System and method for post-stroke rehabilitation and recovery using adaptive surface electromyographic sensing and visualization
A system and method for rehabilitation and recovery using adaptive surface electromyographic sensing and visualization is disclosed. The system uses surface electromyography (sEMG) sensors to identify signals of intent from patients with physical disabilities and then uses these signals to interact with a computer system designed to create repetitive practice in a manner that promotes neurological recovery. A machine teaming module analyses body signals picked up during patient movement attempts and converts these body signals to a visual representation of the intended movement by way of a virtual body or virtual body part displayed on a computer display, display glasses, or the like. The system thus allows for very early patient therapy, providing early benefits to rehabilitation therapy not heretofore possible. The virtual reality or augmented reality environment provides a patient with very early visual reinforcement of beneficial muscle activation patterns.
1. A system for post-stroke rehabilitation and recovery, the system comprising:
a computer having a processor, memory, and access to computer readable media;
a sensing module comprising at least one sensor configured to sense action or intended action of a human body function to be rehabilitated and configured to provide an output;
a machine learning system stored on the computer readable media and configured to receive the output of the at least one sensor and configured to extract attributes from the received output;
a calibration computer program stored on the computer readable media where the computer program executes the step of: identifying patient motor patterns received from the machine learning system;
a virtual extremity controller stored on the computer readable media and configured to convert the extracted attributes into a digital control signal; and
a virtual extremity renderer stored on the computer readable media and configured to render on a computer display a simulated version of the human body function using the digital control signal;
wherein the virtual extremity renderer positively reinforces the desired muscle activity patterns when consistent but weak muscle activation is detected by the machine learning system in the identified patient motor patterns;
wherein the virtual extremity renderer positively reinforces any amount of muscle activity when very weak muscle activation is detected by the machine learning system in the identified patient motor patterns; and
wherein the virtual extremity renderer presents imagery of desired muscle activity patterns when no measurable muscle activation is detected by the machine learning system in the identified patient motor patterns.
2. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the at least one sensor is an electromyography, sensor.
3. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the machine learning system further comprises a temporal filtering system.
4. The system for post-stroke rehabilitation and recovery as defined in claim 3 , wherein the machine learning system is selected from the group consisting of a Markov chain, a neural network, a feedforward network, a convolutional network, a recurrent network, a temporal convolutional network, a generative network, a Hidden Markov model, a naïve Bayes classifier, a support vector machine classifier, a clustering framework, and combinations thereof.
5. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the computer display is an augmented reality device.
6. The system for post-stroke rehabilitation and recovery as defined in claim 1 , further comprising an auxiliary content renderer stored on the computer readable media.
7. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the simulated version of the human body function is a limb muscle motor function.
8. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the machine learning system is configured to determine health of a patient based on signal strength of the output of the sensing module.
9. The system for post-stroke rehabilitation and recovery as defined in claim 8 , wherein the system is configured to enter one of a normal, weak, and ON/OFT operation mode based on a determined health of the patient.
10. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the machine learning system is further configured to classify the extracted attributes into categories, wherein the categories correspond to at least one of a healthy patient class and a weak patient class.
11. The system for post-stroke rehabilitation and recovery as defined in claim 10 , wherein weights are applied to the extracted attributes based on the category to which the extracted attributes correspond.
12. The system for post-stroke rehabilitation and recovery as defined in claim 11 , wherein the virtual extremity controller is further configured to convert weighted attributes into a digital control signal.
13. The system for post-stroke rehabilitation and recovery as defined in claim 1 , further comprising a patient tracking module stored on the computer readable media.
14. The system for post-stroke rehabilitation and recovery as defined in claim 13 , wherein the patient tracking module is configured to track a number of repetitions the patient accomplishes.
15. The system for post-stroke rehabilitation and recovery as defined in claim 13 , wherein the patient tracking module is configured to assign a proficiency score the attributes extracted by the signal processing module.
16. The system for post-stroke rehabilitation and recovery as defined in claim therein the patient hacking module is configured to provide a time series signal of current and past Patient scores to a computer display.
17. The system for post-stroke rehabilitation and recovery as defined in claim 1 , wherein the virtual extremity renderer is configured to provide the simulated version of the human body function on a computer display in a position that corresponds to an anatomical side of a patient that is non-neglected due to stroke.
18. A method for post-stroke rehabilitation and recovery, the method comprising:
sensing an action or intended action of a human body with a sensor;
providing an output of the sensor to a computer having a processor, memory, and access to computer readable media; wherein the computer is configured to extract attributes from the received output;
identifying, on the computer patient motor patterns from the extracted attributes:
when consistent but weak muscle activation detected in the extracted attributes, then desired muscle activity patterns are positively reinforced on a virtual extremity renderer;
when very weak muscle activation is detected in the extracted attributes, then any amount of muscle activation is positively reinforce on the virtual extremity renderer; and
when no measureable muscle activation is detected in the extracted attributes, then imagery of desired muscle activity patterns are presented on the virtual extremity renderer:
converting on the computer the extracted attributes into a digital control signal; and
rendering on a computer display a simulated version of the human body function using the digital control signal.
19. The method of claim 18 , further comprising the step of rendering on the computer auxiliary content along with the simulated version of the human body function.
20. The method of claim 18 , further comprising the step of assigning on the computer weights to the extracted attributes.
21. The method of claim 18 , further comprising the step of assigning on the computer a proficiency score to the extracted attributes.