DETAILED ACTION
The Applicant’s response, received 19 December 2025, has 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.
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 .
Status of the Claims
Claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are pending.
Claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are rejected.
Priority
This application is a CON of PCT/US2019/022712, filed 18 March 2019, which claims benefit of U.S. provisional application 62/646,784, filed 22 March 2018, and claims benefit of U.S. provisional application 62/664,620, filed 30 April 2018.
Therefore, the effective filing date of the claimed invention is 22 March 2018.
Claim Interpretation
The claim interpretations in the Office action mailed 24 September 2025 are maintained in view of the amendment received 19 December 2025.
Claim 41 recites the limitation “the recurrent neural network is trained on carrier statuses determined by at least one of human analysis (call review) or orthogonal experimental measurements that include long-range PCR and sequencing.” This limitation is interpreted to require the selection of a data set of carrier statuses for use in training a recurrent neural network, and not a wet lab process that generates the data. This limitation is further interpreted as a product-by-process limitation, with the product being the carrier statutes, but not requiring the processing steps of producing the product (e.g., determining by human analysis or experimental measurements).
Claim 1 recites the limitation “anomaly detection model.” This limitation is interpreted to mean an algorithm used to predict whether sequence data is outlier data (Specification, ¶ [0029]).
Claim Objections
The objection to claim 1 in the Office action mailed 24 September 2025 is withdrawn in view of the amendment received 19 December 2025.
Claim Rejections - 35 USC § 112
The rejection of claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, in the Office action mailed 24 September 2025 is withdrawn in view of the amendment received 19 December 2025, as noted below.
The rejection of claim 1 for being indefinite for reciting “training…using the plurality of data arrays, arranged based on their relative position within the given gene…each data arrays’ relative position within the given gene” is withdrawn in view of the amendment received 19 December 2025.
The rejection of claim 38 for being indefinite for reciting the limitation “the copy number data and the SNP data are represented as a single data structure inputted to the recurrent neural network” is withdrawn in view of further consideration.
The Applicant’s amendment received 19 December 2025 has been fully considered, however after further consideration, new grounds of rejection are raised under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, in view of the amendment.
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, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 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 is indefinite for reciting “wherein arranging the training set within the plurality of data arrays according to the order…” because the claim previously recites “generating a training set comprising a plurality of sequences of training copy number data and training SNP data” and “arranging, within the training set, each data array” and therefore it is not clear at the training step why the claim recites that the training set is arranged within the plurality of data arrays.
Claims 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are indefinite for depending from claim 1 and failing to remedy the indefiniteness of claim 1.
Claim 1 is further indefinite for reciting “wherein arranging the training set within the plurality of data arrays according to the order causes sequential memory access during training and execution” because it is not clear as to whether “sequential memory access” is referring to hardware memory, i.e., a physical, tangible component that stores data (e.g., RAM) or else software memory, i.e., an intangible, logical organization of data and instructions (e.g., Long Short-Term Memory (LSTM) implemented as a software-based technique, i.e., an algorithm/neural network architecture). This limitation is interpreted to mean a software-based memory technique, i.e., an algorithm/neural network architecture, and not a hardware memory. It is noted here that a recurrent neural network (RNN) is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory” i.e., a hidden state (memory) where, generally, at each time step t, the RNN takes input xt and the previous hidden state ht-1 to update its current memory ht.
Claims 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are indefinite for depending from claim 1 and failing to remedy the indefiniteness of claim 1.
Claim Rejections - 35 USC § 101
The rejection of claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 under 35 U.S.C. 101 in the Office action mailed 24 September 2025 is maintained with modification in view of the amendment received 19 December 2025.
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 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea and a law of nature without significantly more. The claims recite: (a) mathematical concepts, (e.g., mathematical relationships, formulas or equations, mathematical calculations); (b) mental processes, i.e., concepts performed in the human mind, (e.g., observation, evaluation, judgement, opinion); and (c) a law of nature (e.g., naturally occurring relationships).
Subject matter eligibility evaluation in accordance with MPEP 2106.
Eligibility Step 1: Step 1 of the eligibility analysis asks: Is the claim to a process, machine, manufacture or composition of matter?
Claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are directed to a method (i.e., a process).
Therefore, these claims are encompassed by the categories of statutory subject matter, and thus, satisfy the subject matter eligibility requirements under step 1.
[Step 1: YES]
Eligibility Step 2A: First it is determined in Prong One whether a claim recites a judicial exception, and if so, then it is determined in Prong Two whether the recited judicial exception is integrated into a practical application of that exception.
Eligibility Step 2A Prong One: In determining whether a claim is directed to a judicial exception, examination is performed that analyzes whether the claim recites a judicial exception, i.e., whether a law of nature, natural phenomenon, or abstract idea is set forth or described in the claim.
Independent claim 1 recites the following steps which fall within the mental processes and/or mathematical concepts groupings of abstract ideas:
generating a training set comprising a plurality of sequences of training copy number data and training SNP data, a first subset of the plurality of sequences indicating a first carrier status outnumbered by a second subset of the plurality of sequences not indicating the first carrier status (i.e., mental processes and mathematical concepts);
segmenting the training set into a plurality of arrays (i.e., mental processes);
arranging, within the training set, each data array that represents the copy number data according to a sequential characteristic that is in an order that corresponds to its relative position within a given gene in a genome (i.e., mental processes);
training a recurrent neural network using the plurality of data arrays, wherein each data array comprises data that is arranged based on their relative position within the given gene, to generate carrier statuses using the training set that is arranged in accordance with each data arrays’ relative position within the given gene, wherein arranging the training set within the plurality of data arrays according to the order causes sequential memory access during training and execution (i.e., mathematical concepts);
generating an input vector representing SNP or copy number data for that individual (i.e., mental processes and mathematical concepts);
executing an anomaly detection model configured to determine outlier data corresponding to sequencing data or a genetic variation (i.e., mental processes and mathematical concepts);
determining an absence of outlier data (i.e., mental processes); and
responsive to determining an absence of outlier data, generate the respective carrier status of the individual (i.e., mental processes and mathematical concepts).
Independent claim 1 further recites a law of nature by associating an individual’s genomic data (copy number and single nucleotide polymorphism (SNP)) with a phenotype (carrier status), i.e., a genotype-phenotype correlation (MPEP 2106.04(b)).
Dependent claims 6, 7, 10, 11, 12, 13, 14, 20, 21, 36, 37, 39-41, and 136-138 further recite the following steps which fall within the mental processes and/or mathematical concepts groupings of abstract ideas, as noted below.
Dependent claim 6 further recites:
the genome comprises a gene associated with the respective carrier status and/or a pseudogene corresponding to the gene (i.e., mental processes); and
the copy number data is indicative of a number of copies of genetic material corresponding to the gene and/or the pseudogene that are detected during sequencing of the genome at a plurality of locations across the genome (i.e., mental processes).
Dependent claim 7 further recites:
the SNP data is indicative of a number of sequencing reads from the gene that have a single nucleotide polymorphism relative to a reference sequence at one or more locations across the gene (i.e., mental processes).
Dependent claim 10 further recites:
determining a plurality of carrier statuses of the individual, including the respective carrier status (i.e., mental processes and mathematical concepts).
Dependent claim 11 further recites:
the copy number data comprises one or more copy number values; the SNP data comprises one or more SNP values (i.e., mental processes).
Dependent claim 12 further recites:
each input of the recurrent neural network associated with the one or more copy number values corresponds to a sequencing probe used to sequence the gene (i.e., mental processes).
Dependent claim 13 further recites:
each input of the recurrent neural network associated with the one or more SNP values corresponds to a location in the gene associated with one or more probes used to sequence the gene (i.e., mental processes);
Dependent claim 14 further recites:
in accordance with a determination that the respective carrier status is determined with a confidence level below a confidence level threshold, flagging the determined respective carrier status for review (i.e., mental processes and mathematical concepts); and
in accordance with a determination that the respective carrier status is determined with a confidence level above the confidence level threshold, forgoing flagging the determined respective carrier status for review (i.e., mental processes and mathematical concepts).
Dependent claim 20 further recites:
at least one of the copy number data for a given gene or for a given pseudogene or the SNP data for a given gene or for a given pseudogene are represented as at least one data structure having an ordering characteristic (i.e., mental processes and mathematical concepts); and
at least one of the copy number data or the SNP data are ordered within the data structure in an order that corresponds to a sequence of genetic material in the gene (i.e., mental processes and mathematical concepts).
Dependent claim 21 further recites:
the copy number data includes a first copy number data value corresponding to a first position in the gene, and a second copy number data value corresponding to a second position in the gene, the first position in the genome having a relative position with respect to the second position in the gene (i.e., mental processes);
the first copy number data value has the relative position with respect to the second copy number data value in the data structure (i.e., mental processes);
the SNP data includes a first SNP data value corresponding to the first position in the gene, and a second SNP data value corresponding to the second position in the gene, the first position in the genome having a relative position with respect to the second position in the gene (i.e., mental processes); and
the first SNP data value has the relative position with respect to the second SNP data value in the data structure (i.e., mental processes).
Dependent claim 36 further recites:
determining a plurality of carrier statuses of the individual, including the respective carrier status (i.e., mental processes and mathematical concepts); and
the plurality of carrier statuses is represented as a data structure having an ordering characteristic, and the plurality of carrier statuses is ordered within the data structure in an order that corresponds to a sequence of genetic material in the genome (i.e., mental processes and mathematical concepts).
Dependent claim 37 further recites:
the plurality of carrier statuses includes a first carrier status associated with a first position in the genome, and a second carrier status associated with a second position in the genome, the first position in the genome having a relative position with respect to the second position in the genome (i.e., mental processes); and
the first carrier status has the relative position with respect to the second carrier status in the data structure (i.e., mental processes).
Dependent claim 39 further recites:
determining the respective carrier status of an individual is further based on copy number data for a pseudogene in the genome of the individual corresponding to the gene, and SNP data for the pseudogene (i.e., mental processes).
Dependent claim 40 further recites:
the gene and the pseudogene are selected from the group consisting of CYP21A2/CYP21A1P, GBA/psGBA, PMS2/PMS2CL and SMN1/SMN2 (i.e., mental processes).
Dependent claim 41 further recites:
the recurrent neural network is trained on carrier statuses determined by at least one of human analysis (call review) or orthogonal experimental measurements that include long-range PCR and sequencing (i.e., mathematical concepts).
Dependent claim 136 further recites:
the training set comprises at least one thousand samples (i.e., mental processes).
Dependent claim 137 further recites:
an ordering of a carrier status array of at least one sample of the training set corresponds to an ordering of a sequence of training copy number data in the at least one sample (i.e., mental processes).
Dependent claim 138 further recites:
training the recurrent neural network is further based on a weighted cross-entropy loss function (i.e., mathematical concepts).
The abstract ideas recited in the claims are evaluated under the broadest reasonable interpretation (BRI) of the claim limitations when read in light of and consistent with the specification. As noted in the foregoing section, the claims are determined to contain limitations that can practically be performed in the human mind with the aid of a pen and paper (e.g., generating training sets of data can be done with pen and paper; and arranging data into arrays and/or vectors according to a sequential characteristic that is in an order that corresponds to its relative position within a given gene in a genome, can be done with pen and paper), and therefore recite judicial exceptions from the mental process grouping of abstract ideas. Additionally, the recited limitations that are identified as judicial exceptions from the mathematical concepts grouping of abstract ideas (e.g., Specification, ¶¶ [0032] - [0034], and FIGS. 4A & 4B describe using machine learning algorithms in making carrier status determinations based on SNP and copy number data, in particular ¶ [0033] which shows equations used in the long-short term memory (LSTM) cells in each layer of the neural network, and weight matrix and bias vector parameters that can be learned during training; and ¶ [0034] which shows the utilization of weighted cross-entropy loss functions when training an RNN) are abstract ideas irrespective of whether or not the limitations are practical to perform in the human mind. Furthermore, a law of nature correlating a genotype-phenotype association is identified at Eligibility Step 2A Prong One.
Therefore, claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 recite an abstract idea and a law of nature.
[Step 2A Prong One: YES]
Eligibility Step 2A Prong Two: In determining whether a claim is directed to a judicial exception, further examination is performed that analyzes if the claim recites additional elements that when examined as a whole integrates the judicial exception(s) into a practical application (MPEP 2106.04(d)). A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. The claimed additional elements are analyzed to determine if the abstract idea is integrated into a practical application (MPEP 2106.04(d)(I)). If the claim contains no additional elements beyond the abstract idea, the claim fails to integrate the abstract idea into a practical application (MPEP 2106.04(d)(III)).
The judicial exceptions identified in Eligibility Step 2A Prong One are not integrated into a practical application because of the reasons noted below.
Dependent claims 6, 7, 12, 13, 14, 21, 36, 37, 40, 41, 136, 137, and 138 do not recite any elements in addition to the judicial exception, and thus are part of the judicial exception.
The additional elements in independent claim 1 include:
a computer;
receiving an electronic request to determine a carrier status of an individual (i.e., receiving data);
inputting a vector (i.e., inputting data);
inputting copy number data for a gene in a genome of the individual and SNP data for the gene (i.e., inputting data);
executing the recurrent neural network; and
outputting data (i.e., outputting data).
The additional elements in dependent claims 2, 10, 11, 20, 38, 39, include:
the gene is sequenced to determine the copy number data and the SNP data (claim 2);
the recurrent neural network includes an output for each carrier status of the plurality of carrier statuses (claim 10);
the recurrent neural network includes an input for: each copy number value of the one or more copy number values, and each SNP value of the one or more SNP values (claim 11);
the data structure is provided as input to the recurrent neural network to generate the respective carrier status of the individual (i.e., inputting data) (claim 20);
the copy number data and the SNP data are represented as a single data structure inputted to the recurrent neural network (i.e., inputting data) (claim 38); and
the recurrent neural network is configured to further receive, as inputs, the copy number data for the pseudogene and the SNP data for the pseudogene (claim 39).
The additional element of a computer (claim 1) invokes a computer and/or computer-related components merely as tools for use in the claimed process, and therefore is not an improvement to computer functionality itself, or an improvement to any other technology or technical field, and thus, does not integrate the judicial exceptions into a practical application (MPEP 2106.04(d)(1)).
The additional elements of receiving data (claim 1), inputting data (claims 1, 20, and 38), and outputting data (claim 1); are merely pre-solution and/or post-solution activities used in the claimed process – nominal additions to the claims that do not meaningfully limit the claims, and therefore do not add more than insignificant extra-solution activity to the judicial exceptions (MPEP 2106.05(g)).
The additional element of gene sequencing to determine the copy number data and the SNP data (claim 2) is merely a pre-solution activity of gathering data for use in the claimed process – a nominal addition to the claims that does not meaningfully limit the claims, and therefore does not add more than insignificant extra-solution activity to the judicial exceptions (MPEP 2106.05(g)).
The additional element of executing a recurrent neural network (claim 1) provides nothing more than mere instructions to implement an abstract idea on a generic computer (MPEP 2106.05(f)), and merely confines the use of the abstract idea to the particular technological environment of neural networks (MPEP 2106.05(h)). The additional elements of the recurrent neural network includes an output for each carrier status of the plurality of carrier statuses (claim 10); the recurrent neural network includes an input for: each copy number value of the one or more copy number values, and each SNP value of the one or more SNP values (claim 11); and the recurrent neural network is configured to further receive, as inputs, the copy number data for the pseudogene and the SNP data for the pseudogene (claim 39); are additional elements that merely further limit the use of a recurrent neural network (claim 1) and therefore provide nothing more than mere instructions to implement an abstract idea on a generic computer (MPEP 2106.05(f)), and merely further confine the use of the abstract idea to the particular technological environment of neural networks (MPEP 2106.05(h)).
Thus, the additionally recited elements merely invoke a computer as a tool; and/or amount to insignificant extra-solution activity; and/or are mere instructions to implement an abstract idea on a generic computer; and/or confine the use of the abstract idea to a particular technological environment; and as such, when all limitations in claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 have been considered as a whole, the claims are deemed to not recite any additional elements that would integrate a judicial exception into a practical application.
Therefore claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are directed to an abstract idea (MPEP 2106.04(d)).
[Step 2A Prong Two: NO]
Eligibility Step 2B: Because the claims recite an abstract idea, and do not integrate that abstract idea into a practical application, the claims are probed for a specific inventive concept. The judicial exception alone cannot provide that inventive concept or practical application (MPEP 2106.05). Identifying whether the additional elements beyond the abstract idea amount to such an inventive concept requires considering the additional elements individually and in combination to determine if they amount to significantly more than the judicial exception (MPEP 2106.05A i-vi).
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception(s) because of the reasons noted below:
Dependent claims 6, 7, 12, 13, 14, 21, 36, 37, 40, 41, 136, 137, and 138 do not recite any elements in addition to the judicial exception(s). The additional elements recited in independent claim 1 and dependent claims 2, 10, 11, 20, 38, and 39 are identified above, and carried over from Step 2A: Prong Two along with their conclusions for analysis at Step 2B. Any additional element or combination of elements that was considered to be insignificant extra-solution activity at Step 2A: Prong Two was re-evaluated at Step 2B, because if such re-evaluation finds that the element is unconventional or otherwise more than what is well-understood, routine, conventional activity in the field, this finding may indicate that the additional element is no longer considered to be insignificant; and all additional elements and combination of elements were evaluated to determine whether any additional elements or combination of elements are other than what is well-understood, routine, conventional activity in the field, or simply append well-understood, routine, conventional activities previously known to the industry, specified at a high level of generality, to the judicial exception, per MPEP 2106.05(d).
The additional elements of a computer (claim 1); receiving data (claim 1), inputting data (claims 1, 20, and 38), and outputting data (claim 1); and a recurrent neural network (claims 1, 10, 11, and 39); are conventional computer components and/or functions (see MPEP at 2106.05(b) and 2106.05(d)(II) regarding conventionality of computer components and computer processes). Additionally, the Specification at para. [0026] explains that because DNA, and thus genes and pseudogenes of interest, can have a sequential character, recurrent neural networks (RNNs) can be especially conducive for use in such applications, because RNNs make use of sequential information in their operation in that the output of a RNN for a given element in a sequence depends on the operations of the RNN during the previous one or more elements in the sequence – such operation that is grounded in sequential operations.
The additional element of gene sequencing (claim 2) is conventional. Evidence for the conventionality is shown by Tayoun et al. (Expert Review of Molecular Diagnostics, 2016, Vol.16:9, pp. 987-999, as cited in the Office action mailed 28 February 2024). Tayoun et al. reviews sequencing-based diagnostics for pediatric genetic diseases (Title; and Abstract), and shows a historical perspective of sequencing-based diagnostics (page 987, column 2, Section 1.2) and further reviews next-generation sequencing (page 988, Section 2; and page 989, Figure 1) diagnostic utility of genomic testing (page 991, Section 4), and key issues (page 996, column 2). Tayoun et al. further shows use of probes in genomic sequencing (page 988, column 2, para. 3; and page 989, Figure 1).
Therefore, when taken alone, all additional elements in claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 do not amount to significantly more than the above-identified judicial exception(s). Even when evaluated as a combination, the additional elements fail to transform the exception(s) into a patent-eligible application of that exception. Thus, claims 1, 2, 6, 7, 10-14, 20, 21, 36-41, and 136-138 are deemed to not contribute an inventive concept, i.e., amount to significantly more than the judicial exception(s) (MPEP 2106.05(II)).
[Step 2B: NO]
Response to Arguments
The Applicant’s arguments/remarks received 19 December 2025 have been fully considered, but they are not persuasive.
The Applicant states on page 8 (para. 2) of the Remarks that the claims are patent-eligible because they are directed toward a particular and novel paradigm for training machine-learning models. The Applicant further states that the Office action alleges that the claims recite a mathematical concept and a mental process at Step 2A Prong One, however when the claims are properly considered in their entirety and in light of the recited computer-specific features disclosed in the Specification (including a position-encoded genomic array during RNN training and execution (paras. [0037] – [0039]; and FIGS. 4A and 5)), it is clear that the claims recite a technical solution implemented in a computing environment, not a mental act or abstract mathematics. The Applicant further states (para. 3) that the recited features define a specific way to train and execute RNNs that is inseparable from the disclosed computing architecture, and that the claimed method changes how the computer stores, retrieves, and feeds data into the neural network, improving execution speed and accuracy, which constitutes an improvement to the functioning of the computer itself. The Applicant further states that under the USPTO’s eligibility framework the claims are not directed toward a mental process or mathematical relationship.
These arguments are not persuasive, because first, with regard to the Applicant’s argument that the claims are patent-eligible because they are directed toward a particular and novel paradigm for training machine-learning models, it is noted that in the above rejection the claims are determined to recite judicial exceptions at Step 2A Prong One that are not integrated into a practical application at Step 2A Prong Two because they are determined to not recite any additional elements that apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception, and when those additional elements are carried over to Step 2B for further evaluation of all relevant considerations discussed in MPEP §§ 2106.05(a) - (c), (e) (f) and (h), and after all additional elements have been evaluated individually and in combination, the claims are deemed to not contribute an inventive concept, i.e., amount to significantly more than the judicial exception. Furthermore, with respect to the Applicant’s assertion that the claims are directed to a particular and novel paradigm, it is noted that the search for a practical application or an inventive concept is different from an anticipation analysis under 35 U.S.C. 102 or an obviousness analysis under 35 U.S.C. 103, and because there are separate and distinct requirements from eligibility, patentability of the claimed invention under 35 U.S.C. 102 and 103 with respect to the prior art is neither required for, nor a guarantee of, patent eligibility under 35 U.S.C. 101 (MPEP 2106.05(I)). Second, with regard to the Applicant’s argument that when the claims are properly considered in their entirety and in light of the recited computer-specific features disclosed in the Specification (including a position-encoded genomic array during RNN training and execution (paras. [0037] – [0039]; and FIGS. 4A and 5)), it is clear that the claims recite a technical solution implemented in a computing environment, it is noted that in the above rejection, the claims were considered as a whole at Step 2A Prong Two, and at Step 2B all of the additional elements were evaluated individually and in combination, with the eligibility analysis at Step 2A Prong Two resulting in a determination that when the claims are considered as a whole (i.e., the analysis takes into consideration all the claim limitations and how those limitations interact and impact each other when evaluating whether the exception is integrated into a practical application), they are deemed to not recite any additional elements that would integrate a judicial exception into a practical application; and with eligibility analysis at Step 2B resulting in a determination that when all additional elements in the claims have been evaluated individually and in combination, they are deemed to not contribute an inventive concept, i.e., amount to significantly more than the judicial exceptions. Third, with regard to the Applicant’s argument that the recited features define a specific way to train and execute RNNs that is inseparable from the disclosed computing architecture, and that the claimed method changes how the computer stores, retrieves, and feeds data into the neural network, improving execution speed and accuracy, which constitutes an improvement to the functioning of the computer itself, it is noted that claims can recite a judicial exception even if they are claimed as being performed on a computer (MPEP 2106.04(a)(2)(III)(C)), i.e., the use of a computer to perform the claimed method at a rate and accuracy that can far outstrip the mental performance of a skilled artisan does not change the nature of the activity being performed (i.e., an abstract idea), and therefore does not materially alter the patent eligibility of the claimed subject matter. Furthermore, a recurrent neural network does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism.
The Applicant states on page 8 (para. 4) of the Remarks that the Examiner alleges that the claims are directed toward a mental process and a mathematical relationship, however, these allegations are in direct conflict with the latest memoranda issued by the Patent Office.
This argument is not persuasive, because first, the August 4, 2025 memorandum regarding reminders on evaluating subject matter eligibility of claims under 35 U.S.C. 101 clearly states (page 1) that the memorandum is not intended to announce any new USPTO practice or procedure and is meant to be consistent with existing USPTO guidance, and further states that Examiners should consult the specific MPEP sections for more thorough information. Second, the rejection above and the rejections of record have been raised in accordance with the eligibility analysis framework provided at MPEP 2106.
The Applicant states (page 8, para. 5 and page 9, para. 1) that the claims cannot be practically performed mentally, and that under the USPTO’s subject matter eligibility guidance, including the August 4, 2025 memorandum from the Deputy Commissioner for Patents, mental processes are considered abstract ideas only when the claim recites limitations that can be performed in the human mind or by a human using pen and paper, such as observations, evaluations, judgements, and opinions, and that the memorandum emphasizes that Examiners are reminded not to expand this grouping in a manner that encompasses claim limitations that cannot practically be performed in the human mind. The Applicant further states that the memorandum specifically states that claim limitations that encompass AI in a way that cannot be practically performed in the human mind do not fall within this grouping. The Applicant further states on page 9 (para. 2) of the Remarks that the claims recite a method carried out entirely by one or more processors for training and executing an RNN and an anomaly detection model to determine a genetic carrier status using large-scale genomic data, and that the steps include generating and segmenting a training set of copy number data and SNP data, arranging the data arrays by the relative position within a gene to cause sequential memory access during training and execution, training the RNN using this arranged data, generating input vectors representing sequencing data for an individual, detecting genetic anomalies using the anomaly detection model, and executing the trained RNN to produce a carrier status output. The Applicant further states that these operations require specific computer hardware and trained machine learning models to process and analyze substantial and complex genomic information, and because every meaningful operation in the claims requires automated, machine-based computation that the human mind is not equipped to perform, the claims are not directed to a mental process under Step 2A Prong One, that is, the claims encompass AI in a way that cannot be practically performed in the human mind, and therefore, do not all within the grouping of mental processes.
These arguments are not persuasive, because first, with regard to the Applicant’s argument that the claims cannot be practically performed mentally, it is noted that at Step 2A Prong One in the rejection above, not every limitation is identified as a mental process, but nonetheless of a limitation is identified as a judicial exception, then the eligibility analysis proceeds to Step 2A Prong Two for determination of whether or not the claim is directed to that judicial exception. Here, as noted at Step 2A Prong One in the rejection above, at least one limitation has been identified as a mental process (e.g., steps of generating, segmenting, and arranging data in arrays as part of a training set of data) prompting further analysis at Step 2A Prong Two. Second, with regard to the Applicant’s argument that the memorandum specifically states that claim limitations that encompass AI in a way that cannot be practically performed in the human mind do not fall within this grouping, it is noted that the example that is referenced in the memorandum is directed to a claim to “a specific, hardware-based RFID serial number data structure” (i.e., an RFID transponder), where the data structure is uniquely encoded (i.e., there is “a unique correspondence between the data physically encoded on the [RFID transponder] with pre-authorized blocks of serial numbers”), which is a fact patter that differs from the instant claims, and therefore should be interpreted based on the fact patterns set forth in the example's claim, as other fact patterns may have different eligibility outcomes, as evidenced in the rejection of the instant claims above. Third, with regard to the Applicant’s argument that the claims recite a method carried out entirely by one or more processors, it is noted that claims can recite a judicial exception even if they are claimed as being performed on a computer (MPEP 2106.04(a)(2)(III)(C)), i.e., the use of a computer to perform the claimed method at a rate and accuracy that can far outstrip the mental performance of a skilled artisan does not change the nature of the activity being performed (i.e., an abstract idea), and therefore does not materially alter the patent eligibility of the claimed subject matter. Fourth, with regard to the Applicant’s argument that the claims are not directed to a mental process because the operations require specific computer hardware and trained machine learning models to process and analyze substantial and complex genomic information, and because every meaningful operation in the claims requires automated, machine-based computation that the human mind is not equipped to perform, the foregoing response to arguments is reiterated, i.e., claims can recite a judicial exception even if they are claimed as being performed on a computer (MPEP 2106.04(a)(2)(III)(C)), i.e., the use of a computer to perform the claimed method at a rate and accuracy that can far outstrip the mental performance of a skilled artisan does not change the nature of the activity being performed (i.e., an abstract idea), and therefore does not materially alter the patent eligibility of the claimed subject matter.
The Applicant states on page 9 (para. 3) of the Remarks that the claims are not directed toward a mathematical relationship, and that the August 4, 2025 memorandum draws a distinction between claims that recite a mathematical relationship vs. claims that merely involve a mathematical relationship, and that specifically, the memorandum explains that a claim recites a mathematical concept when it expressly sets forth or describes a mathematical relationship, equation, formula, or calculation in words, numbers, symbols, or other expressive formats, and that a claim merely involves a mathematical concept when it uses or is based on mathematical ideas but does not explicitly set forth the mathematics itself. The Applicant further states that the guidance emphasizes that a claim is not directed to a mathematical concept simply because it uses data processing techniques or mathematical models internally in a computer implementation, and that the analysis must determine whether the mathematical relationship is what the claim is focused on or whether it is part of a broader technological process. The Applicant further states on page 10 (para. 2) of the Remarks that the pending claims are not directed to a mathematical relationship because they do not set forth or describe any mathematical formula, equation, or algorithm in mathematical terms, and instead, they recite a computer-implemented method for training and executing a recurrent neural network and an anomaly detection model using large-scale genomic data in order to determine the carrier status of an individual, and that while the neural network and anomaly detection model may use mathematical operations internally, such calculations are part of the underlying computer model implementation and are not the focus of the claims. The Applicant further states that the pending claims focus on specific technological actions, such as generating and arranging genomic data arrays by their position within a gene to enable sequential memory access, generating input vectors representing sequencing data for an individual, detecting anomalies in genetic data, and executing the trained model to output a carrier status. The Applicant further states that as the memorandum explains, claims that apply mathematical concepts as part of a specific technological solution, rather than reciting the mathematical relationship itself, are not considered directed to the mathematical concepts grouping. The Applicant further states that because the pending claims use mathematical processes only as part of a larger technological process for genomic data analysis and carrier status prediction, they are not directed toward a mathematical relationship under Step 2A Prong One, and that the Examiner has the burden of showing that the claims recite and not merely involve a mathematical relationship.
These arguments are not persuasive, because first, with regard to the Applicant’s argument that the claims are not directed toward a mathematical relationship, and that the August 4, 2025 memorandum draws a distinction between claims that recite a mathematical relationship vs. claims that merely involve a mathematical relationship, it is noted that the MPEP at 2106.04 II.A.1. is instructive, describing an example of a claim that merely involves, or is based on, an exception is a claim to "A teeter-totter comprising an elongated member pivotably attached to a base member, having seats and handles attached at opposing sides of the elongated member." This claim is based on the concept of a lever pivoting on a fulcrum, which involves the natural principles of mechanical advantage and the law of the lever. However, this claim does not recite (i.e., set forth or describe) these natural principles and therefore is not directed to a judicial exception. In contrast, a claim that recites (i.e., sets forth or describes) training a recurrent neural network recites a mathematical concept because a key aspect of RNN training is weight adjustment, i.e., the training necessitates updating the weights (Wx for input, Wh for hidden state) and biases to reduce loss, which is fundamental to learning. Second, with regard to the Applicant’s argument that while the neural network and anomaly detection model may use mathematical operations internally, such calculations are part of the underlying computer model implementation and are not the focus of the claims, it is noted that at eligibility step 2A Prong One, claims are evaluated under the broadest reasonable interpretation (BRI) of the claim limitations when read in light of and consistent with the specification in order to determine whether a claim recites an abstract idea that falls within at least one of the mental processes and/or mathematical concepts groupings of abstract ideas, as indicated above (see MPEP 2106.04(a)), and if so, then the eligibility analysis proceeds to Step 2A Prong Two in order to determine whether the claim is directed to the judicial exception (i.e., whether or not there are any additional elements that integrate the judicial exception(s) into a practical application). Third, with regard to the Applicant’s argument that as the memorandum explains, claims that apply mathematical concepts as part of a specific technological solution, rather than reciting the mathematical relationship itself, are not considered directed to the mathematical concepts grouping, and further that the Examiner has the burden of showing that the claims recite and not merely involve a mathematical relationship, it is noted that this passage is referencing the improvements consideration (MPEP 2106.04(d)(1), which is a component of the eligibility analysis at Step 2A Prong Two where all limitations in the claim have been considered as a whole (i.e., the analysis takes into consideration all the claim limitations and how those limitations interact and impact each other when evaluating whether the exception is integrated into a practical application) to determine whether a claim recites any additional elements that would integrate a judicial exception into a practical application (MPEP 2106.04(d)). As noted in the above rejection, when all limitations in the instant claims have been considered as a whole, the claims are deemed to not recite any additional elements that would integrate a judicial exception into a practical application, and therefore the instant claims are directed to an abstract idea (MPEP 2106.04(d))
The Applicant further states on page 10 (para. 4) of the Remarks that the claims are patent-eligible because they provide tangible and concrete technical improvements to a computer’s functioning, and that the claims, e.g., claim 1, recite a computer-implemented method that trains and executes a recurrent neural network (RNN) using a novel arrangement of genomic training data, and that this recited arrangement is expressly described in the Specification as improving the accuracy and speed of execution of the RNN. The Applicant further states (para. 5) that claim 1 recites generating a training set that combines copy-number variation (CNV) data with single-nucleotide polymorphism (SNP) data, and that paragraph [0020] of the Specification states that whether an individual has two functional copies of the gene of interest…can be determined using copy number data and single nucleotide polymorphism (SNP) data relating to the gene of interest…in the individual’s genome. The Applicant further states that claim 1 then recites segmenting this training set into arrays and arranging each data array according to a sequential characteristic that corresponds to its relative position within a given gene, and that this arrangement is directly described by Figure 5 (reproduced on page 11 of the Remarks) and in paragraphs [0037] – [0040] of the Specification. The Applicant further states that for example, paragraph [0037] describes that the performance of the RNN-based processes of the disclosure can be improved by representing the carrier status data in array y 504 in a manner having a sequential characteristic that corresponds to the sequence of the genetic material in the gene/pseudogene of interest. The Applicant further states that moreover, paragraph [0039] further discusses that representing the above data can result in better RNN performance (e.g., more accurate carrier status determinations, faster carrier status determinations, etc.) than others. The Applicant further states that claim 1 now additionally recites “wherein arranging the training set within the plurality of data arrays according to the order causes sequential memory access during training and execution” and that this limitation is fully supported by the detailed disclosure of the ordered data arrays (paras. [0037] – [0039]; FIGS. 4A and 5) and the computing architecture used to implement and execute the recurrent neural network (paras. [0042] – [0044]). The Applicant further states on page 11 (bottom) that Figure 4A (reproduced on page 12 of the Remarks) expressly illustrates the sequential nature of RNN processing, and further states on page 12 (para. 1) that this unrolling or unfolding of the RNN demonstrates that the network consumes its inputs strictly in sequence, in alignment with their index positions in the input array, and that when the training set is arranged in an order corresponding to the relative positions of the data within a gene, the processor iterates through that array index order, resulting in sequential memory access during both training and execution. The Applicant further states (para. 2) that as paragraph [0044] of the Specification describes, the computing system includes memory, storage, and processing hardware such as a CPU and optional GPU cores optimized for machine-learning tasks, interconnected via a bus. The Applicant further states that on such hardware, iterating through an ordered array naturally results in sequential iteration through contiguous memory locations, and that this sequential access is a hardware-level behavior that improves execution efficiency and produces faster results, as discussed in the Specification, and therefore, achieving a faster and more efficient performance of the recited RNN is directly linked to how the data is arranged. The Applicant further states (para. 3) that by preserving sequential and positional relationships within genetic sequences, the claimed arrangement enables the RNN to process data in alignment with the biological structure of the gene or pseudogene, and that paragraph [0026] of the Specification explains that DNA, and thus genes and pseudogenes of interest, can have a sequential character, and recurrent neural networks (RNNs) can be especially conducive because RNNs make use of sequential information in their operation. The Applicant further states that the disclosed ordering leverages this characteristic to capture patterns in genomic variation more efficiently, enhancing predictive accuracy, and that (para. 4) this is not a generic data-format step because the position-encoded genomic array and its deliberate order directly control how the processor stores and streams the data into the RNN. The Applicant further states on page 13 (para. 1) that the claimed arrangement integrates the claimed concepts into a practical application that modifies computer operation at both the software and hardware levels, because at the software level, the ordering ensures biologically relevant sequential relationships are preserved in network input, improving model-inference precision (paras. [0037] – [0039] in the Specification); and at the hardware level, the sequential index ordering (shown in FIG. 4A’s input sequence) causes memory to be accessed in sequence, reducing idle cycles and memory stalls during training and inference (para. [0044]). The Applicant further states (para. 2) that the Specification’s performance benefits (e.g., more accurate carrier status determinations, faster carrier status determinations, para. [0039] in the Specification) are directly linked to the claimed arrangement and the resulting sequential memory access, and accordingly (para. 3), the claims go beyond invoking a generic neural network, as alleged by the Examiner, because they recite a specific arrangement of genomic data arrays, exemplified in FIGS. 4A and 5, that changes how the computer stores and retrieves training data during training and execution of the RNN, and this results in measurable hardware-level efficiency gains and software-level accuracy improvements. Finally, the Applicant states that such a specific technical advance (fully disclosed in the Specification) constitutes an improvement to computer operation in a defined technical field and therefore claims patent-eligible subject matter.
These arguments are not persuasive, because first, with regard to the Applicant’s argument that the claims are patent-eligible because they provide tangible and concrete technical improvements to a computer’s functioning, and that the claims, e.g., claim 1, recite a computer-implemented method that trains and executes a recurrent neural network (RNN) using a novel arrangement of genomic training data, and that this recited arrangement is expressly described in the Specification as improving the accuracy and speed of execution of the RNN, it is noted that the purported novel arrangement of genomic training data is a limitation that is identified as a judicial exception at Step 2A Prong One in the above rejection, and further noted the purported improvement to accuracy and speed of execution of the RNN is at most a purported improvement to the machine-learned algorithm, and not an improvement to computer functionality itself, or an improvement to another technology or technical field. Second, with regard to the Applicant’s argument that generating a training set that combines copy-number variation (CNV) data with single-nucleotide polymorphism (SNP) data and segmenting this training set into arrays and arranging each data array according to a sequential characteristic that corresponds to its relative position within a given gene provide an improvement to the performance of the RNN-based processes of the disclosure, it is noted that the claim 1 limitations reciting generating and segmenting a training set of data arrays comprises limitations that are identified as judicial exceptions at Step 2A Prong One in the above rejection, and at Step 2A Prong Two in the above rejection, these judicial exceptions are determined to not be integrated into a practical application when the claims are considered as a whole. Third, with regard to the Applicant’s argument that the new claim 1 limitation reciting “wherein arranging the training set within the plurality of data arrays according to the order causes sequential memory access during training and execution” and that the unrolling or unfolding of the RNN demonstrates that the network consumes its inputs strictly in sequence, in alignment with their index positions in the input array, and that when the training set is arranged in an order corresponding to the relative positions of the data within a gene, the processor iterates through that array index order, resulting in sequential memory access during both training and execution, it is noted that the limitations reciting arranging data of data arrays according to a particular order are limitations that comprise the judicial exceptions identified at Step 2A Prong One in the above rejection, and the description of how the RNN works provided in the Applicant’s argument appears to describe characteristics inherent to the operation of RNNs, e.g., a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism; and a key architectural aspect of RNNs is “unfolding in time,” that is, though visually represented as a single node with a loop, an RNN is “unfolded” (or unrolled) into a feedforward structure during training, creating a layer for each step in the sequence. Fourth, with regard to the Applicant’s argument that iterating through an ordered array naturally results in sequential iteration through contiguous memory locations, and that this sequential access is a hardware-level behavior that improves execution efficiency and produces faster results, as discussed in the Specification, and therefore, achieving a faster and more efficient performance of the recited RNN is directly linked to how the data is arranged, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and therefore if the purported improvements of achieving faster and more efficient performance of the recited RNN is directly linked to how the data is arranged, then it would follow that the purported improvement is to the judicial exception, and not to an improvement to computer functionality itself, or an improvement to another technology or technical field. Furthermore, and as explained in foregoing responses to arguments, the description of how the RNN works provided in the Applicant’s argument appears to describe characteristics inherent to the operation of RNNs, as previously explained. Fifth, with regard to the Applicant’s argument that by preserving sequential and positional relationships within genetic sequences, the claimed arrangement enables the RNN to process data in alignment with the biological structure of the gene or pseudogene, and recurrent neural networks (RNNs) can be especially conducive because RNNs make use of sequential information in their operation and that the disclosed ordering leverages this characteristic to capture patterns in genomic variation more efficiently, enhancing predictive accuracy, and that this is not a generic data-format step because the position-encoded genomic array and its deliberate order directly control how the processor stores and streams the data into the RNN, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism. Sixth, with regard to the Applicant’s argument that the claimed arrangement integrates the claimed concepts into a practical application that modifies computer operation at both the software and hardware levels, because at the software level, the ordering ensures biologically relevant sequential relationships are preserved in network input, improving model-inference precision; and at the hardware level, the sequential index ordering causes memory to be accessed in sequence, reducing idle cycles and memory stalls during training and inference, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism. Seventh, with regard to the Applicant’s argument that the Specification’s performance benefits (e.g., more accurate carrier status determinations, faster carrier status determinations) are directly linked to the claimed arrangement and the resulting sequential memory access, and accordingly, the claims go beyond invoking a generic neural network, as alleged by the Examiner, because they recite a specific arrangement of genomic data arrays, exemplified in FIGS. 4A and 5, that changes how the computer stores and retrieves training data during training and execution of the RNN, and this results in measurable hardware-level efficiency gains and software-level accuracy improvements, and that such a specific technical advance (fully disclosed in the Specification) constitutes an improvement to computer operation in a defined technical field and therefore claims patent-eligible subject matter, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field.
The Applicant states on page 13 (para. 4) of the Remarks that the claimed arrangement changes the operation of the computer hardware and software. The Applicant further states that the Examiner alleges that the claimed advantage of training the RNN is a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field, and that the Office action characterizes features such as generating and arranging the training set, inputting the data into the RNN, and using anomaly detection as mental processes or mathematical concepts that could be practically performed with pen and paper, and concludes that any increased accuracy or speed is merely an enhancement to the model’s conceptual output rather than to the way the computer itself operates. The Applicant further states on page 14 (para. 1) of the Remarks that this characterization misapprehends the nature of the claimed improvement, and that as detailed in the foregoing arguments/remarks and supported by paragraphs [0037] – [0039], [0042] – [0044] in the Specification, and FIGS. 4A and 5, the claimed arrangement of the training set is not simply a mathematical abstraction, but instead, it imposes a specific ordering of the data in computer memory, and that ordering directly causes sequential memory access during training and execution of the RNN (e.g., FIG. 4A). The Applicant further states that sequential access to contiguous memory locations is a recognized hardware-level behavior that results in measurably faster execution, and that these hardware and low-level software effects are independent of the subject matter being classified and constitute improvements to computer functioning itself, not just to the abstract analytical concept. Finally, the Applicant states (para. 2) that the claimed arrangement’s technical benefits are expressly disclosed in the specification, i.e., more accurate carrier status determinations, faster carrier status determinations (para. [0039]), achieved by preserving the sequential characteristic of the genomic data and aligning it with the sequential nature of RNN operation (para. [0026]). The Applicant further states that these effects change how the processor stores, retrieves, and streams data through the network during execution, and that the Examiner’s view omits this direct integration of the claimed ordering into the computer’s physical operation. The Applicant further states (para. 3) that if the Examiner believes that a computer model running faster on a processor due to improved memory-access efficiency does not constitute a technical improvement, the Applicant requests an explanation of how such an optimization would not constitute a technical improvement to computer functionality, however the Office action includes a single conclusory statement alleging that the improvement is to the abstract idea with no explanation. The Applicant further states (para. 4) that because the improvement is tied to memory architecture and processing efficiency (a change in how the computer performs the method), the claims recite a specific advancement in computer technology within the technical field of genomic machine learning and are therefore patent-eligible.
These arguments are not persuasive, because first, with regard to the Applicant’s argument that the claimed arrangement changes the operation of the computer hardware and software, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field. Second, with regard to the Applicant’s argument that the Examiner alleges that the claimed advantage of training the RNN is a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field, and that this characterization misapprehends the nature of the claimed improvement, and that as detailed in the foregoing arguments/remarks and supported by paragraphs [0037] – [0039], [0042] – [0044] in the Specification, and FIGS. 4A and 5, the claimed arrangement of the training set is not simply a mathematical abstraction, but instead, it imposes a specific ordering of the data in computer memory, and that ordering directly causes sequential memory access during training and execution of the RNN, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field. Third, with regard to the Applicant’s argument that sequential access to contiguous memory locations is a recognized hardware-level behavior that results in measurably faster execution, and that these hardware and low-level software effects are independent of the subject matter being classified and constitute improvements to computer functioning itself, not just to the abstract analytical concept, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field. Fourth, with regard to the Applicant’s argument that these effects change how the processor stores, retrieves, and streams data through the network during execution, and that the Examiner’s view omits this direct integration of the claimed ordering into the computer’s physical operation. The Applicant further states (para. 3) that if the Examiner believes that a computer model running faster on a processor due to improved memory-access efficiency does not constitute a technical improvement, the Applicant requests an explanation of how such an optimization would not constitute a technical improvement to computer functionality, however the Office action includes a single conclusory statement alleging that the improvement is to the abstract idea with no explanation, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field. Fifth, with regard to the Applicant’s argument that because the improvement is tied to memory architecture and processing efficiency (a change in how the computer performs the method), the claims recite a specific advancement in computer technology within the technical field of genomic machine learning and are therefore patent-eligible, it is reiterated that the limitations reciting aspects of how the data is arranged comprise the judicial exceptions identified at Step 2A Prong One, and also reiterated that a recurrent neural network is a type of deep learning model designed to process sequential data (like text or time series) by maintaining an internal “memory,” thus an RNN does not fundamentally change how hardware memory (e.g., RAM/storage) works, but instead, it changes how software uses memory by creating an internal, algorithmic “hidden state” that loops back previous data, acting as a functional short-term memory, i.e., a numerical vector (hidden state) stored in RAM during computation, that is to say…not a change to the physical storage mechanism, and therefore the instant claimed improvements comprise a purported improvement to the abstract idea, and not an improvement to computer functionality itself, or an improvement to another technology or technical field.
Double Patenting
The provisional rejection of claim 1 on the ground of nonstatutory double patenting as being unpatentable over claims 1, 41, and 42 of copending Application No. 17/039,826 is maintained and modified to non-provisional.
It is noted that copending application has received a notice of allowance resulting in the issuance of U.S. Patent No.: 12,499,970. Applicants general note the rejection is traversed and the claims are amended. However, upon review the analysis appears to apply to the amended claims. Without any specific arguments, it is unclear what amendments render the rejection as moot.
Conclusion
No claims are allowed.
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.
Inquiries
Any inquiry concerning this communication or earlier communications from the examiner should be directed to STEVEN W. BAILEY whose telephone number is (571)272-8170. The examiner can normally be reached Mon - Fri. 1000 - 1800.
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, KARLHEINZ SKOWRONEK can be reached on (571) 272-9047. 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.
/S.W.B./Examiner, Art Unit 1687
/Joseph Woitach/Primary Examiner, Art Unit 1687