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 6/10/2026, have been fully considered. The following rejections and/or objections are either reiterated or newly applied. They constitute the complete set presently being applied to the instant application.
Applicants have amended their claims, filed 6/10/2026, and therefore rejections newly made in the instant office action have been necessitated by amendment.
Claims 1-7 are the currently pending claims hereby under examination. Claims 1 and 5-6 have been amended.
Claim Objections
Claims 1 and 5 are objected to because of the following informalities:
In claim 1, line 31: “enables the remote cloud system” should be “enables a remote cloud system” since it has not yet been introduced;
In claim 1, line 38: “a remote cloud system” should be “the remote cloud system” for proper antecedent basis;
In claim 5, lines 19-20: “wherein the brainwave data including a plurality of index patterns” lacks a finite verb and should be amended to recite “wherein the brainwave data includes a plurality of index patterns”;
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-4 and 7 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 first recites that “the brainwave physiological signals are converted into a brainwave signal image file divided according to the time sequence into a plurality of sub-images” in lines 7-9. Claim 1 subsequently recites “using the computing terminal to split the electroencephalogram into a plurality of sub-images based on the time sequence” in lines 12-14. The claim therefore introduces “a plurality of sub-images” twice, in connection with two separately recited operations. It is unclear whether these recitations refer to the same plurality of sub-images, two different pluralities of sub-images, or a first plurality that is subsequently divided to produce a second plurality. This ambiguity is material because the claim later recites identifying tags “from the plurality of sub-images,” restoring “each sub-image,” and combining “the restored sub-images.” The claim does not identify which previously recited plurality is used in those subsequent operations. Applicant may resolve this ambiguity by reciting a single operation that converts or divides the brainwave signal image file into the plurality of sub-images, or by expressly defining the relationship between the respective pluralities if two separate sets are intended. For purposes of examination, the Examiner is interpreting that these the recitations are referring to the same plurality of sub-images. Each sub-image is interpreted as a portion of the brainwave signal image file corresponding to a fixed time period of the time sequence.
Claims 2-4 and 7 are rejected by virtue of their dependence from claim 1.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
Claims 1-4 are rejected under 35 U.S.C. 103 as being unpatentable over Hejrati et al. (Behzad Hejrati, Abdolhossein Fathi, and Fardin Abdali-Mohammadi, “Efficient Lossless Multi-Channel EEG Compression Based on Channel Clustering,” Biomedical Signal Processing and Control 31 (2017): 295-300), hereinafter referred to as Hejrati, and further in view of Khalid et al. (Beenish Khalid et al., “EEG Compression Using Motion Compensated Temporal Filtering and Wavelet Based Subband Coding,” IEEE Access 8 (2020): 102502-102511), hereinafter referred to as Khalid, Gürkan et al. (Hakan Gürkan, Umit Guz, and B. Siddik Yarman, “EEG Signal Compression Based on Classified Signature and Envelope Vector Sets,” International Journal of Circuit Theory and Applications 37 (2009): 351-363), hereinafter referred to as Gürkan, CCITT Recommendation H.120, (CCITT Recommendation H.120, “Codecs for Videoconferencing Using Primary Digital Group Transmission,” Fascicle III.6 (1988)), hereinafter referred to as H.120, and Coleman et al. (US 2019/0113973 A1), hereinafter referred to as Coleman.
Regarding claim 1, Hejrati teaches a method for transmitting compressed brainwave physiological signals (Hejrati, Title: “Efficient Lossless Multi-Channel EEG Compression Based on Channel Clustering”; Abstract: “EEG is widely used in telemedicine and neurological research. However, transmitting large EEG data is a challenge, due to high redundancy between channels”; Hejrati discloses a method for transmitting compressed EEG signals for telemedicine applications) comprising: using a brainwave cap to detect a plurality of brainwave physiological signals and generating an electroencephalogram based on a time sequence of the plurality of brainwave physiological signals using a computing terminal (Hejrati, page 295, Introduction: “Electroencephalogram (EEG) demonstrates the electrical activities of the brain by one-dimensional signals gathered by electrodes placed on the scalp,” wherein Hejrati discloses EEG acquisition with electrodes placed on the scalp, corresponding to a brainwave cap, and generating an electroencephalogram comprising a time sequence of brainwave physiological signals; page 298, Experimental Results: the dataset was collected from subjects while performing a motor imagery task and the signals were recorded using a 64-channel system, disclosing a conventional multichannel electrode system used to acquire EEG data; page 295, Introduction: “Intelligent systems are usually parts of remote telemedicine, where the data can be remotely transferred from the person’s mobile phone to any destination, such as a remote terminal in a hospital,” demonstrating the use of a computing terminal); using the computing terminal to split the electroencephalogram into a plurality of portions based on the time sequence, each portion corresponding to a time period of the time sequence (Hejrati, page 296, Proposed Method: “First, the time-domain data of all channels are divided into N symbol blocks”; Hejrati teaches partitioning the electroencephalogram into discrete, time-ordered representations corresponding to defined temporal intervals).
Also regarding claim 1, Hejrati does not expressly disclose that the brainwave physiological signals are converted into a brainwave signal image file divided according to the time sequence into a plurality of sub-images, each sub-image being a divided portion of the brainwave signal image file corresponding to a fixed time period of the time sequence. Rather, Hejrati teaches dividing the time-domain EEG data into N symbol blocks, which provides time-ordered data portions, but does not teach converting those blocks into a two-dimensional brainwave signal image file or arranging fixed-length EEG sample blocks as sub-images corresponding to fixed time periods.
Khalid fills this gap by teaching conversion of a one-dimensional EEG signal into a sequence of two-dimensional image frames. During preprocessing, Khalid divides the EEG signal into non-overlapping segments consisting of N samples, arranges the samples in alternating forward and reverse rows to preserve adjacent-sample correlation, and constructs a sequence of N by N frames (Khalid, page 102504, Section II.A, “Pre-Processing”; Figure 2). Because each frame contains a fixed number of consecutive samples from a time-sampled EEG signal, each frame corresponds to a fixed time period of the EEG time sequence.
Khalid further arranges the EEG frames into successive groups of pictures, or GOPs, and applies motion-compensated temporal filtering to exploit temporal redundancy between the ordered frames (Khalid, Abstract; pages 102503 through 102505, Section II; Figures 1 through 3). Figure 3, on page 102505, depicts the one-dimensional EEG data as an ordered sequence of frames F1 through Ff and divides the frames into successive GOPs. Accordingly, Khalid teaches converting the brainwave physiological signals into a brainwave signal image file divided according to the time sequence into a plurality of sub-images corresponding to fixed time periods.
It would have been prima facie obvious before the effective filing date of the claimed invention to have modified Hejrati in view of Khalid by arranging the consecutive samples of Hejrati’s fixed-length EEG blocks as two-dimensional frames and organizing the resulting frames according to their temporal order. Hejrati already divides EEG data into time-domain blocks for compression, and Khalid applies its frame-conversion procedure to the same type of time-sampled EEG data so that temporal and spatial redundancy can be exploited. The combination would have been implemented using Khalid’s disclosed preprocessing procedure and would predictably have improved compression efficiency while preserving the correspondence between each frame and its original time period.
Also regarding claim 1, with respect to the method being performed in a biological feedback training system and the subject being under brain training in the biological feedback training system, the modified Hejrati teaches acquiring and compressing EEG information for telemedicine, brain-computer-interface applications, and neurological research but does not expressly disclose performing those operations while the subject is undergoing brain training in a biological feedback training system.
Coleman fills this gap by teaching a wearable EEG headset having one or more electrodes for collecting brainwaves from a user (Coleman, ¶[0053]); user effectors that provide vibration, sound, visual indications, or other feedback (Coleman, ¶[0056]); and real-time feedback concerning the user’s current mental state that assists the user in achieving a particular mental state and enables interaction with a meditation training application (Coleman, ¶[0057]). Coleman specifically teaches that a user wears a brainwave headset connected to a mobile phone while meditating, an application processes the user’s brainwaves, and neurofeedback helps the user achieve deeply meditative states and speeds the user’s learning of meditation states (Coleman, ¶[0346]).
It would have been prima facie obvious to have implemented the EEG acquisition and compression operations of the modified Hejrati in Coleman’s meditation training system. Coleman already acquires and processes EEG data during training, while the modified Hejrati provides a compatible procedure for converting that EEG data into time-ordered frames and compressing the frames. The combination would predictably have reduced the EEG data processed or transmitted during training while retaining Coleman’s real-time neurofeedback.
Also regarding claim 1, with respect to using the computing terminal to identify at least one static feature tag and a plurality of associated dynamic displacement tags based on the time sequence from the plurality of sub-images according to a plurality of static feature tags and a plurality of dynamic displacement tags stored in a brainwave database, wherein the at least one static feature tag comprises a static background value fixed over time of the electroencephalogram, and each of the plurality of associated dynamic displacement tags comprises a difference value of an associated sub-image at a corresponding fixed time period of the electroencephalogram based on the time sequence relative to the at least one static feature tag, the modified Hejrati teaches generating reference information and corresponding difference information from EEG data.
Hejrati calculates the difference between each EEG channel and a corresponding cluster centroid, compresses the differences, and transmits each compressed difference together with an index identifying the applicable centroid (Hejrati, page 296, Proposed Method: “the difference between each channel and the corresponding cluster’s centroid is calculated”; Equations 4 through 6; “SendData = (ADi, k)”). Hejrati reconstructs each channel by adding the decoded difference to the corresponding centroid reference (Hejrati, pages 296 and 298, Equations 7 through 10; Figures 1 and 2).
Khalid applies corresponding reference-and-difference processing to the time-ordered EEG frames. Khalid selects a reference frame Fn, searches an adjacent frame Fn+1 for matching blocks, and generates horizontal and vertical motion vectors mapping the associated frame to the reference frame (Khalid, page 102504, Section II.B, Equation 1). Khalid calculates a temporally decomposed high-pass frame according to:
tH(x,y) = 1/2[MCn+1(x,y) − Fn(x,y)]
where Fn is the reference frame and MCn+1 is the motion-compensated associated frame (Khalid, page 102504, Equation 2). The high-pass frame therefore contains difference values of an associated fixed-time EEG frame relative to the reference frame. Khalid also generates a low-pass frame containing information retained for reconstruction and preserves the low-pass and high-pass outputs in their corresponding positions within the ordered GOP structure (Khalid, page 102504, Equation 3; page 102505, Figure 3).
However, the modified Hejrati does not expressly disclose identifying the reference and difference representations according to plural stored static feature tags and dynamic displacement tags maintained in a brainwave database. The modified Hejrati also does not expressly disclose retaining a background reference value unchanged over multiple fixed-time EEG frames and calculating difference values relative to that maintained background value.
Gürkan fills the stored-brainwave-representation and identification gaps by teaching classified signature and envelope vector sets, referred to as CSEVS, generated from EEG frames using k-means clustering (Gürkan, Abstract; pages 354 through 355, Section 2.2; Figure 1). Gürkan constructs two separately stored categories of reusable EEG vectors: a classified signature set containing plural classified signature vectors and a classified envelope set containing plural classified envelope vectors (Gürkan, pages 353 through 355). The transmitter and receiver possess the same CSEVS, and each EEG frame is represented by an index R identifying an entry in the classified signature set, an index K identifying an entry in the classified envelope set, and a gain coefficient Ci (Gürkan, Abstract; pages 352 through 357). The encoder selects the vectors that minimize the modeling error and stores the corresponding indices and gain coefficient (Gürkan, page 356, Section 2.3, Steps 2a through 5; Equations 14 through 17). The receiver retrieves the identified vectors from the stored CSEVS and reconstructs the associated EEG frame according to XAi = CiΦKΨR (Gürkan, pages 356 through 357, Equation 18; Figure 3). Gürkan further teaches using an EEG training database to generate the CSEVS (Gürkan, page 358). Gürkan thereby teaches a brainwave-database architecture containing two distinct stored collections of reusable EEG representations and corresponding identifiers for selecting entries from both collections.
Gürkan does not expressly characterize its original classified signature and envelope vectors as background values and background-relative difference values. H.120 supplies the maintained-background and background-relative-difference representations to be stored using Gürkan’s two-set architecture. H.120 adaptively selects among a motion-compensated interframe prediction value Mi, a background prediction value Bi, and an intraframe prediction value Ii to minimize probable prediction errors (H.120, page 44, Section 3.6.2.2; page 45, Figure 10/H.120 and Equations 3-1 through 3-3). Figure 10/H.120 shows a prediction error e generated by subtracting the selected prediction value from the input and shows reconstruction by adding the prediction error to the selected prediction value. When the background prediction value Bi is selected, the prediction error e therefore comprises a difference value relative to the background prediction value.
H.120 further defines an update-control parameter under which the preceding-frame background prediction value is retained for the consecutive k−1 frames following an update frame (H.120, pages 46 through 47, Section 3.6.2.3; Equations 3-7 and 3-8; Figure 11/H.120). The background prediction value thereby remains fixed over multiple successive frame periods before being updated.
It would have been prima facie obvious before the effective filing date of the claimed invention to have applied Gürkan’s disclosed frame-modeling, clustering, storage, and indexing procedure separately to the two EEG-derived frame populations produced by the modified Hejrati: the maintained background frames and the background-relative prediction-error frames. The modification would not merely store the H.120 frames unchanged in Gürkan’s CSEVS. Rather, each background frame and prediction-error frame would be processed using Gürkan’s disclosed frame-modeling procedure, under which an EEG frame vector Xi is represented using a selected classified signature vector, a selected classified envelope vector, and a gain coefficient Ci. The signature and envelope vectors derived from the background-frame population would be clustered to form an indexed background CSEVS, and the signature and envelope vectors derived from the prediction-error-frame population would be clustered to form a separately indexed prediction-error CSEVS (Gürkan, pages 352 through 355, Sections 2.1 and 2.2; Figures 1 and 2; Equations 1 through 13).
The stored background CSEVS would contain plural reusable EEG background representations corresponding to the plurality of static feature tags, and the stored prediction-error CSEVS would contain plural reusable EEG difference representations corresponding to the plurality of dynamic displacement tags. For each fixed-time EEG frame, the computing terminal would select indices RB and KB and a gain coefficient CB identifying the applicable stored background representation and indices RE and KE and a gain coefficient CE identifying the applicable stored background-relative prediction-error representation. The selected background representation would remain fixed for the associated frame periods according to H.120’s background-update procedure, and the selected prediction-error representation would provide the difference value for the corresponding fixed-time EEG frame. At the receiver, the background and prediction-error representations would be reconstructed using Gürkan’s Equation 18 and then integrated using H.120’s disclosed prediction-error add-back operation. Thus, both tag types would be stored and identified using Gürkan’s EEG-specific storage and indexing technique, while H.120 would supply the background-prediction and background-relative-error relationship between the two stored representation types.
A person of ordinary skill would have had a concrete reason to apply Gürkan’s classification procedure separately to the background and prediction-error frame populations. Gürkan expressly teaches that constructing CSEVS from clustered centroid vectors reduces the size of the vector sets and drastically reduces the computational complexity of the searching and matching process while achieving significant compression ratios (Gürkan, page 352; pages 354 through 355, Section 2.2). Khalid and H.120 already divide the EEG image sequence into a comparatively stable background or reference component and a time-varying difference component to remove temporal redundancy. Those operations produce repeated or similar background patterns and repeated or similar prediction-error patterns across the ordered EEG frames. Applying Gürkan’s clustering separately to those two populations would reduce the number of distinct background and prediction-error representations that must be maintained or transmitted, because recurring representations could be replaced by compact indices to corresponding centroid-based models. This is the same reduction in stored vector-set size, search complexity, and transmitted data that Gürkan identifies as the benefit of its classified-vector technique.
Although H.120 concerns videoconferencing rather than physiological-signal analysis, H.120 is reasonably pertinent to the particular compression problem presented by the modified Hejrati. Khalid itself converts one-dimensional EEG samples into an ordered sequence of two-dimensional image frames, organizes those EEG frames into GOPs, and applies motion-compensated temporal filtering to exploit interframe redundancy. After applying Khalid, the artisan is confronting the same technical problem addressed by H.120: efficiently encoding and reconstructing an ordered sequence of two-dimensional frames by maintaining background information, encoding differences from the applicable prediction, transmitting motion and mode information, and controlling decoder reconstruction. H.120 consequently would logically have commended itself to the attention of a person implementing Khalid’s GOP-based, motion-compensated EEG-frame compression.
The proposed modification would have preserved the respective functions of the references. Khalid and H.120 would continue to separate the ordered EEG image sequence into background and prediction-error information; Gürkan would continue to model EEG-derived frames using classified signature and envelope vectors, cluster similar vectors into reusable centroid-based sets, and identify the selected representations through compact indices; and H.120 would continue to reconstruct each frame by integrating the applicable background and prediction-error information. Because Gürkan’s procedure operates on EEG frame vectors and the background and prediction-error frames are EEG-derived numerical frame vectors produced by the preceding compression operations, applying the same disclosed modeling and clustering procedure to each frame population would have involved the predictable use of Gürkan’s technique for its stated data-reduction purpose. The result would have been fewer distinct background and prediction-error representations, reduced searching and matching complexity, reduced transmission volume, and retention of the information needed for frame-by-frame reconstruction.
Also regarding claim 1, with respect to using the computing terminal to generate at least one superimposed group tag, wherein the at least one superimposed group tag is a grouping integration command that associates each of the plurality of associated dynamic displacement tags with its corresponding fixed time period in the time sequence relative to the at least one static feature tag, such that the electroencephalogram is transformed into the compressed brainwave physiological signals comprising the identified at least one static feature tag, the plurality of associated dynamic displacement tags, and the at least one superimposed group tag, the modified Hejrati teaches the ordered EEG frames and the reference, difference, motion, and database-index information needed to represent those frames.
Khalid preserves the order and temporal-decomposition relationship of the EEG frames within each GOP (Khalid, pages 102504 through 102505, Figures 2 and 3; Equations 1 through 3), and Gürkan provides ordered sequences of signature indices, envelope indices, and gain coefficients that identify the stored vectors used to reconstruct each EEG frame (Gürkan, pages 356 through 357, Figures 2 and 3; Equation 18). However, the modified Hejrati does not expressly disclose a transmitted grouping integration command directing the decoder how the applicable background representation and associated difference information are to be interpreted and integrated for each ordered frame.
H.120 fills this gap by teaching that the coder determines the operating modes, transmits those modes with the coded data as a combination of commands, and causes the decoder to reproduce the signal according to the received commands and data (H.120, page 53, Section 3.6.4.3; Table 6/H.120). H.120 multiplexes prediction-error data e and motion-vector data v with coding-mode data m and provides a format for each frame containing frame synchronization, frame-mode data, line-mode data, motion-vector data, and prediction-error data (H.120, page 54, Section 3.6.5.1; Figures 14 and 15/H.120). The frame-sync word designates the start of a frame (H.120, page 54, Section 3.6.5.2.1). The frame-mode data include background-refresh and background-update commands controlling transfer to and updating of the background frame memory (H.120, page 55, Section 3.6.5.2.2; Figures 16 and 17/H.120). H.120 separately defines the motion-vector and prediction-error data included in the frame representation (H.120, pages 57 through 58, Sections 3.6.5.2.5 and 3.6.5.2.6; Figures 21 and 22/H.120).
When applied to Khalid’s ordered EEG GOPs and Gürkan’s database indices, H.120’s frame synchronization, coding-mode commands, background-memory commands, motion-vector data, and prediction-error data collectively form grouping and control information that identifies the beginning of each ordered frame representation, identifies whether and how the maintained background is used or updated, and identifies the difference and motion information interpreted for that frame. The resulting grouping integration command associates each background-relative difference representation with its corresponding fixed-time EEG frame and the selected stored background representation required to restore that frame.
It would have been prima facie obvious to have packaged the modified Hejrati information in H.120’s command-controlled, frame-synchronized structure. Khalid already requires preservation of the GOP organization, frame order, temporal-decomposition relationship, and motion-vector association; Gürkan already provides ordered indices and coefficients identifying the stored EEG representations; and H.120 provides a conventional command structure communicating how corresponding background, motion, and prediction-error information is interpreted by a decoder. The modification would have included, for each ordered EEG frame or GOP, the applicable frame synchronization and mode commands, the index of the static background representation, the corresponding prediction-error representation, and associated motion and gain information. Each component would have continued to perform its known function, predictably enabling reliable frame-by-frame reconstruction while preserving the EEG time sequence.
Also regarding claim 1, with respect to the at least one superimposed group tag enabling the remote cloud system to restore each sub-image of the electroencephalogram by integrating the corresponding associated dynamic displacement tag with the at least one static feature tag at the corresponding fixed time period according to the time sequence, and to reconstruct the electroencephalogram by combining the restored sub-images according to the time sequence; and using the computing terminal to transmit the identified at least one static feature tag, the plurality of associated dynamic displacement tags, and the at least one superimposed group tag instead of the detected plurality of brainwave physiological signals to a remote cloud system for analysis according to the time sequence so as to reduce the volume of data required in transmission to the remote cloud system and enable real-time biological feedback to the subject; wherein the method enables the biological feedback training system to provide the real-time biological feedback to the subject and improves efficiency of the brain training by transmitting the compressed brainwave physiological signals, the modified Hejrati teaches compression, transmission, and reconstruction of time-ordered EEG information.
Hejrati teaches compressing EEG signals to address the difficulty of transmitting large EEG datasets and transmitting the compressed information from a person’s mobile phone to a remote destination such as a hospital terminal (Hejrati, Abstract; page 295, Introduction). Hejrati reconstructs the EEG by decoding the difference information, adding each difference to the applicable centroid reference, and applying inverse DPCM (Hejrati, pages 296 and 298, Equations 7 through 10; Figures 1 and 2). Khalid encodes the spatiotemporally decomposed EEG frames into a scalable bitstream and evaluates the reconstructed EEG against the original EEG, thereby confirming reconstruction of the ordered EEG frames (Khalid, pages 102503 through 102510, Sections II and III). Gürkan similarly retrieves the stored EEG vectors identified by R and K and reconstructs each EEG frame according to Equation 18 (Gürkan, pages 356 through 357).
H.120 further teaches that the predictive encoder converts the input signal into a prediction-error signal e using motion vector v and coding mode m and that the variable word-length coder codes e and v into compressed data while also coding the mode m (H.120, page 42, Section 3.6.1; Figure 9/H.120). At the decoder, the variable word-length decoder decodes e, v, m, and frame-memory-parity information p, and the predictive decoder reproduces the signal (H.120, page 43, Section 3.6.1; Figure 9/H.120). Figure 10/H.120 shows reconstruction by adding the decoded prediction error to the selected prediction value, including the maintained background prediction value when background prediction is selected (H.120, pages 44 through 45, Section 3.6.2.2; Figure 10/H.120; Equations 3-1 through 3-3). Applied to the ordered EEG frames of the modified Hejrati, these operations restore the respective EEG sub-images from the selected static background representation and associated difference information and reconstruct the EEG by combining the restored frames in Khalid’s GOP and frame order.
However, the modified Hejrati does not expressly disclose transmitting the compressed static-reference identifiers, associated background-relative difference information, and grouping commands to a remote cloud system that performs frame restoration, EEG reconstruction, and analysis and enables real-time biological feedback based on the cloud analysis.
Coleman fills this gap by teaching that collected biosignal data may be processed and analyzed by a remotely located server, cloud-based server, or SAAS platform (Coleman, ¶[0058]). Coleman further teaches that functions performed by the client device may instead be performed by the SAAS platform, that sensor data may be transmitted to the SAAS platform for processing, analysis, and storage, and that bandwidth limitations may require processing to be performed on the client before transmission (Coleman, ¶[0357]). Coleman’s SAAS platform contains a cloud-based database and analyzer that store raw, preprocessed, or analyzed user data and process the sensor data (Coleman, ¶[0359]-[0360]). As established above, Coleman also provides real-time feedback concerning the user’s mental state and uses neurofeedback to help the user achieve meditative states and speed the user’s learning of meditation states (Coleman, ¶[0057] and [0346]).
It would have been prima facie obvious to have transmitted the compressed static-background index, associated background-relative difference representations, and grouping commands of the modified Hejrati to Coleman’s cloud system for reconstruction and analysis according to the time sequence. The modified Hejrati already produces a compact, ordered, command-controlled EEG representation suitable for transmission instead of the complete detected EEG signals. Coleman teaches moving biosignal processing functions to a cloud system, recognizes bandwidth limitations as a reason to perform preprocessing at the client, and provides real-time neurofeedback based on processed EEG data.
The combination would have been implemented by configuring the computing terminal to generate and transmit the compressed static-reference index, difference, motion, gain, frame-order, and command information; configuring Coleman’s cloud system to maintain the corresponding stored EEG representations, decode the commands and data, restore each fixed-time EEG frame by integrating the applicable background representation and associated difference information, combine the restored frames according to their GOP and frame order, and analyze the reconstructed EEG; and using Coleman’s user effectors to provide real-time biological feedback. The combination would predictably have reduced transmission volume, supported centralized cloud reconstruction and analysis, and improved the efficiency of the brain training by enabling timely biological feedback from the compressed EEG information.
Regarding claim 2, the modified Hejrati teaches the method for transmitting compressed brainwave physiological signals of claim 1, wherein the at least one static feature tag is a static base value of a brainwave physiological signal, and each of the associated dynamic displacement tags is a difference value of the brainwave physiological signal of the associated sub-image relative to the static base value based on the time sequence.
As established regarding claim 1, Khalid converts consecutive EEG samples into ordered two-dimensional EEG frames corresponding to fixed time periods and organizes those frames into successive GOPs (Khalid, pages 102504 through 102505, Section II.A; Figures 2 and 3). H.120 selects a background prediction value Bi and generates a prediction error e by subtracting the selected prediction value from the corresponding input frame; reconstruction is performed by adding the prediction error back to the selected prediction value (H.120, pages 44 through 45, Section 3.6.2.2; Figure 10/H.120; Equations 3-1 through 3-3). H.120 retains the preceding-frame background prediction value for the consecutive k−1 frames following an update frame, such that the background prediction value remains fixed across multiple successive frame periods (H.120, pages 46 through 47, Section 3.6.2.3; Equations 3-7 and 3-8; Figure 11/H.120).
When H.120’s background-prediction procedure is applied to Khalid’s ordered EEG frames, the maintained background representation comprises a static base value of the EEG physiological signal, and the prediction error for each associated EEG frame comprises a difference value of the brainwave physiological signal represented by that sub-image relative to the maintained static base value. Because each Khalid frame corresponds to a fixed period of the EEG time sequence and H.120 associates each prediction error with the background applicable to that frame period, the difference values are determined relative to the static base value based on the time sequence.
In the modified system described regarding claim 1, Gürkan’s modeling, clustering, storage, and indexing procedure is applied separately to the EEG-derived background representations and the EEG-derived background-relative prediction-error representations. The selected static feature tag is the compressed representation of a numerical EEG background value reconstructed from the applicable background-model indices and gain coefficient, while each associated dynamic displacement tag is the compressed representation of a numerical EEG prediction-error value reconstructed from the applicable prediction-error-model indices and gain coefficient. The indices and coefficients are not relied upon merely as abstract labels; together with the corresponding stored CSEVS vectors, they define the numerical background and prediction-error values used to reconstruct the EEG frame under Gürkan’s Equation 18 (Gürkan, pages 356 through 357, Equation 18; Figure 3). Accordingly, the selected static feature tag is a static base value of the brainwave physiological signal maintained for the applicable frame periods, and each associated dynamic displacement tag is a difference value of the brainwave physiological signal of the corresponding EEG sub-image relative to that static base value.
Hejrati further corroborates the suitability of applying a base-and-difference representation to EEG data, although H.120, as applied to Khalid’s time-ordered EEG frames, supplies the maintained static base value and the claimed temporal relationship. Hejrati divides time-domain EEG data into N-symbol blocks, calculates a cluster-centroid reference from the EEG data, calculates each channel difference according to Di = DEi − Ck, and reconstructs the EEG data according to DEi = Di + Ck (Hejrati, page 296, Proposed Method and Equations 4 through 6; page 298, Equations 8 through 10; Figures 1 and 2). Hejrati therefore teaches, in the EEG domain itself, representing brainwave physiological signals using an EEG-derived base value and corresponding difference values.
No additional modification beyond the combination established for claim 1 is required. Claim 2 specifies the nature of the background and prediction-error representations already produced when H.120’s maintained-background procedure is applied to Khalid’s time-ordered EEG sub-images and the resulting representations are modeled, stored, and identified using Gürkan’s two-set EEG database architecture.
Regarding claim 3, the modified Hejrati teaches the method for transmitting compressed brainwave physiological signals of claim 1, wherein the plurality of brainwave physiological signals include power, frequency, current, current source density, asymmetry, coherence, or phase lag, except that Hejrati does not expressly characterize the acquired EEG information using any of the physiological-signal or EEG-feature categories enumerated in claim 3.
Coleman fills this gap by teaching that signal-processed EEG data include EEG band power calculated by the application and used to calculate a score or feedback from the user’s EEG signal (Coleman, ¶[0081]). Coleman further teaches frequency-domain EEG processing that produces a power spectrum, power per EEG band, power variance per EEG band, and peak frequency within a particular band (Coleman, ¶[0146]). Coleman also teaches measuring frequency-band powers and EEG frequency-spectrum characteristics in biosignal data (Coleman, ¶[0401]) and estimating hemispheric asymmetries from EEG signals (Coleman, ¶[0070] and [0151]). Accordingly, Coleman teaches at least the claimed alternatives that the brainwave physiological signals include power, frequency, or asymmetry. Because claim 3 separates the enumerated alternatives using “or,” Coleman’s disclosure of any one alternative is sufficient to satisfy the limitation, and the remaining alternatives of current, current source density, coherence, and phase lag need not be separately addressed.
It would have been prima facie obvious before the effective filing date of the claimed invention to have configured the modified Hejrati to include EEG power, frequency, or asymmetry information among the brainwave physiological information processed and transmitted by the system. The claim 1 combination already incorporates Hejrati’s acquisition and compression of EEG information into Coleman’s biological-feedback training system, and Coleman expressly calculates EEG band-power and frequency-domain features from the acquired EEG signals to generate scores and feedback concerning the user’s brain state. Coleman also identifies the conventional signal-processing operations used to obtain those features, including FFT, wavelet, Hilbert, power-spectrum, and EEG-band processing (Coleman, ¶[0146]). The modification would have been implemented by applying Coleman’s disclosed EEG feature-processing pipeline to the EEG information acquired by the modified Hejrati so that power, frequency, or asymmetry information is included in the brainwave physiological information compressed and transmitted using the architecture established regarding claim 1 and is available after reconstruction for cloud analysis. The benefit would have been preservation of physiologically meaningful EEG features used by Coleman to determine the subject’s brain state and generate real-time biological feedback, while retaining the reduced transmission volume provided by the compression architecture established regarding claim 1.
Regarding claim 4, the modified Hejrati teaches the method for transmitting compressed brainwave physiological signals of claim 1 further comprising using the remote cloud system to integrate the identified at least one static feature tag and the plurality of associated dynamic displacement tags according to the time sequence and the at least one superimposed group tag to restore a plurality of sub-images; and combining the restored sub-images to obtain the electroencephalogram according to the time sequence; wherein the at least one superimposed group tag is a message for integrating the static feature tag and the plurality of associated dynamic displacement tags to restore a plurality of sub-images according to the time sequence.
As established regarding claim 1, Khalid converts the EEG into an ordered sequence of fixed-time EEG frames and preserves the temporal positions and decomposition relationships of those frames within successive GOPs (Khalid, pages 102504 through 102505, Figures 2 and 3; Equations 1 through 3). In the modified system established regarding claim 1, Gürkan’s Equation 18 is applied separately to reconstruct the EEG-derived background representation and the corresponding EEG-derived prediction-error representations from their respective stored vectors, indices, and gain coefficients (Gürkan, pages 356 through 357, Equation 18; Figure 3).
H.120 teaches transmitting the coding modes with the coded data as a combination of commands that causes the decoder to reproduce the signal according to the received commands and data (H.120, page 53, Section 3.6.4.3; Table 6/H.120). H.120’s transmitted frame representation includes frame synchronization, frame-mode data, line-mode data, motion-vector data, and prediction-error data, including background-refresh and background-update commands controlling the background frame memory (H.120, pages 54 through 58, Sections 3.6.5.1 through 3.6.5.2.6; Figures 14 through 22/H.120). When used in the modified system established regarding claim 1 to package the applicable Gürkan background-model indices and coefficients, prediction-error-model indices and coefficients, and associated frame-order, motion, and mode information, H.120’s transmitted frame representation constitutes a machine-readable message directing the decoder regarding which static-background representation and associated dynamic prediction-error representation are to be integrated for each ordered EEG frame.
H.120 further teaches restoring each frame by adding the decoded prediction error to the selected prediction value, including the maintained background prediction value when background prediction is selected (H.120, pages 44 through 45, Section 3.6.2.2; Figure 10/H.120; Equations 3-1 through 3-3). Applied to Khalid’s ordered EEG frames and Gürkan’s stored EEG representations, the decoder uses the grouping message to integrate the applicable static-background representation and associated dynamic prediction-error representation to restore each EEG sub-image and then combines the restored sub-images in Khalid’s GOP and frame order to obtain the electroencephalogram according to the time sequence.
Coleman teaches that biosignal processing and analysis may be performed by a remotely located server, cloud-based server, or SAAS platform and that functions otherwise performed by the client device may instead be performed by the SAAS platform after sensor data are transmitted for processing, analysis, and storage (Coleman, ¶[0058] and [0357]). Coleman further teaches a cloud-based database and analyzer that store and process raw or preprocessed biosignal data (Coleman, ¶[0359]-[0360]). In the combination established regarding claim 1, Coleman’s remote cloud system maintains the corresponding stored EEG representations, decodes H.120’s grouping commands and compressed data, restores the EEG sub-images by integrating the applicable background and prediction-error representations, and combines the restored sub-images according to their temporal order.
No additional modification beyond the combination established for claim 1 is required. Claim 4 recites as express cloud-system operations the same integration, frame restoration, and time-ordered EEG reconstruction that the claim 1 combination already enables and performs.
Claims 5-6 are rejected under 35 U.S.C. 103 as being unpatentable over Coleman et al. (US 2019/0113973 A1), hereinafter referred to as Coleman, and further in view of Gürkan et al. (Hakan Gürkan, Umit Guz, and B. Siddik Yarman, “EEG Signal Compression Based on Classified Signature and Envelope Vector Sets,” International Journal of Circuit Theory and Applications 37 (2009): 351-363), hereinafter referred to as Gürkan.
Regarding claim 5, Coleman teaches a transmission method for compressed brainwave physiological signals in a biological feedback training system comprising using a brainwave cap to detect a plurality of brainwave physiological signals of a subject under brain training in the biological feedback training system and generating an electroencephalogram based on a time sequence of the plurality of brainwave physiological signals using a computing terminal in the biological feedback training system, except that Coleman does not expressly disclose the compact index-pattern compression supplied by Gürkan as discussed below.
Coleman teaches a wearable EEG headset having one or more electrodes that collects brainwaves from a user and communicates with a computing device (Coleman, ¶[0053]). Coleman further teaches EEG headsets having three, four, or more electrodes (Coleman, ¶[0070]); storing raw biosignal data as a time-based, time-coded series for corresponding EEG channels (Coleman, ¶[0091]); and a user wearing a brainwave headset connected to a mobile phone while meditating, with the phone processing the brainwaves and providing neurofeedback that helps the user achieve meditative states and speeds the user’s learning of meditation states (Coleman, ¶[0346]).
Also regarding claim 5, Coleman teaches using the computing terminal to analyze a correlation of the brainwave physiological signals between different channels of the brainwave cap based on the time sequence to generate a plurality of complex electroencephalograms, wherein the correlation analysis comprises at least one of coherence, phase lag, power spectrum, or asymmetry of the brainwave physiological signals.
Coleman teaches acquiring EEG data through an EEG headset having one or more electrodes and explains that using additional electrodes permits collection of additional or more accurate EEG information (Coleman, ¶[0053]). Coleman further teaches EEG headsets having three, four, or more electrodes and using the resulting multichannel EEG information to estimate hemispheric asymmetries (Coleman, ¶[0070]).
Coleman stores EEG channel data as a time-based, time-coded series (Coleman, ¶[0091]). Coleman’s signal-processing pipeline defines processing windows by window size and increment and extracts features from the EEG signal using frequency-domain transformations, including short-time Fourier transforms (Coleman, ¶[0142]-[0143]).
More specifically, Coleman teaches frequency-domain EEG processing that determines phase differences across electrodes within EEG frequency bands (Coleman, ¶[0146]). Determining the phase difference between signals acquired through different electrodes analyzes the temporal phase relationship between different EEG channels and therefore corresponds to the claimed phase-lag correlation analysis. Because claim 5 requires “at least one of” coherence, phase lag, power spectrum, or asymmetry, Coleman’s cross-electrode phase-difference processing is sufficient to satisfy the recited correlation-analysis alternative.
Coleman also teaches determining power spectra and power within EEG frequency bands and estimating hemispheric asymmetries from multichannel EEG information (Coleman, ¶[0146] and [0151]). These disclosures provide additional examples of the claimed correlation-analysis alternatives but are not relied upon independently to establish that a single asymmetry value constitutes a complex electroencephalogram.
Coleman applies its frequency-domain processing to time-coded EEG data using defined processing windows. For each time-positioned processing window, Coleman generates a set of frequency-band features that includes the phase relationships between signals obtained from different electrodes. The resulting set of cross-electrode phase-difference information constitutes an interchannel EEG-correlation representation for that time period. Repeating the processing across successive time-coded windows generates a plurality of such representations according to the EEG time sequence.
This interpretation is consistent with the specification. The specification characterizes a complex electroencephalogram as being generated by algorithmically analyzing correlations between brainwaves obtained from different points or channels, including coherence, phase lag, power, and asymmetry, according to the time sequence (Specification, ¶[0023] and [0028]). The specification further explains that coherence analysis may be performed from the power spectral densities and cross-power spectral density associated with EEG signals obtained at different electrode positions and frequency bands (Specification, ¶[0029]). The specification illustrates an EEG-correlation graph generated using Fourier-transform processing as one representation of the resulting correlation information (Specification, ¶[0033]).
Accordingly, under the broadest reasonable interpretation consistent with the specification, a complex electroencephalogram encompasses a time-positioned interchannel EEG-correlation representation generated from phase-lag, coherence, spectral, or asymmetry analysis. Although the specification illustrates an EEG-correlation graph, claim 5 does not require the complex electroencephalogram to be rendered as a particular line graph, image, matrix, or other graphical format. Coleman’s time-positioned sets of cross-electrode phase-difference information therefore correspond to the claimed plurality of complex electroencephalograms.
Coleman further teaches using the computing terminal to identify a sequence of feature tags from the plurality of complex electroencephalograms according to the time sequence, wherein each feature tag in the sequence characterizes a pattern of the brainwave physiological signals identified from the complex electroencephalograms based on the time sequence. Coleman teaches extracting features from EEG data for use in machine-learning prediction models (Coleman, ¶[0144]-[0146]). Coleman teaches quantizing or discretizing continuous EEG feature values into discrete values (Coleman, ¶[0147] and [0222]). Coleman also defines a brain signature as a set of features that are collected, quantized, stored as a multidimensional representation, and used to classify the user’s state (Coleman, ¶[0181]).
Coleman further teaches feature-event data containing identifications and corresponding values, with the feature events being time-coded so they can be referenced against biosignal data having the same or similar time codes (Coleman, ¶[0219]). A feature event includes the variables forming a data point, and Coleman searches combinations of those feature events to identify matching patterns (Coleman, ¶[0219] and [0222]). Under the broadest reasonable interpretation, each identification and discrete value assigned to an EEG feature corresponds to a feature tag characterizing the applicable EEG pattern, and the ordered time-coded feature events constitute the claimed sequence of feature tags.
Also regarding claim 5, with respect to using the computing terminal to generate a biological feature sequence comprising a plurality of index patterns based on the identified sequence of feature tags and the time sequence according to brainwave data stored in a brainwave database, wherein the brainwave data include a plurality of index patterns with each index pattern composed of a plurality of feature tags and generated by training a neural network using a plurality of electroencephalograms, and the electroencephalogram is transformed into the compressed brainwave physiological signals comprising the biological feature sequence with each index pattern of the biological feature sequence being identified from the brainwave database based on the identified sequence of feature tags, Coleman teaches the feature-combination, stored-pattern, pattern-matching, and neural-network-training aspects of the limitation but does not expressly disclose representing the matched patterns as compact index patterns or transforming the electroencephalogram into compressed brainwave physiological signals comprising such index patterns.
Coleman teaches maintaining a database containing combinations of feature events, finding matching patterns, and using higher-order feature events for classification (Coleman, ¶[0219]). Coleman explains that an EEG session or epoch is represented by multiple features, that continuous features such as EEG alpha power are quantized into discrete bins, and that second-order and higher-order feature events are combinations of multiple discrete feature values (Coleman, ¶[0222]). Coleman further teaches accumulating significant patterns or user-response classifications in a pattern database maintained by the machine-learning module (Coleman, ¶[0232]). Accordingly, Coleman’s stored significant patterns comprise combinations of multiple discrete EEG feature identifications and values corresponding to the claimed combinations of feature tags.
Coleman teaches using machine learning to discover brain signatures from features extracted by analyzing EEG signals across multiple user sessions (Coleman, ¶[0176]). Coleman further teaches building generalized prediction pipelines from labeled training data collected from multiple users and refining those models using accumulated EEG data (Coleman, ¶[0177]-[0178]). Coleman expressly identifies artificial neural networks as one of the machine-learning techniques used to develop its prediction models from the extracted features (Coleman, ¶[0145] and [0152]).
Although Coleman does not expressly require every entry in its pattern database to be generated using the disclosed artificial-neural-network option, it would have been prima facie obvious before the effective filing date to use Coleman’s disclosed artificial neural network to train the EEG prediction model and populate the pattern database with the resulting significant EEG feature patterns or classifications. Coleman teaches artificial neural networks as an available technique for developing the same prediction models, using the same extracted EEG features and training data, that produce the classifications and significant patterns stored by the machine-learning module.
The modification would have been implemented by training Coleman’s artificial neural network using the accumulated EEG feature data from multiple sessions and users, using the trained model to classify combinations of quantized EEG feature values, and storing the resulting significant feature combinations or classifications as reusable entries in Coleman’s pattern database. This would have involved selecting one of Coleman’s expressly identified machine-learning alternatives for its disclosed prediction-modeling purpose and would predictably have produced reusable EEG feature patterns for classification and feedback.
However, Coleman does not expressly disclose characterizing the stored EEG feature patterns as compact index patterns and transmitting a plurality of such index patterns instead of the detected plurality of brainwave physiological signals.
Gürkan fills this gap by teaching an EEG-specific compression architecture in which the transmitter and receiver possess the same stored classified signature and envelope vector sets, referred to as CSEVS, and each EEG frame is represented using an index R identifying a classified signature vector, an index K identifying a classified envelope vector, and a gain coefficient Ci (Gürkan, Abstract; pages 352 through 357). Gürkan selects the stored vectors that minimize the modeling error for the EEG frame and provides binary-coded sequences of the selected indices and gain coefficients to the receiver, which retrieves the corresponding stored vectors and reconstructs the EEG frame according to Equation 18 (Gürkan, pages 356 through 357, Figure 3; Equations 14 through 18). Gürkan thereby teaches replacing complete EEG frame data with compact identifiers that invoke corresponding stored EEG representations.
It would have been prima facie obvious before the effective filing date to have modified Coleman in view of Gürkan by assigning compact indices to Coleman’s stored EEG feature patterns and representing each matched pattern using its database index rather than transmitting the complete underlying EEG data or complete combination of feature values. Coleman already quantizes EEG features, stores combinations of feature events as reusable patterns, identifies matching patterns, recognizes that bandwidth limitations may require client-side processing before transmission, and performs cloud-based pattern analysis (Coleman, ¶[0219], [0222], [0232], and [0357]). Gürkan teaches the compatible EEG-specific technique of maintaining the same stored representations at the transmitter and receiver and transmitting compact indices identifying the selected representations.
The proposed modification would not apply Gürkan’s signature-and-envelope modeling procedure to Coleman’s feature patterns or alter how Coleman generates those patterns. Rather, the modification would apply Gürkan’s disclosed shared-database indexing technique to Coleman’s already-stored EEG feature-pattern entries so that each entry could be invoked through a compact identifier.
The modification would have been implemented by storing Coleman’s neural-network-generated EEG feature patterns as indexed entries in the brainwave database. Each indexed entry would comprise Coleman’s combination of multiple discrete EEG feature tags. For each time-positioned complex electroencephalogram, the computing terminal would identify the corresponding feature tags, match their combination to a stored pattern, and place the compact index identifying that pattern into the biological feature sequence. The resulting ordered sequence of database-identified patterns constitutes the claimed biological feature sequence comprising a plurality of index patterns.
A person of ordinary skill would have had a concrete reason to make this modification because Coleman expressly recognizes bandwidth limitations as a reason to perform processing and analysis on the client before transmission to its SAAS platform (Coleman, ¶[0357]), while Gürkan demonstrates that EEG representations may be stored at both ends and invoked using compact indices to obtain substantial compression (Gürkan, Abstract; pages 356 through 362). The combination would predictably have reduced the amount of EEG information transmitted while preserving the feature-pattern information required for Coleman’s cloud-based classification and feedback. Each reference would continue to perform its established function: Coleman would extract, quantize, classify, store, and match EEG feature patterns, while Gürkan’s shared-database indexing technique would provide compact identifiers for invoking the stored representations.
Also regarding claim 5, Coleman teaches using the computing terminal to transmit the plurality of index patterns comprised in the biological feature sequence instead of the detected plurality of brainwave physiological signals to a remote cloud system for analysis according to the time sequence so as to reduce time and data required in transmission to the remote cloud system and enable real-time biological feedback to the subject; wherein the transmission method enables the biological feedback training system to provide the real-time biological feedback to the subject and improves efficiency of the brain training by transmitting the compressed brainwave physiological signals, as modified by Gürkan’s compact indexing technique.
Coleman teaches that functions performed by the client device may instead be performed by a SAAS platform, that sensor data may be transmitted to the SAAS platform for processing, analysis, and storage, and that bandwidth limitations may require client-side processing before transmission (Coleman, ¶[0357]). Coleman’s SAAS platform includes a cloud-based database and analyzer that store and process raw, preprocessed, or analyzed biosignal data (Coleman, ¶[0359]-[0360]). Coleman also teaches providing real-time feedback concerning the user’s mental state and using neurofeedback to help the user achieve meditative states and speed the user’s learning of those states (Coleman, ¶[0057] and [0346]).
In the modified Coleman system, each index pattern comprises a Coleman stored combination of multiple discrete EEG feature tags together with the compact database identifier assigned to that pattern using Gürkan’s shared-database indexing technique. The identifier is not relied upon as an abstract label alone; together with the corresponding entry in the shared pattern database, it defines and invokes the complete feature-tag combination constituting the index pattern. The computing terminal transmits the ordered compact identifiers for the matched index patterns instead of transmitting the complete detected EEG signals, and the cloud system uses the same indexed pattern database to retrieve and analyze the corresponding EEG feature patterns according to their temporal order. The result would have been reduced transmission time and data volume, scalable cloud analysis, and more efficient real-time biological feedback during brain training.
Regarding claim 6, the modified Coleman teaches using the remote cloud system to compare the biological feature sequence according to the plurality of index patterns transmitted by the computing terminal with the brainwave data stored in the brainwave database, wherein the brainwave data associates each of the plurality of index patterns with at least one predefined category label corresponding to a neurophysiological state or behavioral performance state of the subject (Coleman, ¶[0219], teaching that the system maintains a database containing combinations of feature events, finds matching patterns, and uses the highest-order matching feature events to perform classification; ¶[0177], teaching the use of labeled EEG training data to build brain-state classification algorithms; ¶[0181], teaching a brain signature comprising collected, quantized, and stored features used to classify the user’s state; ¶[0221], identifying predefined categories including “drowsy,” “alert,” and “agitated”; ¶[0222], teaching that a significant combination of feature values may predict a category such as “moderately relaxed”; and ¶[0232], teaching a pattern database containing significant patterns or user-response classifications).
In the modified Coleman established regarding claim 5, each compact index transmitted in the biological feature sequence identifies a corresponding stored Coleman EEG feature pattern. Because Coleman associates each applicable stored pattern with a brain-state or user-response classification, assigning Gürkan’s compact index to the pattern preserves that association. Matching the received index to the corresponding indexed database entry therefore identifies the stored feature pattern and its associated predefined category label.
Coleman teaches that functions performed by the client device may instead be performed by its remote SAAS platform and that sensor data may be transmitted to the SAAS platform for processing, analysis, and storage (Coleman, ¶[0357]). Coleman’s SAAS platform includes a cloud-based database and analyzer that store and process raw, preprocessed, and analyzed biosignal and user-profile data (Coleman, ¶[0359]-[0360]). Accordingly, the modified Coleman teaches performing the claimed comparison at the remote cloud system.
The modified Coleman further teaches using the remote cloud system to determine at least one biological feedback signal based on the at least one predefined category label associated with the index patterns of the biological feature sequence, and to transmit the at least one biological feedback signal to the biological feedback training system for providing real-time biological feedback to the subject (Coleman, ¶[0176], teaching that brain signatures and resulting brain-state classifications provide quantitative or qualitative real-time feedback; ¶[0056]-[0057], teaching user effectors that provide vibration, sound, visual indications, or other real-time feedback concerning the user’s current mental state and assist the user in achieving a desired state; ¶[0368], teaching a cloud multiuser real-time interaction server that processes biosignal data and transmits feedback signals in response to mental-state determinations to provide real-time feedback through network-connected output devices; and ¶[0346], teaching neurofeedback that helps a user achieve meditative states and speeds the user’s learning of those states).
Accordingly, the modified Coleman teaches comparing the received biological feature sequence with the indexed brainwave data stored in the cloud database, identifying the predefined category associated with each indexed EEG feature pattern, determining a biological-feedback signal based on the identified category, and transmitting that signal to the biological-feedback training system for real-time presentation to the subject.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Hejrati et al. (Behzad Hejrati, Abdolhossein Fathi, and Fardin Abdali-Mohammadi, “Efficient Lossless Multi-Channel EEG Compression Based on Channel Clustering,” Biomedical Signal Processing and Control 31 (2017): 295–300), hereinafter referred to as Hejrati, and further in view of Khalid et al. (Beenish Khalid et al., “EEG Compression Using Motion Compensated Temporal Filtering and Wavelet Based Subband Coding,” IEEE Access 8 (2020): 102502–102511), hereinafter referred to as Khalid, Gürkan et al. (Hakan Gürkan, Umit Guz, and B. Siddik Yarman, “EEG Signal Compression Based on Classified Signature and Envelope Vector Sets,” International Journal of Circuit Theory and Applications 37 (2009): 351–363), hereinafter referred to as Gürkan, CCITT Recommendation H.120, (CCITT Recommendation H.120, “Codecs for Videoconferencing Using Primary Digital Group Transmission,” Fascicle III.6 (1988)), hereinafter referred to as H.120, Coleman et al. (US 2019/0113973 A1), hereinafter referred to as Coleman, and further in view of Tidare et al. (Jonatan Tidare, Elaine Åstrand, and Martin Ekström, “Evaluation of Closed-Loop Feedback System Delay: A Time-Critical Perspective for Neurofeedback Training”, Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies, BIODEVICES (2018): 187–193), hereinafter referred to as Tidare.
The modified Hejrati teaches claim 1 as shown above.
Regarding claim 7, the modified Hejrati does not expressly disclose that the method maintains a round-trip packet latency of less than three seconds, thereby providing reduced end-to-end communication delay. As discussed above regarding claim 1, the modified Hejrati teaches the method for transmitting compressed brainwave physiological signals recited in claim 1. The findings and rationales for combining Hejrati, Khalid, Gürkan, H.120, and Coleman set forth regarding claim 1 are incorporated herein.
Coleman teaches a biological-feedback training system in which EEG information is processed and feedback concerning the user’s mental state is provided in real time (Coleman, ¶[0057] and [0346]). Coleman further teaches transmitting biosignal information to a remote cloud-based server or SAAS platform for processing and analysis (Coleman, ¶[0058] and [0357]-[0360]).
Coleman also teaches measuring network round-trip delay. Coleman teaches synchronizing different components by sending timing information back and forth until the system obtains an adequate estimate of network delay, including by computing round-trip delay time and clock offset (Coleman, ¶[0200]). Coleman therefore recognizes round-trip network delay as a measurable parameter in its distributed EEG-processing system but does not expressly specify maintaining that delay below three seconds.
Tidare fills this gap by teaching an EEG neurofeedback system specifically designed to evaluate and minimize closed-loop delay between EEG acquisition and the return of feedback to the subject. Tidare explains that a real-time closed-loop feedback system should provide neurofeedback with minimal delay and that excessive delay may reduce the subject’s ability to control and use the feedback system (Tidare, page 187, Abstract and Introduction).
Tidare teaches continuously streaming EEG data from EEG-acquisition software to an external signal-processing program and returning visual or tactile feedback to the subject. Tidare’s EEG-acquisition software uses TCP/IP to stream EEG data and event markers to an external processing program. The EEG packets are buffered for either 20 or 50 milliseconds before transmission, the received data are forwarded to the signal-processing thread, and feedback is transmitted back to the game engine (Tidare, pages 188 through 189, Sections 2.1 through 2.5; Figure 3).
Tidare expressly measures both a pure closed-loop system delay, principally reflecting delay caused by data streaming, and a full closed-loop system delay, which includes the streaming protocol, stimulation software, EEG packet-wait time, and presentation of the returned feedback (Tidare, page 189, Section 2.6).
Tidare performed 50 repetitions for each closed-loop-delay measurement (Tidare, page 189, Section 2.6). Tidare reports full closed-loop delays of approximately 90 to 131 milliseconds for multichannel EEG configurations, a worst-case delay of approximately 294 milliseconds for another tested configuration, and a maximum delay of approximately 424 milliseconds under a stressed configuration (Tidare, pages 190 through 191, Results; Figures 5 through 7). Each of these measured closed-loop delays is substantially less than three seconds.
Under the broadest reasonable interpretation consistent with the specification, the claimed round-trip packet latency encompasses the closed-loop interval beginning with transmission of the subject’s physiological-signal information and ending with receipt of the resulting feedback. The specification describes the relevant interval as the time between the subject generating the signal and receiving the feedback signal after transmission, comparison, and return of the feedback (Specification, ¶[0039]-[0041]).
Tidare measures the corresponding closed-loop interval because EEG information is transmitted through a TCP/IP streaming protocol to the processing component and the resulting feedback is transmitted back through the system. Moreover, Tidare’s full closed-loop measurement includes EEG packet buffering and transmission, processing and stimulation software, and presentation of the returned feedback. Because the maximum measured delay for the complete loop remained approximately 424 milliseconds, the packet-transmission portion of that complete delay was necessarily also less than three seconds.
It would have been prima facie obvious before the effective filing date of the claimed invention to implement the EEG transmission and feedback loop of the modified Hejrati using Tidare’s packet-based streaming, bounded packet-buffering, and closed-loop-delay-monitoring practices. The modified Hejrati already transmits a reduced-volume EEG representation to Coleman’s remote cloud system and returns real-time biological feedback to the subject. Coleman further teaches measuring round-trip network delay, while Tidare teaches measuring the complete EEG acquisition-to-feedback loop and demonstrates maximum measured delays substantially below three seconds. A person of ordinary skill would have had reason to combine these teachings because Tidare teaches that minimizing closed-loop delay improves the subject’s ability to control and use neurofeedback and emphasizes testing system delay before conducting real-time brain-computer-interface experiments. The combination would have been implemented by using Coleman’s round-trip-delay measurement to monitor communication between the computing terminal and remote cloud system, transmitting the compressed EEG information using packet-based streaming, bounding packet-buffering and processing intervals, and returning the resulting feedback through the monitored communication loop. Tidare’s reported maximum full closed-loop delay of approximately 424 milliseconds demonstrates that the claimed three-second ceiling was readily achievable using conventional EEG-neurofeedback streaming practices. The benefit would have been a monitored EEG transmission and feedback loop that maintains round-trip packet latency below three seconds, thereby reducing end-to-end communication delay and allowing the subject to respond effectively to the biological feedback during brain training.
Response to Arguments
35 U.S.C. §101
Applicant's arguments filed 6/10/2026, pages 7-15, regarding the previous 101 Rejections of claims 1-7 have been fully considered. Applicant’s argument that the claims do not recite a judicial exception is not persuasive. Nevertheless, the amendments materially alter the eligibility analysis. Upon reconsideration of the amended claims as a whole, the recited mathematical concepts are integrated into a practical application at step 2A, Prong Two. Accordingly, the rejections of claims 1-7 under 35 U.S.C. §101 have been withdrawn.
Applicant's Argument: Applicant argues that claims 1 through 7 do not recite a mathematical concept because the claims do not recite a mathematical formula and the claimed operations cannot practically be performed in the human mind. Applicant further argues that any alleged exception is integrated into a practical application directed to EEG compression and real-time biological feedback, including an alleged improvement in the effective throughput of the brainwave cap's detection pipeline. Applicant also argues that the number of references cited in the prior-art rejections demonstrates that the claimed combination was not well-understood, routine, or conventional, while separately recognizing that subject-matter eligibility and nonobviousness are distinct inquiries.
Examiner's Response: Applicant's argument that the claims recite no mathematical concept is not persuasive. The recited operations include generating difference values, performing correlation analysis, extracting features, matching patterns, and training a neural network. Such operations can recite mathematical calculations or mathematical relationships even when no equation appears in the claim. Further, the mathematical-concept category is not limited to calculations that a human can practically complete mentally. The asserted inability to perform the entire claimed operation in the human mind therefore does not remove the recited mathematical concepts from the judicial-exception analysis.
Examiner's Response: Nevertheless, the amendments place claim 1 in a materially different posture at Step 2A, Prong Two. Considered as an ordered combination, amended claim 1 converts EEG signals into an image file, divides the image file into fixed-time sub-images, uses a fixed background value and corresponding difference values, generates a grouping command associating the difference values with their respective time periods, reconstructs the EEG signals from the compressed representation, and transmits that representation instead of the raw EEG signals to reduce the volume of transmitted data. These limitations recite a particular technical mechanism for addressing EEG-data transmission, rather than merely stating a desired result or directing use of an abstract idea on a generic computer.
Examiner's Response: Current USPTO guidance instructs that a claim need not recite numerical parameters or implementation detail at the level of source code. The claim must include the components or steps that provide the disclosed improvement, and the claim must be evaluated as a whole and as an ordered combination without dismissing potentially meaningful limitations merely as generic computer components. See MPEP §§ 2106.04(d)(1) and 2106.05(a); Advance Notice of Change to the MPEP in Light of Ex parte Desjardins, dated December 5, 2025. Applying that guidance, the specific compression, grouping, reconstruction, and reduced-data transmission steps now recited in claim 1 sufficiently reflect a technological improvement and integrate the recited mathematical concepts into a practical application.
Examiner's Response: Applicant's repeated reliance on improving the effective throughput of the brainwave cap's detection pipeline is not commensurate with the scope of claim 1 because the amended claim does not recite that asserted pipeline-throughput result. The absence of that phrase is not by itself dispositive because a claim need not expressly state the improvement. Here, withdrawal is instead supported by the particular technical steps that the claim does recite and by the claim's express transmission of the compressed representation in place of the raw signals to reduce transmitted data volume.
Examiner's Response: Applicant is correct that subject-matter eligibility and nonobviousness are distinct inquiries. For that same reason, the number of references used in a rejection under 35 U.S.C. § 103 does not establish whether an additional element or ordered combination was well-understood, routine, or conventional for purposes of Step 2B. The number of references is not evidence that the claim satisfies or fails the Step 2B inquiry. Because the claims are no longer directed to a judicial exception after application of Step 2A, Prong Two, however, it is unnecessary to reach Step 2B. Applicant's Berkheimer arguments and the remaining Step 2B arguments are therefore moot. In view of the amendments and upon reconsideration of the claims as a whole, the rejection of claims 1 through 7 under 35 U.S.C. § 101 has been withdrawn.
35 U.S.C. §103
Applicant's arguments filed 6/10/2026, pages 15-44, regarding the previous 103 Rejections of claims 1-7 have been fully considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. That is, there are new grounds of rejection. Additionally:
Prior Grounds of Rejection
Applicant's Argument: Applicant argues that the former combinations based on Hejrati, Coleman, Normile, and the FCC report fail to teach the amended claims. Applicant particularly argues that Normile is nonanalogous art, that Normile's vector-quantization technique is incompatible with Hejrati's lossless EEG compression, that Normile does not teach the claimed static background value or superimposed group tag, and that the FCC report does not teach a physiologically meaningful neurofeedback-latency requirement.
Examiner's Response: The prior grounds of rejection relying on Normile and the FCC report have been withdrawn and replaced by the grounds stated in the present Office Action. Therefore, Applicant's arguments directed specifically to Normile's codebook, Normile's alleged nonanalogous status, the compatibility of Normile with Hejrati, Normile's block headers, and the FCC report's broadband-testing context are moot. The present rejection of claims 1 through 4 relies on the combined teachings of Hejrati, Khalid, Gürkan, H.120, and Coleman. Claims 5 and 6 are rejected over Coleman in view of Gürkan. Claim 7 is rejected over Hejrati, Khalid, Gürkan, H.120, and Coleman, and further in view of Tidare. The remaining arguments have been considered to the extent that they address limitations or combination issues that remain relevant to the present grounds.
Claims 1 Through 4
Applicant's Argument: Applicant argues that Hejrati operates on one-dimensional numerical EEG signals and uses a centroid that is spatially defined and recomputed for each block, rather than a static background value fixed over time. Applicant further argues that the cited art fails to teach conversion of the EEG into fixed-time sub-images, stored static and dynamic feature tags, a superimposed group tag that associates each difference value with its time period and static background, and reconstruction of the EEG at a remote cloud system. Applicant contends that the claimed B/M/G structure must be considered as a unified architecture and cannot be assembled by mapping isolated teachings from separate references.
Examiner's Response: Applicant's arguments do not address the references according to the respective teachings for which they are relied upon in the present rejection. The rejection does not rely on Hejrati's cluster centroid alone as the claimed time-fixed background, nor does it rely on Hejrati alone for the image-domain or grouping-command limitations. Hejrati supplies the acquisition, block processing, compression, transmission, and base-and-difference reconstruction of multichannel EEG data. Khalid expressly converts consecutive one-dimensional EEG samples into ordered two-dimensional EEG frames corresponding to fixed sample intervals and organizes those frames into successive groups of pictures. Gürkan supplies an EEG-specific database containing stored reusable representations and corresponding indices selected to represent EEG frames. H.120 supplies a background prediction value retained across successive frame periods, background-relative prediction errors, transmitted mode and synchronization commands, and reconstruction by adding the applicable prediction error to the selected background prediction. Coleman supplies the biological-feedback training context, remote cloud processing and analysis, and return of real-time neurofeedback. The rejection addresses the combined teachings, not whether any one reference independently anticipates the claim. See In re Keller, 642 F.2d 413, 425 (CCPA 1981); In re Mouttet, 686 F.3d 1322, 1332 through 1333 (Fed. Cir. 2012).
Examiner's Response: The newly recited image conversion does not distinguish the claim from the present combination. Khalid's preprocessing divides a time-sampled EEG signal into fixed numbers of consecutive samples and arranges those samples as two-dimensional frames. Each resulting frame is a divided image portion associated with a fixed period of the EEG time sequence. Applicant's observation that Hejrati itself processes numerical channel vectors therefore does not address the modification expressly supplied by Khalid.
Examiner's Response: The phrase "a static background value fixed over time of the electroencephalogram" is given patentable weight, but it does not require a neurological baseline having a particular physiological meaning, a value transmitted only once for an entire recording, or a value that can never be updated. Under the broadest reasonable interpretation consistent with the specification, the language encompasses an EEG-derived background representation maintained unchanged for the associated successive frame periods. H.120 retains the preceding-frame background prediction value for the consecutive k minus 1 frames following an update frame and determines a prediction error relative to that maintained value. When this procedure is applied to Khalid's ordered EEG frames, the maintained value is a static background value of the EEG for those fixed-time periods, and each prediction error is a difference value of the associated EEG frame relative to that value. Gürkan's separate stored vector sets and indices provide the claimed stored and identified representations of the resulting EEG-derived background and difference information.
Examiner's Response: The present rejection also does not treat the superimposed group tag as an undifferentiated header. H.120 transmits frame synchronization, frame-mode commands, background-refresh and background-update commands, motion-vector information, and prediction-error information. In the stated combination, this machine-readable command and control information identifies the applicable maintained background, identifies the corresponding difference information for each ordered EEG frame, and directs the decoder regarding how the information is to be interpreted and integrated. The decoder then adds the difference representation to the applicable background representation to restore the frame, and the restored frames are combined in Khalid's frame and group order. Coleman teaches placing the corresponding reconstruction and analysis functions at the remote cloud system. These combined teachings supply the claimed association of each dynamic displacement tag with its corresponding fixed time period relative to the static feature tag and the claimed time-ordered cloud reconstruction.
Examiner's Response: Applicant's characterization of the B/M/G terminology as a unified architecture does not require that the entire architecture appear in a single reference. Section 103 expressly permits consideration of the claimed subject matter as a whole in view of the combined teachings of the prior art. The present rejection explains how the identified background representation, difference representations, stored indices, temporal ordering, grouping commands, add-back operation, transmission, and cloud reconstruction operate together in the modified system. To the extent Applicant argues only that Hejrati, Khalid, Gürkan, H.120, or Coleman individually lacks the complete architecture, such an individual attack does not overcome the combined teaching. In re Keller, 642 F.2d at 425.
Examiner's Response: The dependent-claim arguments are likewise unpersuasive. Regarding claim 2, H.120's maintained EEG background and background-relative prediction error, modeled and identified using Gürkan's stored EEG representations, provide the recited static base value and associated difference values. Regarding claim 3, Coleman expressly teaches EEG power, frequency, and asymmetry information. Because the claim recites the listed signal types in the alternative, teaching one alternative is sufficient. Regarding claim 4, H.120's received commands and prediction-error add-back restore the respective frames, Khalid preserves their temporal order, and Coleman places the reconstruction and analysis at the remote cloud system, as fully mapped in the rejection.
Analogous Art, Motivation, and Operability
Applicant's Argument: Applicant argues that image and video coding art is outside the field of EEG compression, that combining lossless EEG compression with other compression techniques would destroy Hejrati's intended function, that no person of ordinary skill would have combined the references, and that the number and diversity of the references demonstrate hindsight reconstruction.
Examiner's Response: The argument is not persuasive as applied to the present grounds. Khalid and Gürkan are directed specifically to EEG compression, and Coleman and Tidare are directed to EEG-based feedback systems. H.120 is reasonably pertinent to the particular problem faced by the claimed invention. The specification and claim concern reducing the amount of data required to transmit and reconstruct an ordered EEG image sequence. Khalid itself converts EEG samples into two-dimensional frames, organizes them into groups of pictures, and applies motion-compensated temporal filtering derived from image and video coding to exploit interframe redundancy. Once Khalid's disclosed EEG-frame representation is used, an artisan faces the same frame-coding problem addressed by H.120: maintaining background information, encoding frame differences and motion information, transmitting decoder-control commands, and reconstructing ordered frames. H.120 therefore logically would have commended itself to the attention of a person implementing Khalid's EEG-frame compression. See MPEP § 2141.01(a).
Examiner's Response: The stated combination also does not require bodily incorporation of every component of each reference or replacement of Hejrati's entire compression system. The test is what the combined teachings would have suggested to a person of ordinary skill, not whether the references can be physically combined without alteration. In re Keller, 642 F.2d at 425; In re Mouttet, 686 F.3d at 1332. The modification applies Khalid's disclosed EEG-frame conversion, Gürkan's disclosed EEG-vector modeling and indexing, and H.120's disclosed background-prediction and command-controlled reconstruction to EEG-derived frame information. Each technique continues to perform its established function. Moreover, the claims do not require lossless compression, medical-diagnostic fidelity, or preservation of every advantage Hejrati identifies. Applicant has not shown that the stated combination would be inoperative for the claimed purpose of compressing, transmitting, reconstructing, and analyzing EEG information for biological feedback.
Examiner's Response: The rationales for the modifications arise from the references themselves rather than from Applicant's disclosure. Hejrati and Khalid seek to reduce redundancy in transmitted EEG data. Gürkan teaches that clustered reusable EEG vector sets reduce stored-set size, search complexity, and transmitted data. H.120 teaches background prediction and command-controlled reconstruction of ordered frames. Coleman recognizes bandwidth limitations, client-side preprocessing, cloud analysis, and real-time neurofeedback. Tidare teaches minimizing and measuring the complete EEG acquisition-to-feedback delay. These express teachings provide reasons to make the respective modifications and support a reasonable expectation that the modified system would compress, transmit, reconstruct, and analyze EEG information as claimed. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 417 through 418, 421 (2007). Reliance on several references does not, without more, establish hindsight or nonobviousness. In re Gorman, 933 F.2d 982 (Fed. Cir. 1991).
Claims 5 and 6
Applicant's Argument: Applicant argues that the prior art does not teach cross-channel correlation analysis producing complex electroencephalograms, feature tags identified from those complex electroencephalograms, or a brainwave database containing index patterns composed of feature tags and generated by training a neural network using electroencephalograms. Applicant further argues that amended claim 6 recites objective category labels, neurophysiological or behavioral-performance states, and transmission of a biological-feedback signal, rather than a mental process.
Examiner's Response: The argument does not overcome the present rejection over Coleman in view of Gürkan. Coleman processes multichannel EEG data in time-positioned windows and expressly determines phase differences across electrodes within EEG frequency bands. That operation is a phase-lag analysis of the relationship between signals from different channels. Claim 5 requires at least one of coherence, phase lag, power spectrum, or asymmetry, so Coleman's phase-difference processing satisfies the recited alternative. Coleman also teaches power spectra and hemispheric asymmetry. Consistent with the specification, the resulting time-positioned sets of interchannel relationship information are complex electroencephalograms. The claim does not require a particular graphical rendering, matrix, or image format for each complex electroencephalogram.
Examiner's Response: Coleman further teaches extracting, identifying, quantizing, and time-coding EEG features; storing combinations of feature events as significant patterns; matching feature combinations to stored patterns; using EEG data from multiple sessions and users as labeled training data; and using artificial neural networks as a disclosed machine-learning option. The rejection explains why it would have been obvious to use Coleman's expressly disclosed artificial-neural-network option to train the EEG prediction model and populate the pattern database with the resulting reusable combinations or classifications of EEG feature values. Gürkan supplies the compatible EEG-specific technique of assigning compact indices to stored reusable EEG representations, maintaining corresponding representations at the transmitter and receiver, and transmitting the indices instead of the full EEG information. In the combination, each compact index identifies a stored Coleman pattern composed of multiple discrete EEG feature tags and generated through the selected neural-network training procedure. The ordered identifiers form the claimed biological feature sequence.
Examiner's Response: Regarding claim 6, Coleman associates stored EEG feature patterns with classifications including drowsy, alert, agitated, and relaxed states; performs the comparison and classification using a cloud-based analyzer; determines feedback based on the classified state; and transmits real-time feedback signals to network-connected output devices. Assigning Gürkan's compact index to the stored Coleman pattern preserves the association between that indexed pattern and its predefined state or performance label. Thus, the amended terminology of claim 6 is expressly addressed by the present rejection. Applicant's separate assertion that the amended language is not a mental process relates to the withdrawn rejection under 35 U.S.C. § 101 and does not establish nonobviousness under 35 U.S.C. § 103.
Claim 7 and Alleged Unexpected Results
Applicant's Argument: Applicant argues that a round-trip packet latency of less than three seconds is a physiologically meaningful result caused by the claimed B/M/G compression architecture, rather than a generic network threshold. Applicant further characterizes the sub-three-second result as unexpected objective evidence of nonobviousness.
Examiner's Response: The FCC report is not relied upon in the present rejection. Tidare instead concerns the same EEG neurofeedback context recited by claim 7. Tidare streams EEG packets to an external processing program, returns visual or tactile feedback, measures the complete closed-loop delay including packet buffering, transmission, processing, stimulation software, and presentation of the returned feedback, and reports measured full-loop delays well below three seconds. Coleman additionally teaches measuring round-trip network delay in a distributed EEG-processing system. The present rejection therefore relies on art directed to the claimed physiological-feedback context, not on a generic consumer-broadband threshold.
Examiner's Response: Claim 7 requires maintaining the recited latency and thereby reducing end-to-end communication delay, but it does not require that the threshold be achieved solely by the claimed compression tags, does not specify a particular packet payload, bandwidth, channel count, or compression ratio, and does not require a particular cognitive-association window beyond the stated numerical limit. The combination already transmits the reduced EEG representation and returns biological feedback. Implementing Tidare's packet streaming, bounded buffering, and delay-monitoring practices in that feedback loop would have predictably maintained the full closed-loop delay below three seconds, as demonstrated by Tidare's reported measurements.
Examiner's Response: Applicant's assertion of unexpected results has been considered but is entitled to little probative weight. Unexpected results must be supported by factual evidence rather than attorney argument. In re De Blauwe, 736 F.2d 699, 705 (Fed. Cir. 1984); MPEP § 716.01(c). The cited specification passages describe the desired three-second limit and Applicant's asserted causal explanation, but Applicant has not provided comparative testing against the closest prior art, measured results for an embodiment within the scope of the claims, or evidence showing that the result is attributable to the allegedly distinguishing features and is commensurate with the scope of the claims. See MPEP §§ 716.01(b), 716.02(d), and 716.02(e). Moreover, Tidare's measured EEG closed-loop delays below three seconds are evidence that the recited result was expected and readily achievable in the art. On balance, the asserted unexpected result does not outweigh the evidence supporting the prima facie case of obviousness.
Support for the Amendments
Applicant's Argument: Applicant identifies passages of the specification and drawings as support for the amendments to claims 1, 5, and 6 and argues that the amendments add the concrete technical detail requested in the prior Office Action.
Examiner's Response: Applicant's identified support has been considered, and no rejection under 35 U.S.C. § 112(a) based on new matter is made in this Office Action. Written-description support, however, does not by itself establish novelty or nonobviousness over the prior art. The amended limitations are addressed by the rejections above. Separately, amended claim 1 introduces the plurality of sub-images in two recited operations without making clear whether the same plurality or different pluralities are intended. That issue is addressed in the rejection under 35 U.S.C. § 112(b) in this Office Action.
Accordingly, Applicant's arguments do not overcome the present rejections under 35 U.S.C. § 103, and those rejections are maintained.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the
examiner should be directed to AARON MERRIAM whose telephone number is (703) 756-
5938. The examiner can normally be reached M-F 8:00 am - 5:00 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, Jason Sims can be reached on (571)272-4867. 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.
/AARON MERRIAM/Examiner, Art Unit 3791
/MATTHEW KREMER/Primary Examiner, Art Unit 3791
Prosecution Insights