Systems and methods for estimating surface electromyography

1. A biofeedback device adapted for using the activity of one or more muscles to proportionally control a second electromechanical device, the biofeedback device comprising a first detector configured for detecting a first signal that indicates a time series of surface electromyography measurements, designated sEMG(t), for each time t of a plurality of times from a particular location on a subject, a processor configured for processing the first signal using a nonlinear filter for determining a time series of a stable driving signal x(t), a power source, and a feedback system, the feedback system comprising a motorized device, wherein the feedback system is in electrical communication with the processor, wherein

processing the first signal using the nonlinear filter further comprises applying a nonlinear filter that includes a first parameter α that indicates a rate of drift for values of x(t) and an independent second parameter β that indicates frequency of shift in values of x(t); and

the feedback system causes the motorized device to apply a sensory feedback to the subject based on the stable driving function.

2. A biofeedback device as recited in claim 1 , wherein the feedback system applies sensory feedback in a vicinity of the particular location.

3. A biofeedback device as recited in claim 2 , wherein the feedback system applies a vibration as the sensory feedback.

4. A biofeedback device as recited in claim 3 , wherein the first detector is an sEMG electrode and the motorized device comprises a substantively quiet, small vibrating motor adhered to a box that encases the sEMG electrode.

5. A biofeedback device as recited in claim 4 , wherein:

the biofeedback device is portable; and

the feedback system applies the vibration on a constant basis compared to a therapy setting.

6. A biofeedback device as recited in claim 5 , wherein the biofeedback device excludes audio beeping and a visual signal.

7. A biofeedback device as recited in claim 3 , wherein the biofeedback device is configured for:

attaching to the subject that has a condition selected from a group comprising cerebral palsy, brain damage resulting from head trauma, brain damage resulting from oxygen deprivation, paraplegic loss of all or part of a limb, spinal cord injury, amyotrophic lateral sclerosis, stroke, AIDS, sarcopenia, disuse atrophy, glucocorticoid induced muscle loss, prolonged spaceflight, chronic neck pain, back pain, and muscle spasms, muscle tension, tension-type headache and athletic training; and

operating the motorized device to effectively treat the condition.

8. A biofeedback device as recited in claim 2 , wherein determining the stable driving signal x(t) is performed in real time.

9. A biofeedback device as recited in claim 2 , wherein to apply the sensory feedback to the subject based on the stable driving function further comprises driving a motorized prosthetic device attached to the subject in the vicinity of the particular location.

10. A biofeedback device as recited in claim 1 , wherein to apply the sensory feedback to the subject based on the stable driving function further comprises operating a device based on the stable driving function, wherein the subject is able to sense the operation of the device.

11. A biofeedback device as recited in claim 1 , wherein applying the nonlinear filter further comprises applying a recursive method comprising:

determining a measurement model, designated P(emglx), that relates the probability of a rectified sEMG observation, designated emg, given a value of the stable driving signal x;

determining a probability density function of values for the stable driving signal x at each time ti, designated p(x,ti), based on previous measurements sEMG(ti) at previous times ti<t;

propagating the probability density p(x,t−Δt) forward in time an amount Δt to time t to provide an a priori probability density function for the stable driving signal x at time t, designated p(x,t−);

determining a posteriori probability density function of a value of x at time t, designated P(x,t), based on P(emglx) and p(x, t−); and

determining an output value x(t) substantively equal to maximum a posteriori (MAP) value for x that has the highest a posteriori probability, designated argmax(P(x,t)).

12. A biofeedback device as recited in claim 11 , wherein determining the probability density function of values for the stable driving signal x at initial time t 0 before a first measurement of sEMG(t) further comprises determining a constant probability density for an expected range of values of x.

13. A biofeedback device as recited in claim 11 , wherein propagating the probability density p(x,t−Δt) forward in time further comprises:

determining a value ε for bin width of a histogram that represents the probability density function of values for the stable driving signal x; and

evaluating an equation given by

p ( x,t −) =α p ( x−ε,t−Δt ) +(1−2α) p ( x,t−Δt ) +α p ( x+ε,t−Δt ) +β+(1−β) p ( x,t−Δt ).

14. A biofeedback device as recited in claim 11 , wherein determining the measurement model further comprises selecting the measurement model from a group comprising a Poisson measurement model and a half-Gaussian measurement model and an exponential model.

15. A biofeedback device as recited in claim 11 , wherein applying the recursive method further comprises determining p(x,t) by dividing P(x,t) by a constant C, such that ∫ p(x,t) dx =1.

16. A method comprising:

obtaining a time series of surface electromyography measurements sEMG(t) for each time t of a plurality of times from a particular location on a subject;

determining on a processor a time series of a stable driving signal x(t) based on the measurements sEMG(t), further comprising applying a nonlinear filter that includes a first parameter α that indicates a rate of drift for values of x(t) and an independent second parameter β that indicates frequency of shift in values of x(t); and

applying a sensory feedback to the subject based on the stable driving function.

17. A method as recited in claim 16 , wherein applying the sensory feedback to the subject based on the stable driving function further comprises applying the sensory feedback in a vicinity of the particular location.

18. A method as recited in claim 17 , wherein applying the sensory feedback to the subject based on the stable driving function further comprises applying a vibration as the sensory feedback.

19. A method as recited in claim 18 , wherein applying the vibration as the sensory feedback further comprises driving a substantively quiet, small vibrating motor adhered to a box that encases a sEMG electrode.

20. A method as recited in claim 19 , wherein:

a portable system comprises the sEMG electrode, the vibrating motor, a processor configured for determining the stable driving signal x(t), and a battery for powering the vibrating motor and processor; and

applying the vibration as sensory feedback further comprises applying the vibration as sensory feedback by the portable system for a prolonged time compared to a therapy setting.

21. A method as recited in claim 20 , wherein the portable system excludes audio beeping and a visual signal.

22. A method as recited in claim 18 , wherein:

the method further comprises selecting the subject, wherein the subject is determined to have a condition selected from a group comprising cerebral palsy, brain damage resulting from head trauma, brain damage resulting from oxygen deprivation, paraplegic loss of all or part of a limb, spinal cord injury, amyotrophic lateral sclerosis, stroke, AIDS, sarcopenia, disuse atrophy, glucocorticoid induced muscle loss, prolonged spaceflight, chronic neck pain, back pain, and muscle spasms, muscle tension, tension-type headache and athletic training; and

applying the vibration further comprises applying the vibration to effectively treat the condition.

23. A method as recited in claim 17 , wherein determining the stable driving signal x(t) is performed in real time.

24. A method as recited in claim 17 , wherein applying the sensory feedback to the subject based on the stable driving function further comprises driving a motorized prosthetic device attached to the subject in the vicinity of the particular location.

25. A method as recited in claim 16 , wherein applying the sensory feedback to the subject based on the stable driving function further comprises operating a device based on the stable driving function, wherein the subject is able to sense the operation of the device.

26. A method as recited in claim 16 , wherein applying the nonlinear filter further comprises applying a recursive method comprising:

determining a measurement model, designated P(emg|x), that relates the probability of a rectified sEMG observation, designated emg, given a value of the stable driving signal x;

determining a probability density function of values for the stable driving signal x at each time ti, designated p(x,ti), based on previous measurements sEMG(ti) at previous times ti<t;

propagating the probability density p(x,t−Δt) forward in time an amount Δt to time t to provide an a priori probability density function for the stable driving signal x at time t, designated p(x,t−);

determining a posteriori probability density function of a value of x at time t, designated P(x,t), based on P(emglx) and p(x, t−); and

determining an output value x(t) substantively equal to maximum a posteriori (MAP) value for x that has the highest a posteriori probability, designated argmax(P(x,t)).

27. A method as recited in claim 26 , wherein determining the probability density function of values for the stable driving signal x at initial time t 0 before a first measurement of sEMG(t) further comprises determining a constant probability density for an expected range of values of x.

28. A method as recited in claim 26 , wherein propagating the probability density p(x,t−Δt) forward in time further comprises:

determining a value ε for bin width of a histogram that represents the probability density function of values for the stable driving signal x; and

evaluating an equation given by

p ( x,t −) =α p ( x−ε, t−Δt ) +(1−2α) p ( x,t−Δt ) +α p ( x+ε,t−Δt ) +β+(1−β) p ( x,t−Δt ).

29. A method as recited in claim 26 , wherein determining the measurement model further comprises selecting the measurement model from a group comprising a Poisson measurement model and a half-Gaussian measurement model and an exponential model.

30. A method as recited in claim 26 , wherein applying the recursive method further comprises determining p(x,t) by dividing P(x,t) by a constant C, such that ∫ p(x,t) dx =1.