Granted Patent B2

US 12,449,901 · App. 18/674,675 · Granted Oct 21, 2025

Inventors: Krishna Vaughn Shenoy; Jaimie M. Henderson (Redwood City, CA); Francis Robert Willett (Menlo Park, CA)

Assignee: The Board of Trustees of the Leland Stanford Junior University

G06F3/015G06F18/2148G06F18/41G06N7/01G06F2218/00G06N3/08

Abstract

Claims (44)

1. A brain computer-interface for symbol decoding, comprising:

a neural signal recorder implanted into a brain of a user; and

a symbol decoder, the symbol decoder comprising:

a processor; and

a memory, where the memory stores a symbol decoding application capable of directing the processor to:

obtain neural signal data from the neural signal recorder;

temporally bin the neural signal data to create a plurality of neural population time series;

convert a portion of the plurality of neural population time series covering a predetermined time window into a time probability series using a symbol decoding model;

identify a most likely symbol intended by the user from the time probability series as the most likely symbol in the time probability series offset by a delay from a time point in the time probability series at which the probability for a new character exceeds a threshold; and

perform a command associated with the most likely symbol.

2. The brain computer-interface for symbol decoding of claim 1 , wherein the neural signal recorder is a microelectrode array comprising a plurality of electrodes.

3. The brain computer-interface for symbol decoding of claim 2 , wherein the neural signal data describes spikes of neurons in proximity to respective electrodes in the plurality of electrodes.

4. The brain computer-interface for symbol decoding of claim 1 , further comprising an output device selected from the group consisting of: vocalizers, displays, prosthetics, and computer systems.

5. The brain computer-interface for symbol decoding of claim 1 , wherein the symbol model is selected from the group consisting of: recurrent neural networks (RNNs), long short-term memory (LSTM) networks, temporal convolutional networks, and hidden Markov models (HMMs).

6. The brain computer-interface for symbol decoding of claim 1 , wherein the memory further comprises a symbol database comprising:

a plurality of symbols; and

a plurality of commands;

wherein each symbol in the plurality of symbols is associated with a command.

7. The symbol decoding system for brain-computer interfacing of claim 6 , wherein:

the plurality of symbols comprise letters of an alphabet; and

each letter of the alphabet is associated with a command to print the letter to a text string.

8. The brain computer-interface for symbol decoding of claim 7 , wherein the symbols for each letter in the alphabet are difference maximized.

9. A method for decoding symbols from neural activity, comprising:

obtaining neural signal data from a neural signal recorder implanted in a head of a user and configured to record neural signals from a brain of the user;

temporally binning the neural signal data to create a plurality of neural population time series;

converting a portion of the plurality of neural population time series covering a predetermined time window into a time probability series using a symbol decoding model;

identifying a most likely symbol intended by the user from the time probability series as the most likely symbol in the time probability series offset by a delay from a time point in the time probability series at which the probability for a new character exceeds a threshold; and

performing a command associated with the most likely symbol.

10. The method for decoding symbols from neural activity of claim 9 , wherein the neural signal recorder is a microelectrode array comprising a plurality of electrodes.

11. The method for decoding symbols from neural activity of claim 10 , wherein the neural signal data describes spikes of neurons in proximity to respective electrodes in the plurality of electrodes.

12. The method for decoding symbols from neural activity of claim 9 , further comprising performing the command using at least one output device selected from the group consisting of: vocalizers, displays, prosthetics, and computer systems.

13. The method for decoding symbols from neural activity of claim 9 , wherein the symbol model is selected from the group consisting of: recurrent neural networks (RNNs), long short-term memory (LSTM) networks, temporal convolutional networks, and hidden Markov models (HMMs).

14. The method for decoding symbols from neural activity of claim 9 , wherein the symbol and the command are stored in a symbol database comprising:

a plurality of symbols; and

a plurality of commands;

wherein each symbol in the plurality of symbols is associated with a command.

15. The method for decoding symbols from neural activity of claim 14 , wherein:

the plurality of symbols comprise letters of an alphabet; and

each letter of the alphabet is associated with a command to print the letter to a text string.

16. The method for decoding symbols from neural activity of claim 15 , wherein the symbols for each letter in the alphabet are difference maximized.

17. The brain computer-interface for symbol decoding of claim 1 , wherein each bin is 100 ms, and the window is 1 second.

18. The brain computer-interface for symbol decoding of claim 1 , wherein the delay is tuned to the user.

19. The method for decoding symbols from neural activity of claim 9 , wherein each bin is 100 ms, and the window is 1 second.

20. The method for decoding symbols from neural activity of claim 9 , wherein the delay is tuned to the user.

Assignments (3)

Continuity (5)

References Cited (249)