DEVICE FOR RECOVERY OF MOTOR SKILLS

[0002] *****

[0003] The present invention concerns a motor skills recovery interface device and its operation method.

[0004] Field of invention

[0005] More in detail, the invention concerns an interface device for the recovery of the ability to move in patients with injuries of the spinal cord or the peripheral nervous system of an individual.

[0006] In the following, the description will be aimed at the rehabilitation of lower limbs, but it is clear that it should not be considered limited to this specific use.

[0007] Prior art

[0008] As is well known, it is currently not possible to replace or regenerate the nervous system after an injury. This, as is also known, causes a serious motor disability for the patient who suffers this injury, with associated severe clinical, social, family, and economic implications.

[0009] In fact, the inability to replace or regenerate the nervous system after an injury also causes high costs for national healthcare systems. In fact, every year in Italy alone, between 250 and 500 thousand people suffer a spinal cord injury, with an incidence of 40-90 people per million. Males are twice as affected as females, and the most affected population groups are young males (20-29 years old), adolescent females (15-19 years old), and elderly males and females (over 60 years old), with an average age of 42 years. On the other hand, injuries of the peripheral nervous system affect 36.9 people per million inhabitants, if of traumatic origin, or involve between 1 % and 7% of the population, if of non-traumatic origin.

[0010] The most common causes that determine spinal cord injuries are trauma, linked to road accidents (38%), falls (30%), acts of violence (14%), sports injuries (9%).

[0011] Instead, the most frequent causes of peripheral nervous system injuries are related to trauma, compression, metabolic diseases, iatrogenic injuries, infections, autoimmune diseases, hereditary diseases, tumors, or idiopathic causes. Depending on the severity of the damage and its location, in the peripheral nervous system injuries the motor deficit can affect a single muscle or muscle groups; while in spinal cord injuries the most dramatic cases of paraplegia and tetraplegia occur, with the severity of damage directly linked to how close the injury is to the cervical spinal cord.

[0012] If the damage affects the spinal cord, in 59% of cases it results in tetraplegia, with a motor deficit of all four limbs, and in 41 % of cases in paraplegia, with a motor deficit therefore involving only the lower limbs. This results the inability of the affected person to move and the necessity of remaining in bed.

[0013] The severity of direct motor damage is also correlated with indirect damage linked to deep vein thrombosis, pressure ulcers, urinary tract infections, muscle spasms, chronic pain, osteoporosis, cardiorespiratory complications, and depression (the latter affecting up to 20-30% of cases).

[0014] In addition to clinical complications, the patient with a nerve injury also has his / her relational, social, and working life compromised, with an unemployment rate of approximately 90% in the first year and 60% in the long term in patients with spinal cord injury. This is also reflected in the daily life of family members and caregivers, who must offer constant support in personal care, domestic, and social activities.

[0015] For patients, only supporting measures are currently available, linked to hospital or home treatments of indirect clinical complications (with a rate of further hospitalization of 30% in the following years, and an average hospital stay of 22 days), physiotherapy programs, aimed at recovery attempts and infrastructural, and social facilitation measures. All of this generates a cost that, in the first year after the injury, is between $769,351 and $1 ,064,716 dollars for each tetraplegic patient, and around $518,194 dollars for each paraplegic patient. In subsequent years, on average, tetraplegia management costs between $113,423 and $184,891 per patient, and paraplegia costs $68,739 per patient. The total cost for a tetraplegic patient who is 25 years old at the time of injury is between $3,451 ,781 and $4,724,181 over his or her lifetime; while if the patient is 50 years old the total cost ranges from $2,123,154 to $2,596,329. Approximately, $2,310,104 and $1 ,516,052 are respectively the average total costs for a paraplegic patient who is 25 years old at the time of the spinal injury and for a patient who is 50 years old, over their entire lifetime.

[0016] To overcome these problems, dielectric elastomer actuators, which function as capacitors, have been proposed to replace damaged muscle tissues. In particular, acrylic or silicone elastomers with carbon nanotubes functioning as electrodes achieve mechanical performance similar to human muscle in vitro. However, these solutions present mechanical, electrical, and biological problems that have prevented their clinical application. Therefore, for the application of this technology, a substantial improvement in the actuation properties of the elastomers appears necessary.

[0017] According to another approach, devices are being developed to manage the condition described above and, therefore, improve the lives of patients. In particular, semi-invasive techniques, which use brain signals to activate an exoskeleton, are known.

[0018] These solutions are very functional and promising, but very inconvenient due to the management, costs, and size of the exoskeleton.

[0019] It is, therefore, clear that the sector feels the need to develop new solutions for patients who have suffered a serious trauma to the spinal cord or peripheral nervous system.

[0020] Scope of the invention

[0021] In light of the above, it is, therefore, aim of the present invention to propose a neural interface device between the brain and muscles, which allows the motor skills recovery quickly in patients with tetraplegia, paraplegia, or with serious disabilities linked to peripheral nervous system injuries. Another aim of the invention is to propose an interface device that is easy to handle and can be manufactured at low costs.

[0022] Object of the invention

[0023] It is, therefore, specific object of the present invention a motor skills recovery interface device, comprising: one or more cortical sensors, each capable of detecting a respective cortical signal coming from a respective portion of the cerebral cortex of a patient; a decoding unit, operatively connected to said one or more cortical sensors, configured to process each of said cortical signals transmitted by each of called one or more cortical sensors; one or more muscle electrostimulators, each applicable to a muscle to be stimulated; wherein said decoding unit is configured to process one or more actuation signals of said one or more muscle electro-stimulators on the basis of the cortical signal received by said one or more cortical sensors.

[0024] Always according to the invention, said cortical sensor may be of the electroencephalographic type.

[0025] Still according to the invention, said cortical sensor may be of the epidural electrocorticography (ECoG) type.

[0026] Advantageously according to the invention, each of said cortical sensors may comprise a wireless transmission module, for signal transmission, said decoding unit can comprise a transceiver module, configured to receive the signals transmitted by said wireless transmission modules of said cortical sensors, and to transmit the activation signals for said at least one muscle electro-stimulator, and central control unit, programmed to execute an open supervised deep learning algorithm for processing said electroencephalographic signals received from cortical sensors by means of said transceiver module, wherein said central control unit is made to produce specific stimuli for each of said one or more muscle electrostimulators to be activated.

[0027] Preferably according to the invention, each of said cortical sensors may comprise one or more electrodes, each designed to extract the signal from the cerebral cortex of a patient.

[0028] Furthermore, according to the invention, said cortical sensors may comprise an inductive antenna for power supply.

[0029] Advantageously according to the invention, said one or more muscle electro- stimulators may be of the Compex type and / or of the implantable type.

[0030] Always according to the invention, said decoding unit may comprise a central control unit, programmed to execute an open supervised deep learning algorithm for processing the signals received from said cortical sensors, wherein said central control unit is capable of producing a stimulus signal for said one or more muscle electro-stimulators.

[0031] Still according to the invention, said decoding unit may be configured to process one or more actuation signals of said one or more electro-muscle stimulators on the basis of the electroencephalographic signal received from said one or more cortical sensors by means of an multilinear switching Aksenova / Markov algorithm, so as to connect the cortical activation signal to the intramuscular functional electrical stimulator that can be associated with the muscles of a user which guarantee the volitional movement to which a specific cortical activation signal corresponds.

[0032] It is a further object of the present invention a method of operation of an interface device as defined above, comprising the following steps:

[0033] A. detecting a cortical signal of the electroencephalographic or epidural electrocorticography (ECoG) type by means of said one or more cortical sensors, so as to generate one or more corresponding cortical signals;

[0034] B. sending said one or more corresponding cortical signals by means of said wireless transmission modules of each of said cortical sensors;

[0035] C. receiving said one or more corresponding cortical signals from said decoding unit;

[0036] D. processing said one or more corresponding cortical signals by means of said decoding unit to obtain one or more actuation signals for said one or more muscle electro-stimulators;

[0037] E. transmitting said one or more actuation signals for said one or more muscle electro-stimulators;

[0038] F. carrying out an electrostimulation of the muscle on which are applicable by said one or more muscle electro-stimulators, corresponding to the cortical signals processed by said decoding unit.

[0039] Always according to the invention, said step D may be performed by said central control unit executing a machine learning algorithm.

[0040] Still according to the invention, said machine learning algorithm may be a multilinear switching Aksenova / Markov algorithm, suitable for connecting the cortical activation signal to the intramuscular functional electrical stimulator that can be associated with the muscles of a user, which guarantee the volitional movement to which a specific cortical activation signal corresponds.

[0041] It is a further object of the present invention a computer program comprising instructions which, when the program is executed by a computer, cause the computer to execute steps C, D, and E of the method described above.

[0042] It is a further object of the present invention a computer readable storage medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of the method described above.

[0043] Brief description of the figures

[0044] The present invention will be now described, for illustrative but not limitative purposes, according to its preferred embodiments, with particular reference to the figures of the enclosed drawings, wherein: figure 1 shows a block diagram of a motor skills recovery device, according to the present invention; figure 2 shows a schematic view of a sensor; figure 3 shows a schematic view of a decoding unit; and figure 4 shows the motor skills recovery device, according to the invention, applied to a subject.

[0045] Detailed description

[0046] In the various figures the similar parts will be indicated with the same numerical references.

[0047] Referring to figure 1 , the diagram of a motor skills recovery interface device according to the invention, hereinafter generically indicated with the numerical reference 1 , can be observed.

[0048] The interface device 1 is technically a neural interface that connects the motor cortex C to the muscular system M of patient P, exploiting the wireless technology and the artificial intelligence.

[0049] The interface device 1 essentially comprises a decoding unit 2, a plurality of cortical detectors / sensors 3, connected to said decoding unit 2, individually indicated with 31 , 32, ..., 3n, and at least one or more muscle electro-stimulators 4, individually indicated with 41 , 42, ... , 4n. The cortical sensors 31 , 32, ..., 3n are applicable to the cerebral cortex and are of the electroencephalographic (ECG) type, or, in the embodiment described, said cortical sensors 31 , 32, ..., 3n are external devices that detect a cortical signal transducing it into electroencephalographic signals SE1 , SE2, ..., SEn of electroencephalographic (EEG) type.

[0050] In cases where the detected signal is not sufficiently precise or "clean", internal cortical sensors ECoG, which must be implanted by means of a neurosurgical intervention, through an operculum of approximately 5 cm for each cerebral hemisphere can be used. In this case, using surgical navigation, the cortical sensor is inserted on the surface of the brain in an epidural location. The cortical sensor ECoG is patch-shaped with 64 micro-electrodes, generally located in both cerebral hemispheres, in the primary motor cortical area (indicated in the literature as M1 ).

[0051] Said cortical sensors 3 are made of biocompatible silicone and platinum, suitable for a long-term implantation. Their function is to detect the activation of the cerebral cortex C through epidural electrocorticography (ECoG), detecting and obtaining the corresponding electrical signals SE1 , SE2, ..., SEn.

[0052] Each of said cortical sensors 3 also includes a wireless transmission module. In particular, with reference to figure 2, a cortical electroencephalographic sensor 31 (or also of the ECoG type) can be observed, which have a plurality of electrodes 311 , each intended to be in contact, as mentioned above, with the cerebral cortex, and a wireless transmission module 312, which can be of different types, such as Bluetooth®, Wi-Fi, or other types of protocols, possibly appropriately encrypted, in order to limit any interference from external signals.

[0053] In particular, the electronic components of the cortical activation sensor are housed in a titanium casing (50 mm diameter, 7-12 mm thick, and a convex external face). An array of 64 platinum-iridium (90:10) recording electrodes for epidural ECoG (2 mm diameter, 4-4.5 mm pitch) and five reference electrodes are positioned on the flat internal face of the device. ECoG data are recorded thanks to an integrated circuit specific for the application, which allows multi-channel amplification and digitization with an input noise of less than 0.7 pV root mean square in the range from 0.5 Hz to 300 Hz .

[0054] The data are transmitted via a very high frequency antenna (402-405 MHz). Power supply is remotely provided via a 13.56 MHz inductive high-frequency antenna. Both antennas are embedded in a silicone flap that extends into the subcutaneous space. In order to guarantee the stability of the high-frequency signal (586 Hz), given the limited bandwidth, 32 out of 64 contacts are used for each system. The signal SE1 detected, for example, by the detected cortical sensor 31 is wirelessly sent from each of said one or more cortical sensors 3 via the respective transmission modules 312, to the external decoding unit 2.

[0055] The external decoding unit 2 will use an open supervised deep learning mechanism to process the data extracted from the cortical activation sensor: using the multilinear switching Aksenova / Markov algorithm, the extracted data will be decoded. The patient will be asked to think about performing specific movements of the lower limb, and, in particular, movements linked to the activation of the gluteal muscles, femoral quadriceps, and plantar flexor. This model will allow the cortical activation signal to be connected to the intramuscular functional electrical stimulator implanted in the muscles that guarantee the volitional movement corresponding to that specific cortical activation signal.

[0056] The algorithm mentioned above consists of a hidden Markov model classifier (HMM) defined as "gate" and a set of independent regression models, the "experts". The "gate" will define the intention to activate the specific muscle or muscle chain, while the "expert" will define the width of the contraction. The multilinear switching model will predict the state and assume that the sequence of states follows a first- order Markov chain hypothesis. Therefore, the probability of a state at any time will depend on the combination of the previous state and the cortical signal just acquired.

[0057] The multilinear switching Aksenova / Markov algorithm is therefore an algorithm for the analysis of high-dimensional data. This method offers an effective means of reducing the dimensionality of data while maintaining crucial information.

[0058] Specifically, the algorithm is based on the idea of passing (or “switching”) between different linear representations of the data. This “switching” occurs in such a way as to maximize the information retained during the dimensionality reduction.

[0059] For example, if you have a data set with many variables, the algorithm might choose to represent some of these variables in one way and others in another, based on how to maximize the information retained. This "switching" between different representations is carried out iteratively, until an optimal reduced representation of the data is reached. The processing of the central control unit 21 is carried out to produce a specific stimulus for the muscle electro-stimulator 4 to be activated (e.g., if the activation of the motor cortical area specific for the left rectus femoris is detected, the decoding unit 2 sends the activation signal only to the electro-stimulator of left rectus femoris).

[0060] Said decoding unit 2 also includes a transceiver module 22, configured to receive the signals SE1 , SE2, ..., SEn transmitted by said cortical sensors 3 and to transmit the activation signals for the at least one muscle electro-stimulator 4.

[0061] Also in this case, the transceiver module 22 can be of different types, according to the need, such as Bluetooth®, Wi-Fi, Infrared or other types of protocols.

[0062] The decoding unit can be a generic computer, suitably programmed, such as a PC, a tablet, or a specific device, having, for example, a suitably programmed FPGA.

[0063] The muscle electro-stimulators 4 are connected to said decoding unit 2 in wireless mode.

[0064] Each muscle electro-stimulator 41 , 42, ..., 4n is a peripheral device that is applied to the skin overlying the muscles M to be activated (see also figure 4). In cases of difficult transmission of the contraction impulse (such as in cases of obese patients with significant subcutaneous adipose tissue), the muscle electro-stimulator 4 can be surgically implanted on the muscle belly.

[0065] The signal transmitted from said decoding unit 2 to the electro-stimulators 4 (or to one of them) determines the contraction of the corresponding muscle, thus bypassing the spinal cord and the peripheral nervous system.

[0066] Specifically, the external electro-stimulator is of the Compex-type wireless electro-stimulator type. Instead, what can be implanted if necessary is composed of the following features: an implantable wireless control unit and an interface with the electrodes. It is surgically implanted in the muscle belly in correspondence with the neuromotor plate, which is identified through a special probe. The electrodes receive neural signals coming from the cortical activation sensor and these are reprocessed by the artificial intelligence system. This determines the activation of the electrostimulator and the creation of an electric current that activates the specific muscle or muscle chain corresponding to the intended volitional movement, producing muscle contraction and restoring movement.

[0067] In other words, the interface device 1 is a brain-muscle neural interface that bypasses the damaged nervous system through a wireless signal, connecting the patient's motor cortex to his own muscles, ensuring the recovery of the ability to move.

[0068] The operation of the interface device 1 described above takes place as follows.

[0069] The cortical sensors 31 , 32, ... , 3n detect an electroencephalographic (EEG) signal SE1 , SE2, ... , SEn. Each signal SE1 , SE2, ..., SEn detected by each cortical sensor 31 , 32, ..., 3n is wirelessly sent via the transmission modules 311 , 312, ..., 31 n to the external decoding unit 2. The central control unit 21 , through the open supervised deep learning algorithm, processes the signals SE1 , SE2, ..., SEn, so as to produce a specific stimulus for the electro-stimulator 4 to be activated. In the central control unit 21 , and in particular the deep learning algorithm, is previously trained on the signals of the patient P.

[0070] Subsequently, the corresponding stimulation signal SE1 , SE2, ..., SEn is transmitted and reproduced by one more muscle electro-stimulators 41 , 42, ..., 4n to activate the muscle M on which it is applied and stimulate the appropriate movement corresponding to the detected cortical stimulus.

[0071] In the interface device 1 , even in the presence of a injury of the spinal cord or of the peripheral nervous system, the muscle contraction is guaranteed through said one or more muscle electro-stimulators 4. The activation of each muscle electro-stimulator 41 , 42, ..., 4n is obtained by the wireless communication with a corresponding cortical sensor 31 , 32, ..., 3n coupled to the cerebral cortex C, which detects the activity of the part of the cerebral cortex C responsible for controlling those specific muscles M. For example, the contraction of the patient's right biceps brachii would be guaranteed by muscle electro-stimulators 4 wirelessly activated by cortical sensors 3, which have detected the stimulus to the contraction of that muscle M in the motor cortex C.

[0072] In particular, said decoding unit 2 sends processed actuation signals SM1 , SM2, ..., SMn, for each of said muscle stimulators 41 , 42, ..., 4n.

[0073] On the basis of the above, the interface device 1 allows to detect the signals coming from the cortex C, to transform them, or "translate" them into electrostimulatory signals for actuation of the corresponding muscle M.

[0074] The central control unit 21 of the decoding unit 2 then "interprets" the electroencephalographic or epidural electrocorticography signals.

[0075] Advantages

[0076] The advantages of the application of the wireless brain-muscle interface realized by the interface device are articulated on different levels: clinical, psychosocial, family, and economic.

[0077] Clinically, a patient with a motor deficit linked to a damage of the spinal cord or peripheral nervous system would recover the ability to move in the body areas affected by the nerve injury. Furthermore, the resolution of the state of immobility would allow the prevention of the development of indirect clinical damage linked to deep vein thrombosis, pressure ulcers, urinary tract infections, muscle spasms, chronic pain, osteoporosis, cardiorespiratory complications, and depression.

[0078] From a psychosocial point of view, the patient would have the possibility of recovering a social and working life that would otherwise be limited by immobility or the impossibility of carrying out some movements. Furthermore, the so-called caregivers and family members would have the possibility of regaining space and time in their lives, otherwise aimed at supporting the sick person.

[0079] A further advantage of the solution according to the present invention is given by the elimination of hospital and home management costs of the sick person, which would guarantee important savings for national healthcare systems.

[0080] The present invention has been described for illustrative but not limitative purposes, according to its preferred embodiments, but it is to be understood that modifications and / or changes can be introduced by those skilled in the art without departing from the relevant scope as defined in the enclosed claims.

Claims

CLAIMS1. Motor skills recovery interface device (1 ), comprising: one or more cortical sensors (3; 31 , 32, ..., 3n), each capable of detecting a respective cortical signal (SE1 , SE2, ..., SEn) coming from a respective portion of the cerebral cortex (C) of a patient (P); a decoding unit (2), operatively connected to said one or more cortical sensors (3; 31 , 32, ..., 3n), configured to process each of said cortical signals (SE1 , SE2, ..., SEn) transmitted by each of called one or more cortical sensors (3; 31 , 32, ... , 3n); one or more muscle electro-stimulators (4), each applicable to a muscle to be stimulated (M); wherein said decoding unit (2) is configured to process one or more actuation signals (SM1 , SM2, ..., SMn) of said one or more muscle electro-stimulators (4; 41 , 42, 4n) on the basis of the cortical signal received by said one or more cortical sensors (31 , 32, ..., 3n).

2. Interface device (1 ) according to the preceding claim, characterized in that said cortical sensor (3; 31 , 32, ..., 3n) is of the electroencephalographic type.

3. Interface device (1 ) according to any one of the preceding claims, characterized in that said cortical sensor (3; 31 , 32, ..., 3n) is of the epidural electrocorticography (ECoG) type.

4. Interface device (1 ) according to any one of the preceding claims, characterized in that each of said cortical sensors (3; 31 , 32, ..., 3n) comprises a wireless transmission module (312), for signal transmission, in that said decoding unit comprises a transceiver module (22), configured to receive the signals transmitted by said wireless transmission modules (312) of said cortical sensors (3), and to transmit the activation signals for said at least one muscle electro-stimulator (4), and a central control unit (21 ), programmed to execute an open superviseddeep learning algorithm for processing said electroencephalographic signals received from cortical sensors (3) by means of said transceiver module (22), wherein said central control unit (21 ) is made to produce specific stimuli for each of said one or more muscle electro-stimulators (4) to be activated.

5. Interface device (1 ) according to any one of the preceding claims, characterized in that each of said cortical sensors (3; 31 , 32, ..., 3n) comprises one or more electrodes (311 ), each designed to extract the signal from the cerebral cortex (C) of a patient (P).

6. Interface device (1 ) according to any one of the preceding claims, characterized in that said cortical sensors (3; 31 , 32, ..., 3n) comprise an inductive antenna for power supply.

7. Interface device (1 ) according to any one of the preceding claims, characterized in that said one or more muscle electro-stimulators (4) are of the Compex type and / or of the implantable type.

8. Interface device (1 ) according to any one of the preceding claims, characterized in that said decoding unit (2) comprises a central control unit (21 ), programmed to execute an open supervised deep learning algorithm for processing the signals received from said cortical sensors (3; 31 , 32, ... , 3n), wherein said central control unit (21 ) is capable of producing a stimulus signal for said one or more muscle electro-stimulators (4).

9. Interface device (1 ) according to the preceding claim, characterized in that said decoding unit (2) is configured to process one or more actuation signals (SM1 , SM2, ..., SMn) of said one or more electro-muscle stimulators (4; 41 , 42, .., 4n) on the basis of the electroencephalographic signal received from said one or more cortical sensors (31 , 32, ..., 3n) by means of an multilinear switching Aksenova / Markov algorithm, so as to connect the cortical activation signal to the intramuscular functional electrical stimulator that can be associated with the muscles of a user which guarantee the volitional movement to which a specific cortical activation signal corresponds.

10. Method of operation of an interface device (1 ) according to any one of the preceding claims, comprising the following steps:A. detecting a cortical signal of the electroencephalographic or epidural electrocorticography (ECoG) type by means of said one or more cortical sensors (3; 31 , 32, ..., 3n), so as to generate one or more corresponding cortical signals (SE1 , SE2, ... , SEn);B. sending said one or more corresponding cortical signals (SE1 , SE2, ..., SEn) by means of said wireless transmission modules (312) of each of said cortical sensors (3; 31 , 32, ..., 3n);C. receiving said one or more corresponding cortical signals (SE1 , SE2, ..., SEn) from said decoding unit (2);D. processing said one or more corresponding cortical signals (SE1 , SE2, ..., SEn) by means of said decoding unit (2) to obtain one or more actuation signals (SM1 , SM2, ..., SMn) for said one or more electrodes - muscle stimulators (4; 41 , 42, ... , 4n);E. transmitting said one or more actuation signals (SM1 , SM2, ..., SMn) for said one or more muscle electro-stimulators (4; 41 , 42, ..., 4n);F. carrying out an electrostimulation of the muscle (M) on which are applicable by said one or more muscle electro-stimulators (4; 41 , 42, ..., 4n), corresponding to the cortical signals (SE1 , SE2, ..., SEn) processed by said decoding unit (2).

11. Method according to the preceding claim, when dependent on claim 4, characterized in that said step D is performed by said central control unit (21 ) executing a machine learning algorithm.

12. Method according to the preceding claim, characterized in that said machine learning algorithm is a multilinear switching Aksenova / Markov algorithm, suitable for connecting the cortical activation signal to the intramuscular functional electrical stimulator that can be associated with the muscles of a user, which guarantee the volitional movement to which a specific cortical activation signal corresponds.

13. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to execute steps C, D, and E of the method according to any one of claims 10 and 11 .

14. A computer readable storage medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of the method according to any one of claims 10 and 11 .

Images