Cognitive state machine for prosthetic systems

• US 6,952,687 B2
• Filed: 07/10/2002
• Issued: 10/04/2005
• Est. Priority Date: 07/10/2001
• Status: Expired due to Fees

• Current Assignee: California Institute of Technology
• Original Assignee: California Institute of Technology
• Inventors: Shenoy, Krishna V., Burdick, Joel W., Andersen, Richard A., Pesaran, Bijan, Cao, Shiyan, Mitra, Partha, Meeker, Daniella
• Primary Examiner(s): Davis, George B.
• Application Number:
  US10/193,649
• Publication Number:
  US 20030023319A1
• Time in Patent Office: 1,182 Days
• Field of Search: 706/12, 706/47, 623/24, 623/25
• US Class Current: 706/12
• CPC Class Codes: A61F 2/72 Bioelectric control, e.g. myoelectric
• CPC Class Codes: A61F 2002/704 computer-controlled, e.g. robotic control
• CPC Class Codes: A61F 2002/705 Electromagnetic data transfer for implantable prostheses A61F2250/0002
• CPC Class Codes: G06F 3/015 Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection

Abstract

A prosthetic system may use a decoder to predict an intended action, such as a reach, from processed signals generated from measured neural activity. The decoder may included a cognitive state machine, which transitions between cognitive states based on transition rules.

Claims

1. A method comprising:

receiving a processed signal generated from measured neural activity;

transitioning between states in a state sequence in a cognitive state machine in response to said processed signal; and

generating a command signal indicative of an intended movement corresponding to said state sequence.

2. The method of claim 1, wherein the neural activity comprises single unit activity.

3. The method of claim 1, wherein the neural activity comprises local field potential activity.

4. The method of claim 1, further comprising:

classifying the processed signal into discretely valued states corresponding to a cognitive state.

5. The method of claim 1, wherein the neural activity comprises a single unit activity and a local field potential activity.

6. The method of claim 1, further comprising:

classifying the processed signal into continuously valued states corresponding to a cognitive state.

7. The method of claim 1, wherein the states correspond to cognitive states that are observable.

8. The method of claim 1, wherein the states correspond to a cognitive state that are hidden.

9. The method of claim 1, wherein said transitioning comprises transitioning between states according to transition rules.

10. The method of claim 1, wherein said transitioning comprises transitioning between states defined by transition rules described by a transition matrix that determines probabilities.

11. The method of claim 1, wherein said generating comprises generating the command signal from a state sequence according to the optimization of a cost function.

12. The method of claim 1, wherein said generating comprises generating the command signal from a state sequence according to the optimization of a cost function that maximizes expected probability over all sequences.

13. The method of claim 1, wherein said generating comprises generating the command signal from a state sequence according to the optimization of a cost function that minimizes risk over all sequences.

14. The method of claim 1, further comprising:

classifying the processed signal into a period classification.

15. The method of claim 1, further comprising:

classifying the processed signal into a direction classification indicative of an intended direction for said movement.

16. The method of claim 1, wherein said transitioning comprises transitioning from a planning state to a movement state.

17. The method of claim 1, wherein said intended movement comprises a reach.

18. The method of claim 1, wherein said transitioning comprises transitioning between states in response to receiving a classification corresponding to a first state in the sequence for a predetermined time period.

19. The method of claim 1, wherein said transitioning comprises transitioning from a planning state to a movement state in response to receiving a planning classification for a predetermined time period.

20. The method of claim 1, wherein said transitioning comprises transitioning from a planning state to a movement state in response to receiving a planning classification and a direction classification for a predetermined time period.

21. The method of claim 1, wherein said transitioning comprises transitioning from a planning state to a movement state in response to receiving a planning classification for a predetermined time period and a cognitive go signal.

22. A method comprising:

observing neural signals from a subject in a period in which the subject forms an intention to perform a movement;

recording a plurality of observations;

determining a plurality of underlying cognitive states in response to said observations; and

generating a cognitive state transition model including said cognitive states.

23. The method of claim 22, wherein said generating comprises optimizing a mathematical model including said observations using a cost function.

24. The method of claim 22, wherein said determining comprises generating a mathematical model including said observations.

25. The method of claim 22, wherein said observing comprises observing spectra generated from electrical activity measured in a subject'"'"'s brain.

26. The method of claim 25, wherein said electrical activity comprises spike train activity.

27. The method of claim 25, wherein said electrical activity comprises local field potential activity.

28. The method of claim 25, wherein said observations comprise spectra.

29. The method of claim 25, wherein said observations comprise functions of spectra.

30. Apparatus comprising:

a state machine operative to transition between a plurality of cognitive states in response to signals generated from neural activity, said cognitive states including a termination state; and

a signal generator operative to generate a control signal indicative of an intended movement in response to the state machine transitioning through a state sequence.

31. The apparatus of claim 30, wherein said neural activity comprises single unit activity.

32. The apparatus of claim 30, wherein said neural activity comprises local field potential activity.

33. The apparatus of claim 30, wherein said cognitive states comprise a baseline state and a planning state.

34. The apparatus of claim 30, wherein the cognitive state machine is further operative to transition from the planning state to the movement state in response to receiving signals indicative of a planning period for a predetermined period of time.

35. The apparatus of claim 30, wherein the cognitive state machine is further operative to transition from the planning state to the movement state in response to receiving signals indicative of a planning period and a direction for a predetermined period of time.

36. The apparatus of claim 30, wherein the cognitive state machine is further operative to transition from the planning state to the movement state in response to receiving signals indicative of a planning period for a predetermined period of time and receiving a cognitive go signal.

37. The apparatus of claim 30, further comprising a classifier operative to classify said signals into one of a plurality of classifications.

38. The apparatus of claim 37, wherein the classifier is further operative to select a classification of highest probability from a plurality of classifications.

39. The apparatus of claim 37, wherein said classifications comprise classifications corresponding to cognitive states.

40. The apparatus of claim 37, wherein said classifications comprise classifications corresponding to directions.

41. The apparatus of claim 30, wherein said intended movement comprises a reach.

42. A system comprising:

an implant operative to measure neural activity;

a signal processor operative to transform said measured neural activity into a processed signal;

a decoder operative to predict an intended movement from said processed signal, the decoder including

a state machine operative to transition between a plurality of cognitive states in response to the processed signal, said cognitive states including a termination state;

a signal generator operative to generate a control signal indicative of the intended movement in response to the state machine transitioning through a state sequence; and

a controller operative to control a device to perform a movement corresponding to-the intended movement in response to said control signal.

43. The system of claim 42, wherein the neural activity comprises electrical activity in a subject'"'"'s brain.

44. The system of claim 42, wherein the device comprises a robotic limb.

45. The system of claim 42, wherein the intended movement comprises an intended reach.

46. An article comprising a machine-readable medium including machine-executable instructions, the instructions operative to cause a machine to:

receive a processed signal generated from measured neural activity;

transition between states in a state sequence in a cognitive state machine in response to said processed signal; and

generate a command signal indicative of an intended movement corresponding to said state sequence.

47. The article of claim 46, wherein the neural activity comprises electrical activity in a subject'"'"'s brain.

48. The article of claim 46, wherein said transitioning comprises transitioning from a planning state to a movement state.

49. An article comprising a machine-readable medium including machine-executable instructions, the instructions operative to cause a machine to:

observe neural signals from a subject in a period in which the subject forms an intention to perform a movement and performs the movement;

record a plurality of observations;

determine a plurality of underlying cognitive states in response to said observations;

generate a cognitive state transition model including said cognitive states; and

determine the control signal corresponding to a said state sequence.

50. The article of claim 49, wherein the instructions operative to cause the machine to observe comprise instructions operative to cause the machine to observe spectra generated from electrical activity measured in a subject'"'"'s brain.