DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Election/Restriction
Applicant's election without traverse of Group I (claims 1-19) and Species A (claim 13) in the response filed 07/15/2026 is acknowledged. Claims 10, 11, 12, 19, 20-25 are hereby withdrawn from further consideration pursuant to 37 CFR 1.142(b), as being drawn to a nonelected invention, there being no allowable generic or linking claim.
Status of Claims
Claims 10, 11, 12, 19, 20-25 are withdrawn.
Claims 1-9 and 13-18 are under examination.
Priority
This application claims the benefit of priority to US provisional application number 63/241,645, filed September 8, 2021.
Information Disclosure Statement
The information disclosure statement (IDS) document(s) submitted is in compliance with the provisions of 37 CFR 1.97. Accordingly, the IDS document(s) has/have been fully considered by the examiner.
Drawings
The replacement drawings filed 11/28/2022 are acceptable.
Specification
The specification (on pages 78-79) recites a section entitled “REFERENCES”. Applicant is reminded that the listing of references in the specification is not a proper information disclosure statement. 37 CFR 1.98(b) requires a list of all patents, publications, applications, or other information submitted for consideration by the Office, and MPEP § 609.04(a), subsection I. states, “the list may not be incorporated into the specification but must be submitted in a separate paper.” Therefore, unless the references have been cited by the examiner on form PTO-892, they have not been considered. Accordingly, the specification is objected to as it contains non-standard section headings. See 37 CFR 1.77(b). This objection may be overcome by amending section headings to comply with 37 CFR 1.77(b).
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 1-9 and 13-18 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
The Supreme Court has established a two-step framework for this analysis, wherein a claim does not satisfy § 101 if (1) it is “directed to” a patent-ineligible concept, i.e., a law of nature, natural phenomenon, or abstract idea, and (2), if so, the particular elements of the claim, considered “both individually and ‘as an ordered combination,” do not add enough to “transform the nature of the claim into a patent-eligible application.” Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1353 (Fed. Cir. 2016) (quoting Alice, 134 S. Ct. at 2355).
Guidance: Step 1. Under the broadest reasonable interpretation, the claimed invention (claims 1 being representative) is directed to a method for performing a process (determine a treatment regimen) and therefore falls within one of the four statutory categories.
A. Guidance Step 2A, Prong 1
The Revised Guidance instructs us first to determine whether any judicial exception to patent eligibility is recited in the claim. The Revised Guidance identifies three judicially-excepted groupings identified by the courts as abstract ideas: (1) mathematical concepts, (2) certain methods of organizing human behavior such as fundamental economic practices, and (3) mental processes. In this case, the following steps of claim 1 recite an abstract idea:
encoding, independently, the first set of SNPs and the second set of SNPs by: labeling each subject as either a disease case or a control case based on the known disease outcome for the subject, and labeled each SNP in each subject as either homozygous with minor allele, heterozygous allele, or homozygous with the dominant allele;
optionally applying one or more filter to the first encoded set to create a first modified set of SNPs; training the deep neural network using the first encoded set of SNPs or the first modified set of SNPs; and validating the deep neural network using the second encoded set of SNPs.
Mental Processes
Under MPEP §2111, during patent examination, claims must be interpreted in their broadest reasonable manner consistent with the specification. This means that examiners consider the claim language in light of the specification as understood by a person of ordinary skill in the art, ensuring that the claims are not unduly narrowed by implicit limitations not explicitly recited in the claim (37 CFR 1.75(d)(1)).
Under the BRI, the recited act of encoding SNP data (via labeling as disease or control) is generically recited and sets forth or describes observing and/or analyzing data (which scientists can perform using their brains or a pencil and paper). As such, this step encompasses a mental process of observing data and/or manipulating data. MPEP 2106.04(a)(2), section III.
Under the BRI, the recited act of optionally applying is generically recited and sets forth or describes analyzing data (which scientists can perform using their brains or a pencil and paper). As such, this step encompasses a mental process of manipulating data. MPEP 2106.04(a)(2), section III.
Under the BRI, the recited acts of training and validating a deep neural network are generically recited and set forth or describe analyzing data (which scientists can perform using their brains or a pencil and paper). As such, this step encompasses a mental process of manipulating data. MPEP 2106.04(a)(2), section III.
It is important to note that “Claims that recite performing information analysis as well as the collection and manipulation of information related to such analysis, have been determined by our reviewing court to be an abstract concept that is not patent eligible. See SAP, 898 F.3d, 1165, 1167, 1168 (Claims reciting "[a] method for providing statistical analysis" (id. at 1165) were determined to be "directed to an abstract idea" (id. at 1168)); see also Content Extraction & Transmission LLC v. Wells Fargo Bank, Nat'l Ass 'n, 776 F.3d 1343, 1345, 1347 (Fed. Cir. 2014). [Step 2A, Prong 1: YES].
Mathematical Concept
With regards to training and validating a deep neural network, this step is recited at a high level of generality (without any technological details or rules directed to how they are performed or the structure of the neural network). Moreover, the training and validating are based upon the ‘encoded’ information which amounts to a mathematical correlation. This position is also supported by Ioffe et al. (IDS filed 01/22/2024), which teaches methods for accelerating deep neural training that includes assigning and adjusting weights of parameters that define the structure of the network. See entire. It is important to note that a mathematical concept need not be expressed in mathematical symbols, because “[w]ords used in a claim operating on data to solve a problem can serve the same purpose as a formula.” In re Grams, 888 F.2d 835, 837 and n.1, 12 USPQ2d 1824, 1826 and n.1 (Fed. Cir. 1989). As such, this step encompasses a mathematical concept of manipulating information through mathematical correlations or calculations. MPEP 2106.04(a)(2) Section I [Step 2A, Prong 1: YES].
B. Guidance Step 2A, Prong 2
Having made that determination, under the 2019 Guidance, the examiner next determines whether there are additional elements beyond the recited abstract idea(s) that integrate them into a practical application.
In this case, the additional steps/elements that are not part of the abstract idea are as follows:
collecting a first set of SNPs from at least 1,000 subjects with a known disease outcome from a database and a second set of SNPs from at least 1,000 other subjects with a known disease outcome from a database;
With regards to said collecting, this step is recited at a high level of generality and requires collecting data for use by the abstract idea. Accordingly, this step amounts to insignificant extra-solution activity and is not indicative of an integration into a practical application. See MPEP 2106.05(g).
With regard to the claimed ‘database’, this element is generically recited and merely used as a tool to obtain information or perform the abstract idea. Moreover, applicant is reminded that “generic computer components such as a computer and database do not satisfy the inventive concept requirement.” See MPEP 2106.05(f) and 2106.05(h). Therefore, the additionally recited steps/elements amount to insignificant extra-solution activity that does not apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. Even when viewed in combination, these additional steps/elements do not integrate the recited judicial exception into a practical application. See MPEP 2106.04(d)(1) for a list of considerations when evaluating whether additional elements integrate a judicial exception into a practical application. [Step 2A, Prong 2: NO].
C. Guidance Step 2B:
This part of the eligibility analysis evaluates whether the claim as a whole amount to significantly more than the recited exception i.e., whether any additional element, or combination of additional elements, adds an inventive concept to the claim. See MPEP 2106.05.
As discussed above, the non-abstract steps/elements amount to nothing more than insignificant extra-solution activity. Moreover, applicant’s own specification teaches well understood, routine, and conventional elements for collecting SNP data [see at least ¶0057]. Moreover, courts have also recognized the following laboratory techniques as well-understood, routine, conventional activity in the life science arts when they are claimed in a merely generic manner (e.g., at a high level of generality) or as insignificant extra-solution activity:
v. Analyzing DNA to provide sequence information or detect allelic variants, Genetic Techs., 818 F.3d at 1377; 118 USPQ2d at 1546;
Taken alone or in combination, there is nothing unconventional with regards to the above non-abstract elements/steps. See MPEP 2106.05(d)(Part II). Therefore, the independent claim(s) as a whole do not amount to significantly more than the exception itself and are not patent eligible. [Step 2B: NO].
Dependent Claims
Dependent claims 2-9 and 13-18 have also been considered under the two-part analysis but do not include additional steps/elements appended to the judicial exception that are sufficient to amount to significantly more than the judicial exception(s) for the following reasons. Regarding claim(s) 2-9, these claims further limit type or amount of data being collected for use by the abstract idea (i.e. insignificant extra-solution activity). Accordingly, these steps are not indicative of an integration into a practical application and do not amount to significant more for reasons set forth above (Step 2A, prong 2 analysis, and Step 2B analysis). Regarding claim(s) 13-18, these are all directed to limitations that further limit the specificity of the abstract idea set forth above (e.g. the type of neural network and training). As such, these claims also recite mathematical concepts for reasons discussed above in the Step 2A (prong 1) analysis. Therefore, the claims as a whole are not patent eligible. For additional guidance, applicant is directed generally to MPEP 2106.
Claim rejections - 35 USC § 112a
The following is a quotation 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 35 U.S.C. 112 (pre-AIA ), first paragraph:
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 1-9 and 13-18 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, or for pre-AIA the inventor(s), at the time the application was filed, had possession of the claimed invention. The written description requirement is separate and distinct from the enablement requirement. The specification must: (1) describe the claimed invention in a manner understandable to a person of ordinary skill in the art, and (2) show that the inventor actually invented the claimed subject matter.
Claim 1 is directed to a method of training a deep neural network for estimating polygenic risk scores for a disease. Regarding claim(s) 1, the specification fails to provide written description support for the following steps:
encoding, independently, the first set of SNPs and the second set of SNPs by: labeling each subject as either a disease case or a control case based on the known disease outcome for the subject, and labeled each SNP in each subject as either homozygous with minor allele, heterozygous allele, or homozygous with the dominant allele; optionally applying one or more filter to the first encoded set to create a first modified set of SNPs; training the deep neural network using the first encoded set of SNPs or the first modified set of SNPs; and validating the deep neural network using the second encoded set of SNPs.
Firstly, the claim is not limited to any encoding steps, filtering steps, training steps, or validating steps for achieving the claimed functions, and amounts to functional language specifying desired results and/or specific functions. In other words, it covers all ways of performing the claimed functions and the specification does not provide support for such broad genus limitations. With regards to encoding and filtering, it is unclear how these functions are being performed by merely “labeling” SNPs based on zygosity. The specification does not provide any guidance on the mathematical operations and/or algorithmic processes for performing the claimed functions and the limited discussion merely reiterates the claimed language [0027, 0071]. With regards to training and validating, the specification does not provide sufficient detail with regards to the mathematical operations and/or algorithmic processes that would be implemented for achieving the claimed functions. At best, the specification generally teaches computing a training error using a cross-entropy function and loss function [0061, 0098]. However, this does not provide sufficient guidance as to the type of training being performed and is not commensurate with the full scope of what is being claimed.
With regards to the deep neural network, the specification does not provide any sufficient guidance that would serve to clarify the structure of this neural network, how it was trained, or other specific details with regards to how it operates for estimating polygenic risk scores associated with a disease, i.e. the invention is essentially using a black box to achieve the claimed functions. The specification does disclose embodiments wherein “the deep neural network comprises a linearization layer on top of a deep inner attention neural network. In some aspects, the linearization layer computes an output as an element-wise multiplication product of input features, attention weights, and coefficients. In some aspects, the network learns a linear function of an input feature vector, coefficient vector, and attention vector…In some aspects, all hidden layers of the multi-layer neural network use a non-linear activation function, and wherein the attention layer uses a linear activation function. In some aspects, the layers of the inner attention neural network comprise 1000, 250, or 50 neurons before the attention layer” [0026]; and teaches an “example LINA model for structured data, which uses an input layer and multiple hidden layers to output the attention weights in the attention layer. The attention weights are then multiplied with the input features element-wise in the linearization layer and then with the coefficients in the output layer. The crossed neurons in the linearization layer represent element-wise multiplication of their two inputs. The incoming connections to the crossed neurons have a constant weight of 1” [0017, FIG. 10]. However, these teachings are not commensurate in scope with what is being claimed and it is improper to import narrowing limitations found in the specification into the claims. See MPEP 2111.01. Additionally, it is noted with particularity that the claimed invention does not even require a computer processor. Even if it is assumed that a computer or CPU is inherent for implementing the claimed functions, the specification does not provide a sufficient algorithm corresponding to each of the claimed functions. Such a disclosure must include a general purpose computer or computer component along with the algorithms that the computer uses to perform each claimed specialized function. Thus, one of ordinary skill in the art would not have recognized that the inventor possessed the full scope of structural and functional limitations to achieving the full scope of what is claimed or other ways of performing the claimed functions.
Methods for training a deep neural network to predict disease risk factors are are not trivial. For example, Orr et al. (Neural Networks: Tricks of the Trade, 1998, pp.1-425) teaches that training a neural network requires making many seemingly arbitrary choices such as the number and types of nodes, layers, learning rates, training and test sets, and so forth. These choices can be critical, yet there is no foolproof recipe for deciding them because they are largely problem and data dependent. Many training techniques work well for nets of small to moderate size. However, when problems consist of thousands of classes and millions of examples, not uncommon in bioinformatics applications, many of these techniques break down [see e.g., , page 4 and Chapter 17, pages 368-378].
Islam et al. (Artificial Intelligence: Emerging Trends and Applications, 2018, Ch. 17, pp. 333-351) teaches deep learning methods for predicting phenotypic traits and diseases from omics data. In particular, unlike the claimed method, Islam teaches collecting gene expression data from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) database as well as copy number alteration (CAN) data from over 2,000 breast cancer patients, inputting the data into a well-defined DNN model, as well as well-defined methods for training and validating the DNN model to predict breast cancer subtypes [Section 3.1, entire].
Larochelle et al. (journal of Machine Learning Research, 2009, 1, pp. 1-40) additionally teaches that training deep multi-layered neural networks is known to be hard. Larochelle teaches drawbacks of the standard learning strategy—consisting of randomly initializing the weights of the network and applying gradient descent using backpropagation—is known empirically to find poor solutions for networks with 3 or more hidden layers, and the benefits of unsupervised learning strategies [See Introduction, pages 1-2, and Section 3.1].
Cohen et al. (US2018/0068083; Pub. Date: 03/08/2018) teaches a method for predicting a likelihood of having ovarian cancer in a patient. Unlike the claimed method, Cohen teaches storing a set of data comprising patient record parameters and diagnostic indicators of whether or not the patient has been diagnosed with cancer; selecting a subset of the plurality of parameters for inputs into a machine learning system; randomly partitioning the set of data into training data and validation data; generating a classifier using a machine learning system based on the training data and the subset of inputs, wherein each input has an associated weight; determining whether the classifier meets a predetermined ROC statistic, specifying a sensitivity and a specificity, for correct classification of patients. In other words, the training process is much more well-defined in terms of computational operations, weights, parameters, and classifiers.
Furthermore, the claim is not limited to any particular SNPs or disease, i.e. it encompasses all known diseases, nor is it limited to any particular ‘risk scores’, i.e. they encompass both significant and insignificant scores. In this case, however, a review of the specification does not provide any evidence to suggest that applicant has sufficient knowledge of mathematical correlations between SNP data and PRSs for the full scope of diseases presently encompassed by the claim. As such, there is insufficient evidence of possession with regards known or disclosed correlation between structure and function. At best, the specification discusses significant progress made for estimating polygenic risk scores (PRS) for breast cancer [0006 and Table 2]. However, this is not commensurate in scope with what is being claimed. Therefore, after careful consideration, the instant specification fails to disclose that applicant was in possession of a deep neural network that could estimate a polygenic risk for scores for any disease.
For the reasons discussed above, the specification does not satisfy the written description requirement with respect to the full scope of what is being claimed. For more information regarding the written description requirement, see MPEP §2161.01- §2163.07(b).
Claim rejections - 35 USC § 112b
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 1-9 and 13-18 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 pre-AIA the applicant regards as the invention. Claims that depend directly or indirectly from claim(s) XXX are also rejected due to said dependency.
Claim 1 recites (in the preamble) “A computer-implemented method of training a deep neural network for estimating a polygenic risk score for a disease, the method comprising…”. In this case, the body of the claim does not recite any positive process limitation that actually result in stated purpose of “estimating…risk scores for a disease”. As such, it is unclear in what way the claimed steps achieve the purpose of the preamble. Applicant is reminded that a preamble is generally not accorded any patentable weight where it merely recites the purpose of a process or the intended use of a structure, and where the body of the claim does not depend on the preamble for completeness but, instead, the process steps or structural limitations are able to stand alone [See MPEP 2111.02]. Clarification is requested via amendment.
Claim 1 recites “collecting a first set of SNPs from at least 1,000 subjects with a known disease outcome from a database and a second set of SNPs from at least 1,000 other subjects with a known disease outcome from a database.” It is unclear what is meant by “collecting…SNPs…from a database” (as physical sequences cannot be stored in a database). The artisan would understand what is meant by collecting SNP data from a database, however, that is not what is being claimed. Clarification is requested via amendment.
Claim 1 recites “encoding, independently, the first set of SNPs and the second set of SNPs by: labeling each subject as either a disease case or a control case based on the known disease outcome for the subject, and labeled each SNP in each subject as either homozygous with minor allele, heterozygous allele, or homozygous with the dominant allele.” In this case, it is unclear as to the metes and bounds of said “encoding”, i.e. in what way is the claimed ‘encoding’ achieved by “labeling” SNPs. Such generic functional claim language amounts to descriptions of problems to be solved and covers all means or methods of performing the claimed function. See MPEP 2111.04. A review of the specification does not describe, to any appreciable extent, any algorithms, equations, or prose equivalent that correspond to the claimed function. Therefore, it is unclear what computational techniques are included or excluded by the claim language such that one of ordinary skill in the art would know how to avoid infringement. Lastly, the phrase “and labeled each SNP in each subject” appears to be grammatically incorrect and should recite “and labeling each…”. Clarification is requested via amendment.
Claim 1 recites “optionally applying one or more filters…”. It is unclear what limiting effect is intended by the term “optionally”, i.e. this is a conditional statement and it is unclear what condition controls whether or not this step is applied. Clarification is requested via amendment. Applicant is reminded that claim scope is not limited by claim language that suggests or makes optional but does not require steps to be performed. See MPEP 2111.04. For purposes of prior art, this phrase is interpreted as not being required and thus is not given patentable weight.
Claim 1 recites “training the deep neural network using the first encoded set of SNPs or the first modified set of SNPs; and validating the deep neural network using the second encoded set of SNPs.” It is unclear as to the metes and bounds of the claimed “training” such that the artisan would recognize what steps are minimally encompassed and what internal structure of the claimed model is intended. For example, the artisan would recognize that “training” a model can encompass any number of different mathematical operations, e.g. feature selection, adjusting model parameters, optimizing loss functions, etc. However, the instant claims do not set forth any steps involved in the method/process of training. As a result, it is unclear what method/process applicant is intending to encompass to achieve the claimed function. In addition, it is unclear as to the metes and bounds of the claimed “deep neural network”. A review of the specification states that “In some aspects, the trained deep neural network comprises at least three hidden layers, and each layer comprises multiple neurons. For example, each layer may comprise 1000, 250, or 50 neurons” teaches “[0024, see also 0027]. However, this is not commensurate is scope with what is claimed and it is improper to import narrowing limitations into the claims. MPEP 2111.01. Clarification is requested via amendment. Applicant is encouraged to consider importing limitations from claims 13-17.
Cited Prior Art
The following prior art made of record and not presently relied upon is considered pertinent to applicant' s disclosure. Applicant is reminded that prior art rejections under 35 U.S.C. 102 and/or 35 U.S.C. 103 may be applied in the next Office action in light of applicant's amendments, and that the next Office action can properly be made "Final" if these rejections are necessitated by amendment. See MPEP 706.07.
Hoffman et al. (Nucleic Acids Research, 2019, Vol. 47, No. 20, pp.10597-10611), teaches a deep learning model to accurately predict locus-specific signals from four epigenetic assays using only DNA sequence as input, including collecting SNPs and using them as input to a NN, as well as training and validating the NN [pages 10598-99].
Conclusion
No claims are allowed.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PABLO S WHALEY whose telephone number is (571)272-4425. The examiner can normally be reached between 1pm-9pm EST.
If attempts to reach the examiner by telephone are unsuccessful, the examiner's supervisor, Anita Coope can be reached at 571-270-3614. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PABLO S WHALEY/Primary Examiner, Art Unit 3619
Prosecution Insights