# US 2004/0204769 A1 · Decoding algorithm for neuronal reponses

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

- Number: US 2004/0204769 A1
- Document ID: US20040204769A1
- Status: Published application
- Kind: A1
- Application number: 10/250,543
- Filed: 2004-04-26
- Published: 2004-10-14
- Earliest claimed: 2001-01-12
- Assignees: The United States of America as represented by the Department of Health and Human Services
- Inventors: Barry J. Richmond; Matthew Wiener
- Web page: https://patentgrep.com/patent/US20040204769A1

## CPC

- A61F 2/72: Bioelectric control, e.g. myoelectric
- A61F: FILTERS IMPLANTABLE INTO BLOOD VESSELS; PROSTHESES; DEVICES PROVIDING PATENCY TO, OR PREVENTING COLLAPSING OF, TUBULAR STRUCTURES OF THE BODY, e.g. STENTS; ORTHOPAEDIC, NURSING OR CONTRACEPTIVE DEVICES; FOMENTATION; TREATMENT OR PROTECTION OF EYES OR EARS; BANDAGES, DRESSINGS OR ABSORBENT PADS; FIRST-AID KITS (dental prosthetics A61C)

## Abstract

A device and method for decoding neuronal responses wherein sequences of potentials from neurons are monitored while specific motor tasks are carried out, and these sequences are characterized using order statistics and subsequently the order statistics are used to decode action potentials representing unidentified motor tasks to determine the desired motor task. The method of the invention comprises the steps of monitoring action potentials caused by a motor task being requested by the brain, calculating a spike density function and order tasks for each distinct motor task, to relate action potentials to their specific motor task. The invention also offers methods of formulating instructions for a prosthetic device. This method comprises the steps of learning the neuronal responses of distinct motor tasks by monitoring action potentials caused by a motor task being requested by the brain, calculating a cumulative density function for each distinct motor task, and using order statistics to relate action potentials to their respective motor tasks; monitoring action potentials from at least one neuron of said user wherein the action potentials are caused by the request for an unknown motor task; using said learned neuronal responses to determine which motor task is being requested by the monitored neuron; and formulating instructions on how to carry out the requested motor task. The device of the invention comprises a prosthetic limb, a device capable of making said prosthetic limb carry out motor tasks, a device capable of recording action potentials from neurons, and a device containing instructions for monitoring neurons, calculating cumulative density functions, utilizing order statistics, and determining instructions for various motor tasks.

## Claims

### Claim 1

A method of decoding neuronal responses comprising the steps of:
a. recording action potentials from at least one neuron over time, wherein said action potentials are monitored while specific known motor tasks are being requested by the neuron;
b. using order statistics to relate action potentials to respective unknown motor tasks being requested by the neuron given said recorded data.

### Claim 2

A method of formulating instructions for a prosthetic device comprising the steps of:
a. learning the neuronal responses of a user comprising the steps of recording action potentials from at least one neuron over time, wherein said action potentials are elicited by a predetermined motor task being requested by a neuron, using order statistics and said recorded action potentials to relate other action potentials to their respective unknown motor tasks being requested by the neuron;
b. recording data from action potentials from at least one neuron of said user, wherein the action potentials are elicited by requests for unknown motor tasks;
c. using said learned neuronal responses to determine which motor task is being requested by the monitored neuron, wherein the requested motor task is determined using order statistics; and
d. interpolating among the motor tasks for which the model has been calibrated, by adding the calibrated motions together weighted by the probabilities with which each elicited the action potentials;
e. formulating instructions for conducting the requested motor task.

### Claim 3

A prosthetic device comprising:
a. a prosthetic limb;
b. a device capable of making said prosthetic limb carry out motor tasks;
c. a device capable of recording action potentials from neurons;
d. a device containing instructions for monitoring neurons, utilizing order statistics, and determining instructions for various motor tasks.

### Claim 4 (depends on 3)

The prosthetic device of claim 3 wherein the device containing instructions comprises an electronic media storage device.

### Claim 5 (depends on 3)

The device of claim 3 wherein the device containing instructions comprises an electronic media storage device and a computer processor.

### Claim 6 (depends on 3)

The device of claim 3 wherein the device capable of recording action potentials from neurons is a microelectrode.

### Claim 7 (depends on 6)

The device of claim 6 wherein the microelectrode is a tungsten microelectrode.

### Claim 8 (depends on 6)

The device of claim 6 wherein the microelectrode records in a range of from about 0.8 to 1.3 mega ohms.

### Claim 9

A method of manipulating a prosthesis using neural responses, said method comprising the steps of:
a. recording a spike train, said spike train comprising neural action potentials over time;
b. determining the spike density function by determining the frequency of action potentials per millisecond from the occurrence of the initial stimuli
c. statistically describing the distribution of the action potentials in said spike train; and
d. repeating said step of statistically describing the distribution of the action potentials for each newly observed spike train.

### Claim 10 (depends on 9)

The method of claim 9, wherein the spike density function is represented by a curve derived from a histogram.

### Claim 11 (depends on 10)

The method of claim 10, wherein the curve is determined by a truncated Gaussian approximation.

### Claim 12 (depends on 9)

The method of claim 9, wherein the action potentials in said spike train are statistically described by
12 h n , k ⁡ ( t ) = ( n k ) ⁢ F ⁡ ( t ) k - 1 ⁢ f ⁡ ( t ) ⁡ [ 1 - F ⁡ ( t ) ] n - k .

### Claim 13 (depends on 9)

The method of claim 9, wherein said spike train was caused by a predetermined stimuli.

### Claim 14 (depends on 9)

The method of claim 9, further comprising determining the stimulus that caused said newly observed spike train.

### Claim 15 (depends on 14)

The method of claim 14, wherein said stimulus that caused said newly observed spike train is determined based on the statistical distribution of the action potentials in the spike train caused by the predetermined stimuli.

### Claim 16 (depends on 15)

The method of claim 15, further comprising the step of instructing a prosthesis to carry out a motion associated with said stimulus that caused said newly observed spike train.

### Claim 17

A method of manipulating a prosthesis using neural responses, said method comprising the steps of:
a. recording a spike train, said spike train comprising neural action potentials over time;
b. determining the spike density function by modeling said spike train as a doubly stochastic Poisson process;
c. statistically describing the distribution of the action potentials in said spike train; and
d. repeating said step of statistically describing the distribution of the action potentials for each newly observed spike train.

### Claim 18 (depends on 17)

The method of claim 17, wherein said spike density function
p(t)=Σwipi(t),
13 wherein ⁢ ⁢ p i ⁡ ( t ) = exp ⁡ ( - μ i ⁢ ∫ t 0 t ⁢ ⁢ ⅆ s ⁢ ⁢ λ ⁡ ( s ) ) ⁡ [ 1 - exp ⁡ ( - μ i ⁢ ∫ t 0 t 0 + δ ⁢ ⁢ t ⁢ ⁢ ⅆ s ⁢ ⁢ λ ⁡ ( s ) ) ] .

### Claim 19

A computer-based method of manipulating a prosthesis using neural responses, said method comprising the steps of
a. recording a spike train, said spike train comprising neural action potentials over time;
b. determining the spike density function by determining the frequency of action potentials per millisecond from the occurrence of the initial stimuli
c. statistically describing the distribution of the action potentials in said spike train; and
d. repeating said step of statistically describing the distribution of the action potentials for each newly observed spike train.

### Claim 20

A computer data product readable by a computing system and encoding instructions for manipulating a prosthesis using neural responses, said method comprising the steps of
a. recording a spike train, said spike train comprising neural action potentials over time;
b. determining the spike density function by determining the frequency of action potentials per millisecond from the occurrence of the initial stimuli
c. statistically describing the distribution of the action potentials in said spike train; and
d. repeating said step of statistically describing the distribution of the action potentials for each newly observed spike train.

### Claim 21 (depends on 20)

The computer data product according to claim 20, wherein the computer data product comprises a computer readable storage medium.
