DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Applicant’s arguments filed in the reply on March 27, 2026 were received and fully considered. Claims 1, 12, and 21 were amended. Please see below for more detail.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 3/27/2026 has been entered.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 3-9, 11-12, and 14-15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Glaser et al (“Real-Time Motor Unit Identification From High-Density Surface EMG”) (“Glaser”) in view of Sun et al (“One-Channel Surface Electromyography Decomposition for Muscle Force Estimation”) (“Sun”) as evidenced by Graimann et al (US 2018/0168477) (“Graimann”).
Regarding Claim 1, while Glaser teaches an apparatus (Abstract) comprising:
a neural interface for obtaining surface electromyography signals of a nervous system of a wearer of the neural interface and outputting digitized signals (p950, II. Data Model, and p952, B. Experimental EMG, “Experimental signals were obtained in a controlled environment with the experimental protocol described in [19]. Eight male subjects participated in the experiment… The surface EMG signals were recorded from tibialis anterior muscle during ankle dorsiflexion. The subject was supine on a bed, with the dominant leg placed in an isometric brace for ankle-joint force measurement… A detection system was combined from two grids of monopolar electrodes each containing 6 5electrodes with 5 mm interelectrode distance. Each electrode had 2 mm in diameter. The grid was placed on top of innervation zone parallel to muscle fibers. All acquired signals were 20 s long and sampled with 2048 samples/s in 12 bit resolution, amplified with EMG amplifier (LISiN—OT Bioelettronica, Torino, Italy) and bandpass filtered between 10 Hz and 500 Hz.” Describes the neural interface used to validate the below analysis steps),
a training step comprising a first algorithm that receives the digitized signals from the neural interface and generates a separation matrix based on first electromyography signals obtained over a first time period by the neural interface (p950, A. Initialization “Denote a block with R samples of observations ӯ(n), 0 < n < R – 1, by
Y
-
R and its correlation matrix by
C
Y
-
R
Y
-
R
. In the initialization part
C
Y
-
R
Y
-
R
-
1
, is computed on the entire block of R samples.” Inverse of the correlation matrix acts as a separation matrix from the R samples of emg signals y, where y is the EMG data with additive zero-mean noise, where the first algorithm would be Equation 6 applied to the block of R samples of observations with an additive zero-mean noise as described in Col. 1);
wherein the training step further comprises an iteration step for refining the separation matrix in real time (p950, III. Real-Time CKC Decomposition, a real-time CKC decomposition is performed and updated as new observation vectors are obtained and decomposed, p951, B. Iterative Part of CKC, “In the iterative part, the Sherman-Morrison-Woodbury formula is employed to iteratively update the inverse of correlation matrix for every new block of observation samples [12], [29]” at a set amount of new blocks of observation samples, p955, C. Time Complexity, a review of the analysis specifically applied to the updating of the correlation matrix found that the updating occurred within 0.02 +/- 0.06 s, thus confirming the refining is occurring within real time); and
a decomposition step that runs in parallel with the training step (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step and as decomposition occurs on the samples, in parallel after a set number of samples Q, the separation matrix is iterated by equation 15 and Equation 8 is updated with a new correlation matrix), wherein the decomposition comprises a second algorithm that receives digitized signals from the neural interface (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step where y(n) represents digitized signals from the neural interface with the additive noise), and the separation matrices can be refined in real time by an iteration step from a training step (p955, Col. 2, “With R and Q set to 5000 and 75 samples, respectively, the initialization part of real-time CKC required 0.75 ± 0.06 s, whereas each update in iterative part required additional 0.02 ± 0.06 s. Thus the decomposition of a 20-s long signal was available within 0.02 s after the end of the signal and the real-time CKC required 0.6 s of processing time for every second of acquired multichannel surface EMG.” the updating of the inverse correlation matrix occurs essentially in real time based on the number of Q samples when applied to synthetic EMG signals),
and (ii) detects one or more motor neuron action potentials for single motor neurons based on second electromyography signals obtained by the neural interface and said separation matrix, wherein said second electromyography signals are generated over a second time period shorter than said first time period when applied to synthetic surface EMG signals (Equation 16 outputs the samples of the train of discharge time of the motor unit, p953-954, A. Synthetic Surface EMG, has a R / first time period length of 5000 samples or 2.5 seconds and has a Q / second time period length of 250 samples which is shorter than first time period. The processing time for the first refinement of the separation matrix is 0.75 ± 0.06 s after 5000 samples and the second refinement after the 250 is 0.02 ± 0.06 s. Thus the motor neuron action potentials obtained by the neural interface are the result of a refined matrix from the initial 5000 samples and represent a shorter time period of 250 samples), and
generate an output in the form of a time-series indicative of motor neuron activity (Figs. 3 and 4 show generated output of motor neuron activity / motor unit discharge patterns),
the steps above achieved by processor modules (p957, Col. 1, “Although consistently identifying fewer MUs than its batch version, real-time CKC shares the robustness in MU identification with batch CKC technique, assuring high accuracy in identification of MU discharges. Its computation complexity depends on the length of the updating blocks of EMG signals, enabling real-time computation on standard personal computers.” And steps performed by processors are carried by software modules as evidenced by Graimann’s [0064] and [0083]).
and while Glaser’s processing time results of the analysis are only applied to synthetic EMG signals, the labeling of the results of the analysis of the experimental EMG signals as real-time CKC indicates this analysis also met the real-time definition enclosed herein (Fig. 4, p955 “the real-time CKC used less than one second of computation time per each second of recorded surface EMG and can, therefore, be considered a real-time method.”).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have the method steps of Glaser be performed by a processor modules as taught by Glaser as a standardization of hardware that can meet the computational complexity requirements for each outlined step.
Yet Glaser fails to teach wherein the output is provided to (i) a prosthetic disposed on the wearer to physically control the prosthetic based on the nervous system of the wearer and wherein the output causes a physical change to the prosthetic.
However Sun teaches a prosthetic control schema (Abstract, Fig. 8) comprising a training step, followed by a surface EMG decomposition step, followed by a motor unit action potential output, and said motor unit action potential being used to physically control a prosthetic based on the nervous system of the wearer, and wherein the output causes a physical change to the prosthetic (Fig. 8, Abstract, p2, Introduction, p6-7, Prosthetic Hand Control, begins with an unsupervised training of learning an orthogonal basis of sEMG signals through reconstruction independent component analysis, a type of blind source separation, enabling a decomposition of one-channel sEMG signals into constituent motor unit action potentials (MUAPs) by the basis vectors, the basis vectors used to create a sparse matrix for decomposing new sEMG signals in real time, sparse matrix used to achieve real-time muscle force estimation for prosthetic hand control, controlling gripping force with multi-channel sEMG-based force estimation controlling prosthetic’s output torque).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to apply the motor unit action potentials derived by Glaser for physical control of an external electrical/electro-mechanical apparatus as taught by Sun as a way to test a training and decomposition routine of multi-channel surface EMG for prosthetics, where multi-channel inputs provide a greater amount of identified motor unit action potential and therefore allows a greater number of output prosthetic controls.
Regarding Claim 3, Glaser and Sun teach the apparatus as claimed in claim 1, wherein the decomposition module comprises a peak detection module for generating the output by processing said one or more motor neuron action potentials to provide a time-series indicative of motor neurons that are fired (See Claim 1 Rejection, for description of applying steps as modules, and p951, I. Initialization and B. Iterative Part of CKC, the identification of discharges above a threshold acts as a peak detection, performed by equations 12 and 17).
Regarding Claim 5, Glaser and Sun teach the apparatus as claimed in claim 1, and Glaser further teaches wherein said neural interface comprises:
an electrode array for measuring surface electrical signals (p952, B. Experimental EMG “A detection system was combined from two grids of monopolar electrodes each containing 6 x 5 electrodes with 5 mm interelectrode distance. Each electrode had 2 mm in diameter. The grid was placed on top of innervation zone parallel to muscle fibers. All acquired signals were 20 s long and sampled with 2048 samples/s in 12 bit resolution, amplified with EMG amplifier (LISiN—OT Bioelettronica, Torino, Italy) and bandpass filtered between 10 Hz and 500 Hz” grids of monopolar electrodes are electrode arrays); and
a signal conditioning module for generating the surface electromyography signals from the surface electrical signals (p952, B. Experimental EMG “A detection system was combined from two grids of monopolar electrodes each containing 6 x 5 electrodes with 5 mm interelectrode distance. Each electrode had 2 mm in diameter. The grid was placed on top of innervation zone parallel to muscle fibers. All acquired signals were 20 s long and sampled with 2048 samples/s in 12 bit resolution, amplified with EMG amplifier (LISiN—OT Bioelettronica, Torino, Italy) and bandpass filtered between 10 Hz and 500 Hz” EMG amplifier is signal conditioning module).
Regarding Claim 6, Glaser and Sun teach the apparatus as claimed in claim 1, and wherein the training module is an adaptive training module (See Claim 1 Rejection, the iterating of the separation matrix in Glaser after Q samples makes the training an adaptive training).
Regarding Claim 7, Glaser and Sun teach the apparatus as claimed in claim 1, wherein the decomposition module detects said motor neuron action potentials by application of said electromyography signals to said separation matrix (See Claim 1 Rejection), and Glaser further teaches wherein a decomposition module detects said motor neuron action potentials by matrix multiplication of electromyography signals and a separation matrix (Equations 8 and 16, ӯ(n) are electromyography signals and
C
^
Y
-
D
+
Q
Y
-
D
+
Q
-
1
is the updated separation matrix).
Regarding Claim 8, Glaser and Sun teach the apparatus as claimed in claim 1, wherein the decomposition module extracts discharge timing of motor neurons (See Claim 1 Rejection¸ Equations 12 and 17), enabling decoding of instructions from individual motor neurons to drive specific outputs at specific times.
Regarding Claim 9, Glaser and Sun teach the apparatus as claimed in claim 1, wherein the training module implement a blind source separation algorithm (See Claim 1 Rejection, convolutive kernel compensation is an example of a blind source separation algorithm).
Regarding Claim 11, Glaser and Sun teach the apparatus as claimed in claim 1, wherein the apparatus is a human-machine interface (See Claim 1 Rejection, Sun’s system provides a human-machine interface between a human’s surface EMG and a machine prosthetic hand).
Regarding Claim 15, Negro, Glaser, and Sun teach the apparatus of claim 1, and Glaser further teaches the apparatus for use in therapy (p955-956, VI. Discussion, “Accurate and rapid feedback on neural drive to muscle response is crucial in many fields of neuromuscular research, including neurorehabilitation [14], [3], ergonomics [24], [28], and training of athletes and astronauts [28]… In this study, a previously introduced and validated CKC decomposition technique [16] has been computationally optimized and algorithmically adapted to real-time online processing. This represents a significant step forward in real-time surface EMG processing, indicating for the first time that the online identification of complete MU discharge patterns is possible during muscle contractions.”).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to apply the real-time motor action potential and discharge timings for the use of therapy as Glaser teaches that accurate feedback of muscle activity is crucial to this field.
Regarding Claim 12, while Glaser teaches a method (Abstract) comprising:
obtaining surface electromyography signals of a nervous system of a wearer of the neural interface and outputting digitized signals (p950, II. Data Model, and p952, B. Experimental EMG, “Experimental signals were obtained in a controlled environment with the experimental protocol described in [19]. Eight male subjects participated in the experiment… The surface EMG signals were recorded from tibialis anterior muscle during ankle dorsiflexion. The subject was supine on a bed, with the dominant leg placed in an isometric brace for ankle-joint force measurement… A detection system was combined from two grids of monopolar electrodes each containing 6 5electrodes with 5 mm interelectrode distance. Each electrode had 2 mm in diameter. The grid was placed on top of innervation zone parallel to muscle fibers. All acquired signals were 20 s long and sampled with 2048 samples/s in 12 bit resolution, amplified with EMG amplifier (LISiN—OT Bioelettronica, Torino, Italy) and bandpass filtered between 10 Hz and 500 Hz.” Describes the neural interface used to validate the below analysis steps),
receiving the digitized signals from the neural interface and generating a separation matrix based on a first algorithm and electromyography signals obtained over a first time period by the neural interface as a training step (p950, A. Initialization “Denote a block with R samples of observations ӯ(n), 0 < n < R – 1, by
Y
-
R and its correlation matrix by
C
Y
-
R
Y
-
R
. In the initialization part
C
Y
-
R
Y
-
R
-
1
, is computed on the entire block of R samples.” Inverse of the correlation matrix acts as a separation matrix from the R samples of emg signals y, where y is the EMG data with additive zero-mean noise, where the first algorithm would be Equation 6 applied to the block of R samples of observations with an additive zero-mean noise as described in Col. 1);
refining the separation matrix as an iteration step in real time (p950, III. Real-Time CKC Decomposition, a real-time CKC decomposition is performed and updated as new observation vectors are obtained and decomposed, p951, B. Iterative Part of CKC, “In the iterative part, the Sherman-Morrison-Woodbury formula is employed to iteratively update the inverse of correlation matrix for every new block of observation samples [12], [29]” at a set amount of new blocks of observation samples, p955, C. Time Complexity, a review of the analysis specifically applied to the updating of the correlation matrix found that the updating occurred within 0.02 +/- 0.06 s, thus confirming the refining is occurring within real time);
receiving the digitized signals from the neural interface and receiving the separation matrix as refined in real time by the iteration step from the training step when applied to synthetic EMG signals (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step where y(n) represents digitized signals from the neural interface with the additive noise, p955, Col. 2, “With R and Q set to 5000 and 75 samples, respectively, the initialization part of real-time CKC required 0.75 ± 0.06 s, whereas each update in iterative part required additional 0.02 ± 0.06 s. Thus the decomposition of a 20-s long signal was available within 0.02 s after the end of the signal and the real-time CKC required 0.6 s of processing time for every second of acquired multichannel surface EMG.” the updating of the inverse correlation matrix occurs essentially in real time based on the number of Q samples when applied to synthetic EMG signals), and detecting one or more motor neuron action potentials for single motor neurons based on a second algorithm and said electromyography signals and said separation matrix when applied to synthetic EMG signals (Equation 16 outputs the samples of the train of discharge time of the motor unit, p953-954, A. Synthetic Surface EMG, has a R / first time period length of 5000 samples or 2.5 seconds and has a Q / second time period length of 250 samples which is shorter than first time period. The processing time for the first refinement of the separation matrix is 0.75 ± 0.06 s after 5000 samples and the second refinement after the 250 is 0.02 ± 0.06 s. Thus the motor neuron action potentials obtained by the neural interface are the result of a refined matrix from the initial 5000 samples and represent a shorter time period of 250 samples); and
Wherein the step of detecting the one or more motor neuron action potentials runs in parallel with the step of generating the separation matrix (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step and as decomposition occurs on the samples, in parallel after a set number of samples Q, the separation matrix is iterated by equation 15 and Equation 8 is updated with a new correlation matrix, Equation 16 outputs the samples of the train of discharge time of the motor unit), and wherein said electromyography signals are provided over a second time period shorter than said first time period when applied to synthetic EMG signals (p953-954, A. Synthetic Surface EMG, has a R / first time period length of 5000 samples or 2.5 seconds and has a Q / second time period length of 250 samples which is shorter than first time period. The processing time for the first refinement of the separation matrix is 0.75 ± 0.06 s after 5000 samples and the second refinement after the 250 is 0.02 ± 0.06 s. Thus the motor neuron action potentials obtained by the neural interface are the result of a refined matrix from the initial 5000 samples and represent a shorter time period of 250 samples),
generating an output in the form of a time-series indicative of motor neuron activity (Figs. 3 and 4 show generated output of motor neuron activity / motor unit discharge patterns),
the steps above achieved by processor modules (p957, Col. 1, “Although consistently identifying fewer MUs than its batch version, real-time CKC shares the robustness in MU identification with batch CKC technique, assuring high accuracy in identification of MU discharges. Its computation complexity depends on the length of the updating blocks of EMG signals, enabling real-time computation on standard personal computers.” And steps performed by processors are carried by software modules as evidenced by Graimann’s [0064] and [0083]).
and while Glaser’s processing time results of the analysis are only applied to synthetic EMG signals, the labeling of the results of the analysis of the experimental EMG signals as real-time CKC indicates this analysis also met the real-time definition enclosed herein (Fig. 4, p955 “the real-time CKC used less than one second of computation time per each second of recorded surface EMG and can, therefore, be considered a real-time method.”).
Glaser fails to teach providing the output to a prosthetic disposed on the wearer to physically control the prosthetic based on the nervous system of the wearer, and wherein the output causes a physical change to the prosthetic.
However Sun teaches a prosthetic control schema (Abstract, Fig. 8) comprising a training step, followed by a surface EMG decomposition step, followed by a motor unit action potential output, and said motor unit action potential being used to physically control a prosthetic based on the nervous system of the wearer, and wherein the output causes a physical change to the prosthetic (Fig. 8, Abstract, p2, Introduction, p6-7, Prosthetic Hand Control, begins with an unsupervised training of learning an orthogonal basis of sEMG signals through reconstruction independent component analysis, a type of blind source separation, enabling a decomposition of one-channel sEMG signals into constituent motor unit action potentials (MUAPs) by the basis vectors, the basis vectors used to create a sparse matrix for decomposing new sEMG signals in real time, sparse matrix used to achieve real-time muscle force estimation for prosthetic hand control, controlling gripping force with multi-channel sEMG-based force estimation controlling prosthetic’s output torque).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to apply the motor unit action potentials derived by Glaser for physical control of an external electrical/electro-mechanical apparatus as taught by Sun as a way to test a training and decomposition routine of multi-channel surface EMG for prosthetics, where multi-channel inputs provide a greater amount of identified motor unit action potential and therefore allows a greater number of output prosthetic controls.
Regarding Claim 14, Glaser and Sun teach the method as claimed in claim 12, further comprising generating the output by processing said one or more motor neuron action potentials to provide a time-series indicative of neurons that are fired (See Claim 1 Rejection, for description of applying steps as modules, and p951, I. Initialization and B. Iterative Part of CKC, the identification of discharges above a threshold acts as a peak detection, performed by equations 12 and 17).
Claim(s) 2 and 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Glaser in view of Sun and further in view of Negro et al (“One-Channel Surface Electromyography Decomposition for Muscle Force Estimation”) (“Negro”).
Regarding Claim 2, while Glaser and Sun teach the apparatus as claimed in claim 1, and Glaser further teaches wherein the training module comprises a module in which the first obtained electromyography signals are extended (p950, Col. 1, the processing of EMG signals-based observation vectors includes extension), their combined efforts fail to teach a convolutive sphering module in which the first obtained electromyography signals are extended and whitened.
However Negro teaches an EMG signal decomposition apparatus (Abstract) comprising:
a neural interface for obtaining surface electromyography signals of a nervous system of a wearer of the neural interface and outputting digitized signals (Abstract, “This approach is then validated using intramuscular signals recorded by novel multichannel thin-film electrodes on the Abductor Digiti Minimi of the hand and Tibilias Anterior muscles, as well as on high-density surface EMG signals recorded by electrode grids on the First Dorsal Interosseous muscle.” measuring performed with neural interface by EMG electrodes);
and a convolutive sphering module in which the first obtained electromyography signals are extended and whitened (p4, Col. 2, the processing of EMG signals to determine the relationship between EMG signals and motor unit action potentials begins with conditioning for blind source separation by extending and whitening EMG signals through convolutive sphering).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to prepare the EMG signals of Glaser by an extending and whitening step as taught by Negro as one of multiple steps to facilitate finding estimated sources for the separation matrix by an intermediate uncorrelation of sources from convolutive sphering.
Regarding Claim 13, while Glaser and Sun teach the method as claimed in claim 12, and Glaser further teaches wherein the training module comprises a module in which the first obtained electromyography signals are extended (p950, Col. 1, the processing of EMG signals-based observation vectors includes extension), their combined efforts fail to teach the training module comprises a convolutive sphering module in which the first obtained electromyography signals are extended and whitened.
However Negro teaches an EMG signal decomposition apparatus (Abstract) comprising:
a neural interface for obtaining surface electromyography signals of a nervous system of a wearer of the neural interface and outputting digitized signals (Abstract, “This approach is then validated using intramuscular signals recorded by novel multichannel thin-film electrodes on the Abductor Digiti Minimi of the hand and Tibilias Anterior muscles, as well as on high-density surface EMG signals recorded by electrode grids on the First Dorsal Interosseous muscle.” measuring performed with neural interface by EMG electrodes);
and a convolutive sphering module in which the first obtained electromyography signals are extended and whitened (p4, Col. 2, the processing of EMG signals to determine the relationship between EMG signals and motor unit action potentials begins with conditioning for blind source separation by extending and whitening EMG signals through convolutive sphering).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to prepare the EMG signals of Glaser by an extending and whitening step as taught by Negro as one of multiple steps to facilitate finding estimated sources for the separation matrix by an intermediate uncorrelation of sources from convolutive sphering.
Claim(s) 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over Glaser in view of Abdelghani et al (US 2014/0330404) (“Abdelghani”) as evidenced by Graimann.
Regarding Claim 21, while Glaser teaches an apparatus (Abstract) comprising:
a neural interface for obtaining surface electromyography signals of a nervous system of a wearer of the neural interface and outputting digitized signals (p950, II. Data Model, and p952, B. Experimental EMG, “Experimental signals were obtained in a controlled environment with the experimental protocol described in [19]. Eight male subjects participated in the experiment… The surface EMG signals were recorded from tibialis anterior muscle during ankle dorsiflexion. The subject was supine on a bed, with the dominant leg placed in an isometric brace for ankle-joint force measurement… A detection system was combined from two grids of monopolar electrodes each containing 6 5electrodes with 5 mm interelectrode distance. Each electrode had 2 mm in diameter. The grid was placed on top of innervation zone parallel to muscle fibers. All acquired signals were 20 s long and sampled with 2048 samples/s in 12 bit resolution, amplified with EMG amplifier (LISiN—OT Bioelettronica, Torino, Italy) and bandpass filtered between 10 Hz and 500 Hz.” Describes the neural interface used to validate the below analysis steps),
a training step comprising a first algorithm that receives the digitized signals from the neural interface and generates a separation matrix based on first electromyography signals obtained over a first time period by the neural interface (p950, A. Initialization “Denote a block with R samples of observations ӯ(n), 0 < n < R – 1, by
Y
-
R and its correlation matrix by
C
Y
-
R
Y
-
R
. In the initialization part
C
Y
-
R
Y
-
R
-
1
, is computed on the entire block of R samples.” Inverse of the correlation matrix acts as a separation matrix from the R samples of emg signals y, where y is the EMG data with additive zero-mean noise, where the first algorithm would be Equation 6 applied to the block of R samples of observations with an additive zero-mean noise as described in Col. 1);
wherein the training step further comprises an iteration step for refining the separation matrix in real time (p950, III. Real-Time CKC Decomposition, a real-time CKC decomposition is performed and updated as new observation vectors are obtained and decomposed, p951, B. Iterative Part of CKC, “In the iterative part, the Sherman-Morrison-Woodbury formula is employed to iteratively update the inverse of correlation matrix for every new block of observation samples [12], [29]” at a set amount of new blocks of observation samples, p955, C. Time Complexity, a review of the analysis specifically applied to the updating of the correlation matrix found that the updating occurred within 0.02 +/- 0.06 s, thus confirming the refining is occurring within real time); and
a decomposition step that runs in parallel with the training step (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step and as decomposition occurs on the samples, in parallel after a set number of samples Q, the separation matrix is iterated by equation 15 and Equation 8 is updated with a new correlation matrix), wherein the decomposition comprises a second algorithm that receives digitized signals from the neural interface (p951, II. Data Model and III. Real-Time CKC Decomposition, Equation 8 is a decomposition step where y(n) represents digitized signals from the neural interface with the additive noise), and the separation matrices can be refined in real time by an iteration step from a training step (p955, Col. 2, “With R and Q set to 5000 and 75 samples, respectively, the initialization part of real-time CKC required 0.75 ± 0.06 s, whereas each update in iterative part required additional 0.02 ± 0.06 s. Thus the decomposition of a 20-s long signal was available within 0.02 s after the end of the signal and the real-time CKC required 0.6 s of processing time for every second of acquired multichannel surface EMG.” the updating of the inverse correlation matrix occurs essentially in real time based on the number of Q samples when applied to synthetic EMG signals),
and (ii) detects one or more motor neuron action potentials for single motor neurons based on second electromyography signals obtained by the neural interface and said separation matrix, wherein said second electromyography signals are generated over a second time period shorter than said first time period when applied to synthetic surface EMG signals (Equation 16 outputs the samples of the train of discharge time of the motor unit, p953-954, A. Synthetic Surface EMG, has a R / first time period length of 5000 samples or 2.5 seconds and has a Q / second time period length of 250 samples which is shorter than first time period. The processing time for the first refinement of the separation matrix is 0.75 ± 0.06 s after 5000 samples and the second refinement after the 250 is 0.02 ± 0.06 s. Thus the motor neuron action potentials obtained by the neural interface are the result of a refined matrix from the initial 5000 samples and represent a shorter time period of 250 samples), and
generate an output in the form of a time-series indicative of motor neuron activity (Figs. 3 and 4 show generated output of motor neuron activity / motor unit discharge patterns),
the steps above achieved by processor modules (p957, Col. 1, “Although consistently identifying fewer MUs than its batch version, real-time CKC shares the robustness in MU identification with batch CKC technique, assuring high accuracy in identification of MU discharges. Its computation complexity depends on the length of the updating blocks of EMG signals, enabling real-time computation on standard personal computers.” And steps performed by processors are carried by software modules as evidenced by Graimann’s [0064] and [0083]).
and while Glaser’s processing time results of the analysis are only applied to synthetic EMG signals, the labeling of the results of the analysis of the experimental EMG signals as real-time CKC indicates this analysis also met the real-time definition enclosed herein (Fig. 4, p955 “the real-time CKC used less than one second of computation time per each second of recorded surface EMG and can, therefore, be considered a real-time method.”).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to have the method steps of Glaser be performed by a processor modules as taught by Glaser as a standardization of hardware that can meet the computational complexity requirements for each outlined step.
Yet Glaser fails to teach wherein the output is provided to a digital system to control actions of a digital avatar in a digital environment based on the nervous system of the wearer.
However Abdelghani teaches a system for decoding intended motor commands from recorded neural signal for virtual environments (Abstract, [0014], [0112]) wherein the output is provided to a digital system to control actions of a digital avatar in a digital environment based on the nervous system of the wearer ([0014], [0076]-[0078], [0112] avatar controlled in a virtual environment based on decoded nervous system signals).
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to apply the motor unit action potentials derived by Glaser for movement replication in an avatar as taught by Abdelghani as a way test a training and decomposition routine of multi-channel surface EMG for prosthetics, where multi-channel inputs provide a greater amount of identified motor unit action potential and therefore allows a greater number of output the state of the user’s motor function.
Response to Arguments
Applicant’s amendments and arguments filed 3/27/2026 with respect to the 35 USC 103 rejections of Claims 1 and 12 with respect to Negro have been fully considered and are persuasive. However, upon further consideration of the reference of Glaser, Examiner maintains that the cited prior art renders the claim language obvious. The independent updating of the correlation matrix (the inverse of a separation matrix) that occurs periodically along with the decomposition of the motor unit action potentials is recognized as a parallel step to refine the trained parameters for said decomposition.
Applicant argues on page 7 that the teachings of Sun fail to teach the ability to generate an output in the form of a time-series indicative of motor neuron activity capable of controlling a multi degree-of-freedom prosthetic in position, velocity, and force. Examiner will note that this level of detail in prosthetic control is not outlined in the written claims and thus this argument is not pertinent. Based on the broadness of the current language, it would be obvious that prosthetics that were previously limited to singular channel inputs for actuating prosthetic movement (Sun) would benefit from a newly realized real-time processing of multiple EMG channels (Glaser) that provides a greater variety of inputs and therefore a greater variety of outputs. Similar arguments apply to Abdelghani for Claim 21.
The rejection stands.
Consequently, Claims 2-9, 11, and 13-15 remain rejected due to their dependency on rejected independent Claims 1 and 12.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAIRO H PORTILLO whose telephone number is (571)272-1073. The examiner can normally be reached M-F 9:00 am - 5:15 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jacqueline Cheng can be reached at (571)272-5596. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/JAIRO H. PORTILLO/
Examiner
Art Unit 3791
/PUYA AGAHI/Primary Examiner, Art Unit 3791
Prosecution Insights