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 .
Specification
The disclosure is objected to because of the following informalities: on page 6 in the description of the First model (attention mapper) states that "In the attention mapper, residual neural network (ResNet) may be used...," omitting the indefinite article "a" before "residual neural network (ResNet)." Appropriate correction is required.
Claim Objections
Claims 2 and 6 are objected to because of the following informalities: claim 2 recites "the feature extraction model is residual neural network (ResNet)" and claim 6 recites "performed using pre-trained residual neural network (ResNet)," each omitting the indefinite article "a" before "residual neural network (ResNet)." Appropriate correction is required.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier.
Claim 1 (and claims 2-4 by dependency): "the feature extraction model [that] receives the ECG data as input and outputs ECG features"
Claimed function: receiving the ECG data as input and outputting ECG features
Corresponding structure: residual neural network (ResNet) formed by stacking multiple 1D convolutional neural network (CNN) layers, combined with Rectified Linear Unit (ReLU) activation functions and batch normalization (BN), in which "n ConV 1D (m)" denotes a 1D CNN operation using n filters of size m, max pooling that compares two values and halves the length by retaining the larger value is used to resize features, and global average pooling computes an average for each channel to output ECG features having a channel length (pg. 6 of specification, "First model (attention mapper)"; FIG. 1(a)-(b); pg. 10 of specification, operation S110), and equivalents thereof
Interpretation: The broadest reasonable interpretation of "feature extraction model" is limited to residual neural network (ResNet) formed by stacking multiple 1D convolutional neural network (CNN) layers, combined with Rectified Linear Unit (ReLU) activation functions and batch normalization (BN), in which "n ConV 1D (m)" denotes a 1D CNN operation using n filters of size m, max pooling that compares two values and halves the length by retaining the larger value is used to resize features, and global average pooling computes an average for each channel to output ECG features having a channel length (pg. 6 of specification, "First model (attention mapper)"; FIG. 1(a)-(b); pg. 10 of specification, operation S110), and equivalents thereof.
Claim 1 (and claims 2-4 by dependency): "the personal identification model [that] receives the ECG features as input and identifies an individual corresponding to the ECG features"
Claimed function: receiving the ECG features as input and identifying an individual corresponding to the ECG features
Corresponding structure: the identification module of the attention mapper, comprising an attention mechanism (e.g., Transformer self-attention per Vaswani et al., "Attention is all you need," NIPS 30 (2017)) that produces an n x n attention distribution matrix whose values sum to 1 in each row, followed by fully connected (FC) layers that produce the personal identification output (pg. 6-7 of specification; FIG. 1(a); pg. 10, operation S110), and equivalents thereof.
Interpretation: The broadest reasonable interpretation of "personal identification model" is limited to the identification module of the attention mapper, comprising an attention mechanism (e.g., Transformer self-attention per Vaswani et al., "Attention is all you need," NIPS 30 (2017)) that produces an n x n attention distribution matrix whose values sum to 1 in each row, followed by fully connected (FC) layers that produce the personal identification output (pg. 6-7 of specification; FIG. 1(a); pg. 10, operation S110), and equivalents thereof.
Claim 1 (and claims 2-4 by dependency): "the arrhythmia classification model [that] receives the ECG features as input and classifies arrhythmia corresponding to the ECG features"
Claimed function: receiving the ECG features as input and classifying arrhythmia corresponding to the ECG features
Corresponding structure: the classification module of the attention mapper, which has the same structure as the identification module -- an attention mechanism (e.g., Transformer self-attention) producing an n x n attention distribution matrix, followed by fully connected (FC) layers that produce the arrhythmia classification output (pg. 6-7 of specification; FIG. 1(a); pg. 10, operation S110), and equivalents thereof.
Interpretation: The broadest reasonable interpretation of "arrhythmia classification model" is limited to the classification module of the attention mapper, which has the same structure as the identification module -- an attention mechanism (e.g., Transformer self-attention) producing an n x n attention distribution matrix, followed by fully connected (FC) layers that produce the arrhythmia classification output (pg. 6-7 of specification; FIG. 1(a); pg. 10, operation S110), and equivalents thereof.
Claim 1 (and claims 2-4 by dependency): "the noise model [that] generates noise with the same length as that of the ECG features"
Claimed function: generating noise with the same length as that of the ECG features
Corresponding structure: the noise model of the noise generator, having a multi-layer perceptron (MLP) structure with three hidden layers and ReLU activation functions, which receives the ECG features extracted by ResNet as input and outputs a vector with the same size as that of the input vector (pg. 7 of specification, "Second model (noise generator)"; FIG. 2; pg. 10, operation S120), and equivalents thereof
Interpretation: The broadest reasonable interpretation of "noise model" is limited to the noise model of the noise generator, having a multi-layer perceptron (MLP) structure with three hidden layers and ReLU activation functions, which receives the ECG features extracted by ResNet as input and outputs a vector with the same size as that of the input vector (pg. 7 of specification, "Second model (noise generator)"; FIG. 2; pg. 10, operation S120), and equivalents thereof
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 5-7 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, at the time the application was filed, had possession of the claimed invention.
Claim 5 recites, in relevant part, the steps of “generating attention distribution for noise” and “generating the noise-added ECG features.” Each of these limitations is recited solely in terms of the result to be achieved. Claim 5 recites no algorithm, operation, model, or other means by which the recited attention distribution for noise is generated, and no algorithm, operation, model, or other means by which the recited noise-added ECG features are generated. Claim 5 further does not require that the attention distribution generated in the third step be used in any manner in generating the noise-added ECG features of the fourth step. Under the broadest reasonable interpretation (MPEP § 2111), claim 5 therefore encompasses the genus of all techniques, performed by any means, that produce an attention distribution for noise and that produce ECG features to which noise has been added.
The specification does not convey possession of that genus. The specification discloses a single species of each of these two steps. With respect to generating attention distribution for noise, the specification discloses only computing the attention distribution ai produced by the attention mechanism of the identification module and the attention distribution ac produced by the attention mechanism of the classification module of a pre-trained “attention mapper,” column-wise summing each n×n attention distribution matrix to form two vectors of size n, subtracting the vector for arrhythmia classification from the vector for personal identification, and applying the softmax function to the result — that is, “the attention distribution may be generated by computing softmax(sumcolumn-wise(ai) - sumcolumn-wise(ac))” (pg. 8 of specification; pg. 11, operation S150, and FIG. 2). With respect to generating the noise-added ECG features, the specification discloses only inputting the extracted ECG features to a noise model having “a multi-layer perceptron (MLP) structure with three hidden layers and ReLU activation functions” that outputs noise of the same length as the ECG features, computing an element-wise product of that noise and the attention distribution to obtain weighted noise, and computing an element-wise sum of the weighted noise and the ECG features (pg. 11, operation S160, and FIG. 2).
The specification does not describe any other species falling within either recited genus, and does not describe any structural feature common to the members of either genus that would permit one of ordinary skill in the art to visualize or recognize the members of the genus. A claim defined entirely by a desired function or result, without a description of the structure or algorithm that achieves that function, is not adequately supported by the disclosure of a single species performing that function. See Ariad Pharmaceuticals, Inc. v. Eli Lilly & Co., 598 F.3d 1336, 1349-51 (Fed. Cir. 2010) (en banc); AbbVie Deutschland GmbH & Co. v. Janssen Biotech, Inc., 759 F.3d 1285, 1300-01 (Fed. Cir. 2014); MPEP § 2163.
Additionally, claim 5 is directed to a computer-implemented method performed by “a computing device comprising at least a processor.” For such a claim, the specification must describe the algorithm by which each claimed function is performed, and the algorithm described must be commensurate in scope with the claimed function. See Vasudevan Software, Inc. v. MicroStrategy, Inc., 782 F.3d 671, 681-83 (Fed. Cir. 2015); MPEP § 2163. The single algorithm disclosed at pages 7-9 and 11 of the specification is not commensurate in scope with the unbounded functional recitations “generating attention distribution for noise” and “generating the noise-added ECG features” of claim 5.
It is acknowledged that the Summary of the specification, at pg. 2, recites the four steps of claim 5 verbatim. That passage, however, merely restates the claim in the same purely functional terms and adds no description of how either recited result is achieved. A restatement of the claimed result is not a written description of the genus that the claim covers, and original claim language reciting only a desired result does not by itself establish possession where the specification fails to describe the means of achieving that result. See Ariad, 598 F.3d at 1349-50; MPEP § 2163.
Claim 6 depends from claim 5 and further limits only the extracting step, reciting that “the extracting of the ECG features is performed using pre-trained residual neural network (ResNet).” That added limitation is described at pages 6 and 11 of the specification. Claim 6 does not, however, further limit either the “generating attention distribution for noise” step or the “generating the noise-added ECG features” step of claim 5. Claim 6 therefore incorporates, and does not cure, the deficiency identified above with respect to claim 5, and is rejected for the same reasons.
Claim 7 depends from claim 5 and recites the specific operations by which “the generating of the attention distribution for noise” is performed, namely generating first and second attention distributions using pre-trained personal identification and arrhythmia classification models, converting those distributions from a matrix form to a vector form, and subtracting the second from the first. Those operations are described at pg. 8 and 11 of the specification, and claim 7 accordingly cures the deficiency identified above as to the third step of claim 5.
Claim 7 does not, however, further limit the step of “generating the noise-added ECG features,” which remains recited solely as a result to be achieved and which, under the broadest reasonable interpretation, continues to encompass the genus of all techniques for producing ECG features to which noise has been added. For the reasons given above with respect to claim 5, the specification's disclosure of the single species at pages 7-9 (steps (4)-(6)) and pg. 11, operation S160 does not reasonably convey possession of that genus. Claim 7 is therefore rejected with respect to that limitation.
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 3-8 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 3 recites the limitation "the noise-added ECG features" in "generating the noise-added ECG features by adding the weighted noise and the ECG features." There is insufficient antecedent basis for this limitation in the claim, as "noise-added ECG features" is not previously recited in claim 3 or in claim 1, from which claim 3 depends. For purposes of examination, "the noise-added ECG features" is interpreted under BRI to mean the ECG features produced by adding the weighted noise to the ECG features extracted by the feature extraction model.
Claim 3 recites, in relevant part, "wherein the training of the noise model comprises training the noise model while freezing the feature extraction model, the personal identification model, and the arrhythmia classification model, and the training of the noise model comprises: …". The claim recites the limitation "the training of the noise model comprises" twice, once to describe the freezing of the three pre-trained models and again to introduce eight enumerated sub-steps, without clarifying the logical relationship between the two clauses. It is unclear whether the freezing limitation is a condition under which the enumerated sub-steps are performed, an additional conjunctive requirement, or a competing, independently sufficient definition of the same claimed training step. One of ordinary skill in the art would not be able to determine the metes and bounds of claim 3 with reasonable certainty. For purposes of examination, the limitation is interpreted under BRI such that the freezing of the feature extraction model, the personal identification model, and the arrhythmia classification model describes the environment in which the eight enumerated sub-steps are performed.
Claim 4 depends from claim 3 and does not cure the antecedent basis deficiency for "the noise-added ECG features" identified above with respect to claim 3. Claim 4 is rejected for the same reasons.
Claim 4 depends from claim 3 and does not cure the ambiguous dual "comprises" construction identified above with respect to claim 3. Claim 4 is rejected for the same reasons.
See the BRI interpretation provided above for claim 3.
Claim 5 recites the limitation "the noise-added ECG features" in its final step, "generating the noise-added ECG features." There is insufficient antecedent basis for this limitation in the claim, as "noise-added ECG features" is not previously recited in claim 5. For purposes of examination, "the noise-added ECG features" is interpreted under BRI to mean the ECG features produced by adding the weighted noise to the ECG features extracted from the target ECG data.
Claim 6 depends from claim 5 and does not cure the antecedent basis deficiency for "the noise-added ECG features" identified above with respect to claim 5. Claim 6 is rejected for the same reasons.
Claim 7 depends from claim 5 and does not cure the antecedent basis deficiency for "the noise-added ECG features" identified above with respect to claim 5. Claim 7 is rejected for the same reasons.
See the BRI interpretation provided above for claim 5.
Claim 7 additionally recites the limitation "the target ECG features" in "inputting the target ECG features to a pre-trained personal identification model" and "inputting the target ECG features to a pre-trained arrhythmia classification model." There is insufficient antecedent basis for this limitation, as neither claim 7 nor parent claim 5 previously recites "target ECG features" — claim 5 recites only "target ECG data" and "ECG features" as separate terms. For purposes of examination, "the target ECG features" is interpreted under BRI to refer to the "ECG features" recited in claim 5 as extracted "corresponding to the target ECG data".
Claim 8 depends from claim 7 and does not cure the antecedent basis deficiencies for "the noise-added ECG features" and "the target ECG features" identified above with respect to claims 5 and 7. Claim 8 is rejected for the same reasons.
See the BRI interpretations provided above for claims 5 and 7.
Claim 8 additionally recites both "the target ECG features" ("generating noise with the same length as that of the target ECG features") and, in the immediately following clause, "the ECG features" ("generating the noise-added ECG features by adding the weighted noise and the ECG features"). The term "the ECG features" has no independent antecedent basis in claim 8 or its parent claims and it is unclear whether it refers to the same entity as "the target ECG features" used earlier in the same claim, rendering the scope of claim 8 uncertain. For purposes of examination, "the ECG features" in the final clause of claim 8 is interpreted under BRI to refer to the same "target ECG features" recited earlier in claim 8 and in parent claim 7.
Appropriate correction is required.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 5 and 6 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 5
Step 1: Claim 5 recites “A method for privacy-preserving electrocardiogram (ECG) data collection, performed by a computing device comprising at least a processor” followed by a series of steps, which falls within the statutory category of a process. See MPEP 2106.03. Accordingly, claim 5 satisfies Step 1 of the eligibility analysis.
Step 2A, Prong 1: Claim 5 is directed to an abstract idea. Specifically, claim 5 recites limitations falling within the mental processes grouping of abstract ideas. The limitations reciting the judicial exception are “extracting ECG features corresponding to the target ECG data”, “generating attention distribution for noise”, and “generating the noise-added ECG features”.
These limitations are recited at the level of a desired result. Claim 5 places no constraint whatsoever on how the ECG features are extracted, how the attention distribution for noise is derived, or how the noise-added features are produced; it recites no model, no architecture, no training operation, and no computational relationship among the three steps. Under the broadest reasonable interpretation (MPEP 2111), each of these steps therefore encompasses an evaluation or judgment that can practically be performed in the human mind or with pen and paper. A person examining an ECG tracing can identify and record its characteristic features; can assign a distribution of weights indicating which portions of those features warrant perturbation; and can add noise values to the recorded feature values. Limitations of this character fall within the mental processes grouping. See MPEP 2106.04(a)(2), subsection III.
It is noted that this characterization does not rest on treating machine-learning operations as mental steps. Consistent with the August 4, 2025 memorandum of Deputy Commissioner Charles Kim, “Reminders on evaluating subject matter eligibility of claims under 35 U.S.C. 101,” the mental processes grouping has not been expanded to encompass limitations that cannot practically be performed in the human mind. Claim 5, unlike claims 1–4 and 6, recites no neural network, no feature extraction model, no personal identification model, no arrhythmia classification model, and no noise model. It is the breadth of claim 5, and not any recitation of artificial intelligence, that places its steps within the mental processes grouping. For the same reason, claim 5 does not merely “involve” an abstract idea; the recited steps set forth the evaluation itself. See MPEP 2106.04, subsection II(A)(1).
Step 2A, Prong 2: The judicial exception is not integrated into a practical application. The additional elements, considered individually and as an ordered combination, do not integrate the judicial exception into a practical application. The additional elements are the preamble recitation of a method “for privacy-preserving electrocardiogram (ECG) data collection”, the recitation that the method is “performed by a computing device comprising at least a processor”, and the step of “receiving target ECG data”.
The specification has been consulted in accordance with MPEP 2106.04(d)(1) and MPEP 2106.05(a), as revised in light of Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision). The first step of that inquiry is satisfied: the specification does disclose an improvement to technology in sufficient detail that one of ordinary skill in the art would recognize it, and the disclosure is not a bare conclusory assertion. The specification identifies a specific technical problem, explaining that personal identification “may be prevented by simply adding noise to ECG data,” but that “if noise is added to the entire data, usability such as arrhythmia classification may be lowered,” so that “noise needs to be added to ECG data such that the added noise only interferes with personal identification and does not interfere with arrhythmia classification” (pg. 2 of specification, Description of Related Art). The specification then discloses a specific technical solution: the attention distribution produced by each of a pre-trained personal identification model and a pre-trained arrhythmia classification model is an “n×n matrix” that “indicates which parts of the ECG features are significant for identification and classification, respectively” (pg. 7 of specification); each matrix is column-wise summed into a vector, and the classification vector is subtracted from the identification vector, such that “a value is larger in a part which classification does not focus while the identification focuses,” yielding a distribution representing “a part in which noise may be added to disturb only identification performance without degrading classification performance” (pg. 8 of specification, Second model (noise generator)).
The second step of the Desjardins inquiry, however, is not satisfied for claim 5, because the claim itself does not reflect the disclosed improvement. The improvement described in the specification resides in the particular manner of deriving the noise-steering distribution from the difference between the two task-specific attention distributions, and in weighting the model-generated noise by that distribution before adding it to the features. Claim 5 recites none of those components or steps. It recites only that an “attention distribution for noise” is generated and that “noise-added ECG features” are generated, without reciting the personal identification model, the arrhythmia classification model, the matrix-to-vector conversion, the subtraction operation, the noise model, the weighting operation, or any functional constraint tying the added noise to the preservation of arrhythmia classification accuracy.
The consequence is confirmed by the specification itself. Because claim 5 imposes no constraint on how the distribution is derived or how the noise is applied, its broadest reasonable interpretation encompasses the very approach the specification disparages, namely “adding random noise to ECG features,” which the specification states “disrupts personal identification, but may also degrade performance of the arrhythmia classification,” so that “usefulness of data is reduced due to the added noise” (pg. 5 of specification, Detailed Description). A claim whose full scope encompasses the deficient approach that the disclosed invention was designed to overcome cannot be said to reflect the improvement over that approach. This is the circumstance addressed in Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316 (Fed. Cir. 2016), where the claims did not contain any limitations addressing the improvements the patentee asserted. Relatedly, claim 5 recites only the idea of a solution or outcome rather than a particular way of achieving that outcome, which weighs against the improvement consideration. See McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1314–15 (Fed. Cir. 2016); MPEP 2106.05(a). It is further noted that the specific mechanism which does reflect the disclosed improvement is recited in dependent claims 7 and 8; the presence of that mechanism in those claims confirms its absence from claim 5.
Turning to the individual additional elements, the recitation that the method is “performed by a computing device comprising at least a processor” amounts to mere instructions to implement the abstract idea on a generic computer, which does not integrate the exception into a practical application. See MPEP 2106.05(f). The specification confirms the generic character of this element, stating that “the computing device may include a personal computer (PC), a server, a laptop computer, a tablet PC, a notebook, a smartphone, a smart watch, a head mounted device (HMD), a smart ring, and smart glasses” (pg. 9 of specification), and further that the described device “can be implemented by using one or more general-purpose computer or designated computer, examples of which include a processor, a controller, an ALU (arithmetic logic unit), a digital signal processor, a microcomputer, an FPGA (field programmable gate array), a PLU (programmable logic unit), a microprocessor, and any other device capable of executing and responding to instructions” (pg. 13 of specification). Because any general-purpose computing device suffices, the claim does not tie the exception to a particular machine that imposes meaningful limits. See MPEP 2106.05(b).
The step of “receiving target ECG data” is mere data gathering that is necessary antecedent activity to the recited exception, and therefore constitutes insignificant extra-solution activity. See MPEP 2106.05(g). The specification confirms that this element is nothing more than conventional data acquisition, stating that “[t]he target ECG data may be received from an external device through a wired/wireless communication network or may also be received from an ECG measurement device implemented in an ECG collection device” (pg. 10 of specification).
In addition, the preamble recitation of “privacy-preserving electrocardiogram (ECG) data collection” merely confines the abstract idea to a particular technological environment and field of use, and imposes no meaningful limit on how the exception is applied. See MPEP 2106.05(h). No transformation or reduction of a particular article to a different state or thing is effected, as the claim manipulates data only. See MPEP 2106.05(c). Considered individually and as an ordered combination, the additional elements do no more than apply the abstract idea using a generic computer upon data obtained by conventional means. Accordingly, claim 5 does not satisfy Step 2A, Prong 2.
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception. With respect to “a computing device comprising at least a processor”, the specification itself supplies an express statement of the well-understood, routine, and conventional nature of this element, describing the invention as implementable on “one or more general-purpose computer or designated computer, examples of which include a processor, a controller, an ALU (arithmetic logic unit), a digital signal processor, a microcomputer, an FPGA (field programmable gate array), a PLU (programmable logic unit), a microprocessor, and any other device capable of executing and responding to instructions” (pg. 13 of specification). See MPEP 2106.05(d)(I)(a). The use of a generic computer to perform generic computer functions has likewise been held insufficient to supply an inventive concept. Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 225–26 (2014).
With respect to “receiving target ECG data”, receiving or transmitting data over a network has been identified by the courts as a well-understood, routine, and conventional computer function. See Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1321 (Fed. Cir. 2016); TLI Communications LLC v. AV Automotive LLC, 823 F.3d 607, 610 (Fed. Cir. 2016); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355 (Fed. Cir. 2014); MPEP 2106.05(d)(II)(i). See MPEP 2106.05(d)(I)(b). The broadest reasonable interpretation of this limitation is not confined to receipt over a network, however, and additionally encompasses acquisition from a locally implemented sensor. The specification establishes that both modes are conventional and that neither is modified, stating that “[t]he target ECG data may be received from an external device through a wired/wireless communication network or may also be received from an ECG measurement device implemented in an ECG collection device” (pg. 10 of specification). See MPEP 2106.05(d)(I)(a). Mere data gathering of this character, performed by any known means, is well-understood, routine, and conventional. See Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 79 (2012) (step of measuring metabolite levels “using any known process” was well-understood, routine, conventional activity previously engaged in by the scientific community); MPEP 2106.05(d).
Considered as an ordered combination, receiving ECG data over a conventional interface and thereafter performing the recited abstract steps on a general-purpose processor adds nothing that is not already present when the elements are considered separately. The elements do not interact in any unconventional manner; the computer is invoked merely as a tool to carry out the exception. Accordingly, claim 5 does not satisfy Step 2B and is rejected under 35 U.S.C. 101.
Claim 6
Step 1: Claim 6 depends from claim 5 and recites a method, which falls within the statutory category of a process. See MPEP 2106.03. Accordingly, claim 6 satisfies Step 1 of the eligibility analysis.
Step 2A, Prong 1: No further abstract idea limitations are recited.
Step 2A, Prong 2: Claim 6 recites the following additional element, “wherein the extracting of the ECG features is performed using pre-trained residual neural network (ResNet)”.
The judicial exception is not integrated into a practical application. For the same reasons as discussed in the Step 2A, Prong 2 analysis of claim 5, the recitation of a computing device comprising at least a processor and the step of receiving target ECG data do not integrate the judicial exception into a practical application. See MPEP 2106.05(f), (g), (h).
For the recited “pre-trained residual neural network (ResNet)”, the Desjardins improvement inquiry has again been applied. As set forth in the analysis of claim 5, the specification does disclose an improvement to technology; however, that improvement resides in the derivation of the noise-steering attention distribution from the difference between the identification and classification attention distributions, and in the weighting of the generated noise by that distribution, and not in the choice of feature extractor. The specification expressly disclaims any criticality to the recited ResNet, stating that “the scope of the present invention is not limited thereto and, depending on example embodiments, an arbitrary feature extraction technique may be used or an arbitrary artificial neural network (ANN) model may be used” (pg. 6 of specification, First model (attention mapper)). The specification further describes ResNet as having been selected for its already-known property, explaining that “residual neural network (ResNet) may be used to prevent degradation in learning performance due to a vanishing gradient or exploding gradient issue” (pg. 6 of specification). ResNet is thus employed for the purpose it was already understood to serve, rather than being configured or modified in any manner that produces the disclosed improvement.
Claim 6 accordingly does not reflect the disclosed improvement. Substituting a known feature extractor into the results-based framework of claim 5 leaves the two operative steps in which the improvement actually resides — the generation of the attention distribution for noise and the generation of the noise-added ECG features — recited as bare outcomes, and claim 6 therefore continues to claim only the idea of a solution rather than a particular way of achieving it. See McRO, 837 F.3d at 1314–15; Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316 (Fed. Cir. 2016). The recitation of ResNet instead amounts to invoking a known computational tool to carry out one step of the abstract idea, which does not integrate the exception into a practical application. See MPEP 2106.05(f). Nor is the recited ResNet a particular machine within the meaning of MPEP 2106.05(b); it is a software model executable on any general-purpose processor of the type the specification describes at page 13.
The specification affirmatively disclaims any criticality to the recited ResNet and locates the disclosed improvement elsewhere; the two limitations in which that improvement resides remain recited in claim 6 as bare results; and the claim as a whole, evaluated as an ordered combination and without oversimplification, covers the idea of a solution rather than a particular way of achieving it. Accordingly, claim 6 does not satisfy Step 2A, Prong 2.
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception. For the same reasons as discussed in the Step 2B analysis of claim 5, the recitation of a computing device comprising at least a processor and the step of receiving target ECG data are well-understood, routine, and conventional, as evidenced by the specification at pg. 13 and 10 respectively, and by Alice Corp., 573 U.S. at 225–26, TLI Communications, 823 F.3d at 610, and Mayo, 566 U.S. at 79.
With respect to the “pre-trained residual neural network (ResNet)”, the specification supplies an express statement establishing its well-understood, routine, and conventional nature. The specification treats ResNet as an off-the-shelf and freely interchangeable component, stating that “an arbitrary feature extraction technique may be used or an arbitrary artificial ne ural network (ANN) model may be used” (pg. 6 of specification), and describes its constituent operations — stacking “multiple 1D convolutional neural network (CNN) layers,” combining the structure with “Rectified Linear Unit (ReLU) activation function and batch normalization (BN),” employing “max pooling” to resize features, and applying “global average pooling” to “compute average for each channel” — as known techniques applied in their ordinary manner (pg. 6 of specification). The specification further states that ResNet “may be used to prevent degradation in learning performance due to a vanishing gradient or exploding gradient issue” (pg. 6 of specification), confirming that the component is employed for a property already recognized in the art as of the effective filing date. See MPEP 2106.05(d)(I)(a).
Applying a known, pre-trained neural network to perform one step of an abstract idea does not furnish an inventive concept. Considered as an ordered combination, receiving ECG data, extracting features by means of a conventional pre-trained ResNet, and thereafter performing the remaining abstract steps on a general-purpose processor adds nothing beyond the sum of the individual parts, and the elements do not interact in any unconventional way. Accordingly, claim 6 does not satisfy Step 2B and is rejected under 35 U.S.C. 101.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim 1 is rejected under 35 USC 103 as being unpatentable over ECG Biosignal Deidentification Using Conditional Generative Adversarial Networks to Jafarlou et al. (hereinafter Jafarlou) in view of Mobile Sensor Data Anonymization to Malekzadeh et al. (hereinafter Malekzadeh), further in view of ResNet-Attention model for human authentication using ECG signals to Hammad et al. (hereinafter Hammad), and further in view of A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics to Osia et al. (hereinafter Osia).
Jafarlou discloses A method for privacy-preserving electrocardiogram (ECG) data collection performed by a computing device comprising at least a processor, the method comprising (Jafarlou: Abstract…Jafarlou proposes a computational framework that transforms an ECG signal so that the individual who produced it can no longer be identified while the clinically relevant content of that signal is retained, which constitutes a method for privacy-preserving electrocardiogram data collection, "In this paper, to address this privacy protection gap, we propose a Generative Adversarial Network (GAN)-based framework for de-identification of ECG signals"):
…the personal identification model receives the ECG features as input and identifies an individual corresponding to the ECG features (Jafarlou: Abstract…Jafarlou premises its framework on the established capability of a model to determine which individual a given ECG originated from, which constitutes the personal identification model … identifies an individual corresponding to the ECG features limitation, "Recent works have shown the feasibility of identifying and authenticating individuals by using ECG as a biometric due to the highly individualized nature of ECG signals"),
the arrhythmia classification model receives the ECG features as input and classifies arrhythmia corresponding to the ECG features,… (Jafarlou: Abstract…Jafarlou trains its de-identifying generator under a loss that expressly retains the information bearing on the subject's cardiac condition, so that a model classifying that cardiac condition from the retained content constitutes the arrhythmia classification model limitation, "We leverage a combination of a standard GAN loss, an Ordinary Differential Equations (ODE)-based, and identity-based loss values to train a generator that de-identifies a ECG signal while preserving structure the ECG signal and information regarding the target cardio vascular condition").
Jafarlou does not expressly disclose, but Malekzadeh does teach:
training a feature extraction model, a personal identification model, and an arrhythmia classification model using learning data (Malekzadeh: p. 4…Malekzadeh trains an encoder jointly against a pre-trained identity classifier and a pre-trained utility-task classifier on common labelled training data, which constitutes training a feature extraction model, a personal identification model, and an arrhythmia classification model using learning data, "La and Ld are utility losses that can be customized based on the app requirements (note that Ld is the only available utility loss if there is no target application), whereas Li is an identity loss that helps the AAE remove user-specific signals")…
training a noise model, (Malekzadeh: p. 1…Malekzadeh trains a dedicated network whose function is to emit a perturbed version of its input under a multi-objective loss, which constitutes training a noise model, "We formulate the anonymization problem using an information-theoretic approach and propose a new multi-objective loss function for training deep autoencoders")
wherein the feature extraction model receives the ECG data as input and outputs ECG features, (Malekzadeh: p. 3…Malekzadeh's encoder takes the raw input data and emits a low-dimensional latent representation of that data, which constitutes the feature extraction model receives the ECG data as input and outputs ECG features limitation, "The Encoder maps X into an identity concealing low-dimensional latent representation Y by getting feedback from a pre-trained classifier, the Encoder Regularizer, which penalizes the Encoder if it captures information corresponding to U into Y"), …
Jafarlou combined with Malekzadeh does not expressly disclose, but Hammad does teach:
…learning data that includes ECG data with a predetermined length (Hammad: Abstract…Hammad trains its ECG models on segments cut to one fixed duration, which constitutes learning data that includes ECG data with a predetermined length, "We have used 2-s duration ECG signals obtained from two ECG databases (Physikalisch-Technische Bundesanstalt [PTB] and Check Your Bio-signals Here initiative [CYBHi]) for authentication"); and …
Jafarlou combined with Malekzadeh-Hammad does not expressly disclose, but Osia does teach:
the noise model generates noise with the same length as that of the ECG features (Osia: p. 4…Osia adds noise directly to the feature vector emitted by the feature extractor, so the added noise is necessarily of the same dimension as that feature vector, which constitutes the noise model generates noise with the same length as that of the ECG features limitation, "Apart from Siamese fine-tuning and dimensionality reduction, the feature extractor module can also add multi-dimensional noise to the feature vector to further increase the privacy").
Jafarlou, Malekzadeh, Hammad and Osia are analogous art because they are all within the same field of endeavor, specifically the machine-learning processing of personally-collected physiological and sensor data under a privacy constraint. They address the same problem solving area of suppressing the identity information such data carries while preserving the information a downstream diagnostic or recognition task depends upon.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to train the ECG de-identification framework of Jafarlou using the multi-objective identity-versus-utility scheme of Malekzadeh, on fixed-duration ECG segments as Hammad collects them, and to inject the resulting noise at the feature vector as Osia teaches.
The suggestion/motivation for doing so would have been provided by Osia, which teaches that noise may be applied to the feature vector itself as one of an enumerated set of privacy mechanisms available to the feature extractor, "Apart from Siamese fine-tuning and dimensionality reduction, the feature extractor module can also add multi-dimensional noise to the feature vector to further increase the privacy" (Osia: p. 4), and by Hammad, which fixes ECG input to a single segment duration so that the feature vector has a stable length for such noise to match, "We have used 2-s duration ECG signals obtained from two ECG databases (Physikalisch-Technische Bundesanstalt [PTB] and Check Your Bio-signals Here initiative [CYBHi]) for authentication" (Hammad: Abstract). This is the application of a known technique to a known method ready for improvement to yield predictable results, the rationale of MPEP § 2143(I)(D).
Claim 2 is rejected under 35 USC 103 as being unpatentable over Jafarlou in view of Malekzadeh, further in view of Hammad and Osia, as applied in the rejection of claim 1 above, and further in view of HADLN: Hybrid Attention-Based Deep Learning Network for Automated Arrhythmia Classification to Jiang et al. (hereinafter Jiang).
Jafarlou combined with Malekzadeh combined with Hammad combined with Osia discloses claim 1.
Hammad further teaches wherein the feature extraction model is residual neural network (ResNet), and the personal identification model…are attention networks (Hammad: Abstract…Hammad builds an ECG model in which a residual convolutional network supplies the features and an attention mechanism performs the identification of the individual, which constitutes the feature extraction model is residual neural network (ResNet) and the personal identification model being an attention network, "In the first model, a convolutional neural network (CNN) is developed and in the second model, a residual convolutional neural network (ResNet) with attention mechanism called ResNet-Attention is designed for human authentication")…
Jafarlou combined with Malekzadeh, Hammad and Osia does not expressly disclose, but Jiang does teach:
…the arrhythmia classification model are attention networks having the same structure (Jiang: p. 3…Jiang classifies arrhythmia with a network that fuses an attention mechanism onto a residual network, which constitutes the arrhythmia classification model being an attention network, "the HADLN architecture was proposed to automatically detect atrial fibrillation based on the fusion of attention mechanism and deep learning model, which combines ResNet, Bi-LSTM, and attention mechanism module"; p. 5…Jiang further reduces that arrangement to a residual network carrying an attention mechanism applied to ECG classification, the same residual-plus-attention structure Hammad applies to identification, which constitutes the two models being attention networks having the same structure, "As a comparison, the ResNet model with attention mechanism, termed as ResNet_A method, is proposed for ECG classification").
Hammad and Jiang are analogous art with the references of record because both are within the same field of endeavor, specifically the deep-learning analysis of electrocardiogram signals, and both address the problem of selecting which parts of an ECG feature representation a task-specific model should weight.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to implement the identification head as the residual-plus-attention network of Hammad and the arrhythmia head as the residual-plus-attention network of Jiang, yielding two task heads of the same structure, as claim 2 recites.
The suggestion/motivation for doing so would have been provided by Jiang, which reports that adding the attention mechanism to a residual network measurably improves ECG classification over the residual network alone, "As a comparison, the ResNet model with attention mechanism, termed as ResNet_A method, is proposed for ECG classification" (Jiang: p. 5), giving a PHOSITA a concrete reason to adopt for the arrhythmia head the same residual-plus-attention structure Hammad had already adopted for the identification head. This is the use of a known technique to improve similar devices in the same way, the rationale of MPEP § 2143(I)(C).
Claims 5 and 6 are rejected under 35 USC 103 as being unpatentable over Jafarlou in view of Malekzadeh, Hammad, Osia, Jiang, and further in view of Task-Specific Adaptive Differential Privacy Method for Structured Data to Utaliyeva et al. (hereinafter Utaliyeva).
Per claim 5, Jafarlou discloses A method for privacy-preserving electrocardiogram (ECG) data collection, performed by a computing device comprising at least a processor, the method comprising:
receiving target ECG data (Jafarlou: Abstract…Jafarlou takes in a subject’s recorded ECG signal and puts it through a computational framework that removes the subject’s identity from that signal while retaining its clinically relevant content, which constitutes a method for privacy-preserving electrocardiogram data collection that receives target ECG data, "In this paper, to address this privacy protection gap, we propose a Generative Adversarial Network (GAN)-based framework for de-identification of ECG signals");…
Jafarlou does not expressly disclose, but Malekzadeh does teach:
…extracting ECG features corresponding to the target ECG data (Malekzadeh: p. 3, § 3.2…Malekzadeh’s encoder takes the raw recorded signal as its input and emits a low-dimensional latent representation of that signal, which constitutes extracting ECG features corresponding to the target ECG data, "The Encoder maps X into an identity concealing low-dimensional latent representation Y by getting feedback from a pre-trained classifier, the Encoder Regularizer, which penalizes the Encoder if it captures information corresponding to U into Y");…
Jafarlou combined with Malekzadeh does not expressly disclose, but Osia does teach:
…generating the noise-added ECG features (Osia: p. 4, § 3.3…Osia’s feature extractor adds noise across the dimensions of the feature vector it emits, so the vector it forwards is the extracted feature representation with noise added to it, "Apart from Siamese fine-tuning and dimensionality reduction, the feature extractor module can also add multi-dimensional noise to the feature vector to further increase the privacy"; p. 5, § 3.3…Osia performs that addition on the feature vector at inference time before the representation leaves the user’s device, "Before uploading to the cloud, the feature extractor adds some noise to the reduced-size feature to obtain the exclusive feature").
Jafarlou combined with Malekzadeh and Osia does not expressly disclose, but with Hammad combined with Utaliyeva does teach:
…generating attention distribution for noise; and… (Hammad: Abstract…Hammad passes the features emitted by its residual network through an attention mechanism, which assigns a weight to each element of those features and so produces an attention distribution over the ECG features, "In the first model, a convolutional neural network (CNN) is developed and in the second model, a residual convolutional neural network (ResNet) with attention mechanism called ResNet-Attention is designed for human authentication"; Utaliyeva: § 4.3.4…Utaliyeva takes a per-element relevance distribution of that kind and makes it the quantity that governs how much noise each element of the representation receives, which is what makes the attention distribution one generated for noise, "In the adaptive sensitivity step, we calibrate the amount of random noise according to the feature importance of each attribute Δi"; § 1…Utaliyeva states that those per-element weightings set the distribution over which the noise is apportioned, "The main idea of the proposed DP method is to adaptively calibrate the amount and distribution of random noise applied to each attribute according to the feature importance for the specific tasks of ML models and different types of data").
Jafarlou, Malekzadeh, Hammad, Osia and Utaliyeva are analogous art because they are from a similar problem solving area in the release of a machine-learned representation of personally collected data under a privacy constraint. They address the same problem of suppressing the identity information such a representation carries while preserving the information a downstream diagnostic task depends upon.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to take the ECG signal received by the de-identification framework of Jafarlou, extract a feature representation of it with the encoder of Malekzadeh, compute a per-element weighting over that representation for the purpose of apportioning noise as Utaliyeva teaches, and emit the representation with that noise added to it as Osia teaches.
The suggestion/motivation for doing so would have been provided by Osia, which teaches that noise applied to the feature vector is an available privacy mechanism of the feature extractor itself, "Apart from Siamese fine-tuning and dimensionality reduction, the feature extractor module can also add multi-dimensional noise to the feature vector to further increase the privacy" (Osia: p. 4, § 3.3), and by Utaliyeva, which teaches that noise so applied should be apportioned across the elements of the representation by the relevance each element carries for the task to be preserved rather than spread uniformly, "The main idea of the proposed DP method is to adaptively calibrate the amount and distribution of random noise applied to each attribute according to the feature importance for the specific tasks of ML models and different types of data" (Utaliyeva: § 1). The running combination already computes such per-element relevance over ECG features, because Hammad passes its residual-network ECG features through an attention mechanism that assigns a weight to each of them, "In the first model, a convolutional neural network (CNN) is developed and in the second model, a residual convolutional neural network (ResNet) with attention mechanism called ResNet-Attention is designed for human authentication" (Hammad: Abstract), so a PHOSITA had the attention weights Utaliyeva’s apportionment calls for already in hand. Furthermore, this is the application of a known technique to a known method ready for improvement to yield predictable results, the rationale of MPEP § 2143(I)(D).
Per claim 6, Jafarlou combined Malekzadeh, Hammad, Osia and Utaliyeva discloses claim 5.
Hammad further teaches wherein the extracting of the ECG features is performed using pre-trained residual neural network (ResNet). (Hammad: Abstract…Hammad extracts the features on which its ECG task operates with a residual convolutional neural network that is trained in advance of the task being performed, which constitutes the extracting of the ECG features being performed using pre-trained residual neural network (ResNet), "In the first model, a convolutional neural network (CNN) is developed and in the second model, a residual convolutional neural network (ResNet) with attention mechanism called ResNet-Attention is designed for human authentication"). The rationale to combine Hammad with Jafarlou is the same as the parent claim.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ALAN CHEN whose telephone number is (571)272-4143. The examiner can normally be reached M-F 10-7.
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, Kamran Afshar can be reached at (571) 272-7796. 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.
/ALAN CHEN/Primary Examiner, Art Unit 2125
Prosecution Insights