# US 9,436,910 B2 · Method, device and database for reconstructing intended activities from neural signals using frequencies of representations

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 9,436,910 B2
- Document ID: US9436910B2
- Status: Granted
- Kind: B2
- Application number: 14/451,478
- Filed: 2014-08-05
- Published: 2015-01-29
- Granted: 2016-09-06
- Earliest claimed: 2012-02-07
- Assignees: Cortec Corporation
- Applicants: Cortec GmbH
- Inventors: Jörn Rickert
- Web page: https://patentgrep.com/patent/US9436910B2

## CPC

- G06N 3/08: Learning methods
- A61B 5/165: {Evaluating the state of mind, e.g. depression, anxiety}
- G06F 16/22: Indexing; Data structures therefor; Storage structures
- H04N 21/42201: {biosensors, e.g. heat sensor for presence detection, EEG sensors or any limb activity sensors worn by the user (input arrangements for interaction with the human body based on nervous system activity detection G06F3/015)}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

There is provided a method for reconstructing intended activities from a first representation of neural signals which is indicative of an intended activity to a second representation, wherein for second representations, a degree of agreement between the first representation and each second representation from a plurality of predetermined second representations that are indicative of intended activities is determined on the basis of a predetermined agreement criterion, and a second representation of neural signals is selected from the plurality of second representations on the basis of the degree of agreement, which selected second representation is the reconstructed intended activity.

## Claims

### Claim 1

A method for reconstruction of intended activities from a first representation of neural signals which is indicative of an intended activity, wherein the first representation is generated using a predetermined method from the neural signals, for each second representation from a plurality of predetermined second representations, each of which being representative of a predetermined activity, wherein the plurality of predetermined second representation are generated from predetermined activities, a degree of correspondence is determined respectively between each of the respective second representations and the first representation according to a predetermined correspondence criterion, and depending on the degree of correspondence, a single second representation of the neural signals from the plurality of second representations is selected as the second representation representative of the intended activity, and the activity represented by the selected representation is reconstructed as the intended activity, wherein the determination of the degree of correspondence in accordance with the correspondence criterion comprises a determination of a total correspondence probability for each second representation from the plurality of second representations, wherein the total correspondence probability from a combination of first partial probabilities, which are created for each second representation from the plurality of second representations and wherein the total correspondence probability for each second representation indicates the probability with which the respective second representation of the first representation is assignable, and wherein forming the first partial probabilities comprises at least one of
determining a partial probability based on the frequency of the second representation within the plurality of predetermined second representations,
determining a partial probability based on semantic relationships of the second representation to already selected second representations, and
determining a partial probability based on a similarity of the second representation to the first representation.

### Claim 2 (depends on 1)

The method of claim 1, wherein the second representation with the greatest total correspondence probability selected.

### Claim 3 (depends on 1)

The method of claim 1, wherein each of the first partial probabilities is weighted by a weighting factor, wherein the weighting factor can be selected differently for each partial probability.

### Claim 4 (depends on 1)

The method of claim 1, wherein prior to the selection of the second representation from the plurality of second representations, it is verified whether the total correspondence probability of the second representation to be selected exceeds a predetermined threshold, and wherein when the predetermined threshold is not exceeded, the first representation the reconstructed intended activity.

### Claim 5 (depends on 1)

The method of claim 1, wherein on the basis of the respectively determined total correspondence probability, a number of second partial probabilities is determined, with which in turn a new total correspondence probability for each second representation from the plurality of second representations is determined.

### Claim 6 (depends on 1)

The method of claim 1, wherein the plurality of predetermined second representations is stored in a database.

### Claim 7

A method of for reconstruction of intended activities from a first representation of neural signals which is indicative of an intended activity, wherein the first representation is generated using a predetermined method from the neural signals, for each second representation from a plurality of predetermined second representations, each of which being representative of a predetermined activity, wherein the plurality of predetermined second representation are generated from predetermined activities, a degree of correspondence is determined respectively between each of the respective second representations and the first representation according to a predetermined correspondence criterion, and depending on the degree of correspondence, a single second representation of the neural signals from the plurality of second representations is selected as the second representation representative of the intended activity, and the activity represented by the selected representation is reconstructed as the intended activity, wherein plurality of predetermined second representations is stored in a database, and wherein for each second representation, a selection frequency, and/or for the second representations stored in the database, a number of selection sequences is stored in the database.

### Claim 8 (depends on 7)

The method of claim 7, wherein first representations are stored in the database, and wherein the stored first representations are correlated in the database to selected second representations, wherein first representations correlated to a second representation are used to determine the similarity of the first representations a second representation.

### Claim 9 (depends on 8)

The method of claim 8, wherein the second representation is replaced by the first representation which has been correlated to the second representation.

### Claim 10 (depends on 1)

The method of claim 1, wherein the first representations and the second representations comprise at least one out of movements, speech, emotions, pictures, smells, and tactile sense.

### Claim 11 (depends on 1)

The method of claim 1, wherein the signals are derived from a neural activity of the brain of a human being.

### Claim 12 (depends on 11)

The method of claim 11, wherein the signals are derived from a man-machine-interface arranged outside the human being.

### Claim 13 (depends on 1)

The method of claim 1, wherein prior to selecting a second representation, it is verified whether the degree of correspondence satisfies a predetermined criterion, and if the degree of correspondence does not satisfy the predetermined criterion, the intended activity is reconstructed from the first representation.

### Claim 14 (depends on 1)

The method of claim 1, wherein the predetermined method comprises applying a predetermined model, with which a neural brain activity is translated into the intended activity, wherein the total of intended activities forms a continuous parameter space.

### Claim 15 (depends on 1)

A database for use in a method for reconstructing intended activities according to claim 1, wherein the database is adapted to store second representations, which respectively are representative of a predetermined activity, and to select stored second representations according to a predetermined correspondence criterion and compare them to first representations, which are generated from a measured neural activity of the brain of a human being.

### Claim 16 (depends on 15)

The database of claim 15, wherein the database is further adapted to store first representations, which have been generated from a measured neural activity of the brain of a human being, and to link the first representations to the second representations stored in the database and/or replace second representations stored in the database by the first representation.

### Claim 17 (depends on 1)

The method of claim 1, wherein a tool comprising at least one out of a prosthesis, speakers and display means is controlled by a selected second representation.

### Claim 18 (depends on 1)

The method of claim 1, wherein each one of the representations respectively represents a sonogram.

### Claim 19 (depends on 1)

The method of claim 1, wherein each one of the representations respectively represents a trajectory of motions.

### Claim 20 (depends on 1)

The method of claim 1, wherein each one of the representations respectively is an image representation.

### Claim 21 (depends on 7)

The method of claim 7, wherein the first representations and the second representations comprise at least one out of movements, speech, emotions, pictures, smells, and tactile sense.

### Claim 22 (depends on 7)

The method of claim 7, wherein the signals are derived from a neural activity of the brain of a human being.

### Claim 23 (depends on 7)

The method of claim 7, wherein a tool comprising at least one out of a prosthesis, speakers and display means is controlled by a selected second representation.

### Claim 24 (depends on 7)

The method of claim 7, wherein each one of the representations respectively represents a trajectory of motions.

### Claim 25 (depends on 7)

The method of claim 7, wherein each one of the representations respectively is an image representation.