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 . This non-final office action on merits is in response to the Patent Application filed on 04/22/2024.
Status of claims
Claim 10 is new. Amendments to claim 6 is acknowledged and have been carefully considered. Claims 1-10 are pending and considered below. This application is a 371 of PCT/CN2023/080283 filed on 03/08/2023.
Information Disclosure Statement
The information disclosure statements (IDSs) filed on 04/22/2024 and 06/08/2026 has been acknowledged. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
Claim Objections
Claims 1 and 7 are objected to because of the following informalities: “a first CNV (copy number variation)” should read “a first copy number variation (CNV)” in claim 1; “to a first CNV” should read “a first copy number variation (CNV)” in claim 7. Appropriate correction is required.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are: a “first-level learning model establishment unit” in claims 1 and 3, a “second-level first learning model establishment unit” in claims 1 and 4, a “second-level second learning model establishment unit” in claims 3 and 4, a “storage unit” in claims 5 and 6, and a “prediction unit” in claims 5 and 6.
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof. A review of the specification shows no corresponding structure described in the specification for the 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph limitation.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1, and 3-6 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 applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claims 1, and 3-6 recites a “first-level learning model establishment unit” in claims 1 and 3, a “second-level first learning model establishment unit” in claims 1 and 4, a “second-level second learning model establishment unit” in claims 3 and 4, a “storage unit” in claims 5 and 6, and a “prediction unit” in claims 5 and 6 without proper description within the specification. Application’s specification does recite a “first-level learning model establishment unit”, a “second-level first learning model establishment unit”, a “second-level second learning model establishment unit”, a “storage unit”, and a “prediction unit”, but does not describe with sufficient detail the specific structure, algorithm, or components that perform the claimed function beyond broad functional language. Thus, limitations lack written description.
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, and 3-6 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.
Claims 1, and 3-6 recites a “first-level learning model establishment unit” in claims 1 and 3, a “second-level first learning model establishment unit” in claims 1 and 4, a “second-level second learning model establishment unit” in claims 3 and 4, a “storage unit” in claims 5 and 6, and a “prediction unit” in claims 5 and 6 and invoke 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. The “first-level learning model establishment unit”, “second-level first learning model establishment unit”, “second-level second learning model establishment unit”, “storage unit”, and “prediction unit” are cited as a part of the information processing method, suggesting physical components. However, the claims and specification do not describe any corresponding structure (hardware and software) on which the “units” operate. Therefore, the claim is indefinite and is rejected under 35 U.S.C. 112(b) or pre-AIA 35 U.S.C. 112, second paragraph.
Applicant may:
(a) Amend the claim so that the claim limitation will no longer be interpreted as a limitation under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph;
(b) Amend the written description of the specification such that it expressly recites what structure, material, or acts perform the entire claimed function, without introducing any new matter (35 U.S.C. 132(a)); or
(c) Amend the written description of the specification such that it clearly links the structure, material, or acts disclosed therein to the function recited in the claim, without introducing any new matter (35 U.S.C. 132(a)).
If applicant is of the opinion that the written description of the specification already implicitly or inherently discloses the corresponding structure, material, or acts and clearly links them to the function so that one of ordinary skill in the art would recognize what structure, material, or acts perform the claimed function, applicant should clarify the record by either:
(a) Amending the written description of the specification such that it expressly recites the corresponding structure, material, or acts for performing the claimed function and clearly links or associates the structure, material, or acts to the claimed function, without introducing any new matter (35 U.S.C. 132(a)); or
(b) Stating on the record what the corresponding structure, material, or acts, which are implicitly or inherently set forth in the written description of the specification, perform the claimed function. For more information, see 37 CFR 1.75(d) and MPEP §§ 608.01(o) and 2181.
For the purpose of examination, Examiner interprets the “detection unit” and “determination unit” to be operated on a general-purpose processor or computing system executing software instructions, and suggests the Applicant recite that the elements in the claims are implemented by specifically programmed processor, dedicated hardware circuits, or other defied computing architecture to provide sufficient structural support under 35 U.S.C. 112(f), if applicable.
Correction and/or clarification is/are required.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-10 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Under step 1, the analysis is based on MPEP 2106.03, and claims 1-6 are drawn to a system, and claims 7-10 drawn to a method. Thus, each claim, on its face, is directed to one of the statutory categories (i.e., useful process, machine, manufacture, or composition of matter) of 35 U.S.C. §101.
Step 2A Prong One
Claim 1 recites the limitations of establishing a first-level learning model, according to a first CNV (copy number variation), a first sample cancer type and a first gender of each of a plurality of first learning sample; using a second gender and a second CNV of each of a plurality of second learning sample as an input of the first-level learning model, so that the first-level learning model outputs a plurality of first output cancer types of each second learning sample; establishing a second-level first learning model, according to the first output cancer types and a second sample cancer type of each of the second learning samples. These limitations, as drafted, are processes that, under their broadest reasonable interpretation, cover performance of the limitation in the mind or by using a pen and paper. Even when considering the “by using a machine learning technique” language, the claim encompasses a user evaluating copy number variation (CNV) information, cancer type information, and gender information associated with learning samples, identifying relationships between such information, generating cancer-type classifications based on those relationships, comparing generated classifications with known cancer types, and refining classification criteria based on the comparisons, all of which are observations, judgements, and classifications. The mere nominal recitation of ‘”by using a machine learning technique” does not take the claim from the mental processes grouping because the claim does note recite any particular machine learning architecture, training algorithm, or technological improvement, but instead recites the result of analyzing information and generating classifications. Thus, the claim recites a mental process which is an abstract idea.
Independent claim 7 recites identical or nearly identical steps with respect to claim 1 (and therefore also recite limitations that fall within this subject matter grouping of abstract ideas), and this claim is therefore determined to recite an abstract idea under the same analysis.
Under Step 2A Prong Two
The claimed limitations, as per claim 1, include:
a first-level learning model establishment unit configured to:
establish a first-level learning model, by using a machine learning technique, according to a first CNV (copy number variation), a first sample cancer type and a first gender of each of a plurality of first learning sample;
use a second gender and a second CNV of each of a plurality of second learning sample as an input of the first-level learning model, so that the first-level learning model outputs a plurality of first output cancer types of each second learning sample; and
a second-level first learning model establishment unit configured to:
establish a second-level first learning model, by using the machine learning technique, according to the first output cancer types and a second sample cancer type of each of the second learning samples.
Examiner Note: underlined elements indicate additional elements of the claimed invention identified as performing the steps of the claimed invention.
The judicial exception expressed in claim 1 is not integrated into a practical application. The claim as a whole merely describes how to generally “apply” the concept of evaluating CNV information, cancer type information, and gender information to generate cancer-type classifications and refine classifications criteria based on the comparison of predicted and known cancer types in a computer environment. The claimed computer components (i.e., a first-level learning model establishment unit configured to, by using a machine learning technique, and a second-level first learning model establishment unit configured to) are recited at a high level of generality and are merely invoked as tools to perform an existing process of analyzing information, identifying relationships between data, generating classifications, and refining classification rules bases on analysis. Simply implementing the abstract idea on a generic computer is not a practical application of the abstract idea. Accordingly, alone and in combination, these additional elements do not integrate the abstract idea into a practical application. The claim is directed to an abstract idea.
Therefore, under step 2A, the claims are directed to the abstract idea, and require further analysis under Step 2B.
Under step 2B
Claim 1 does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed with respect to Step 2A, the claim as a whole merely describes how to generally “apply” the concept of evaluating CNV information, cancer type information, and gender information to generate cancer-type classifications and refine classifications criteria based on the comparison of predicted and known cancer types in a computer environment. Thus, even when viewed as a whole, nothing in the claim adds significantly more (i.e., an inventive concept) to the abstract idea. The claim is not patent eligible.
Claims 2 and 9-10 recite no further additional elements, and only further narrow the abstract idea. The previously identified additional elements, individually and as a combination, do not integrate the narrowed abstract idea into a practical application for reasons similar to those explained above, and do not amount to significantly more than the narrowed abstract idea for reasons similar to those explained above.
Claims 3-6, and 8 recite the additional element of a “first-level learning model establishment unit” in claim 3, a “second-level first learning model establishment unit” in claim 4, a “second-level second learning model establishment unit” in claims 3 and 4, a “storage unit” in claims 5 and 6, and a “prediction unit” in claims 5 and 6, and “by using machine learning technology” in claims 3 and 8. However, these additional elements amount to implementing an abstract idea on a generic computing device. As such, these additional elements, when considered individually or in combination with the prior devices, do not integrate the abstract idea into a practical application or amount to significantly more than the abstract idea.
Thus, as the dependent claims remain directed to a judicial exception, and as the additional elements of the claims do not amount to significantly more, the dependent claims are not patent eligible.
Therefore, the claims here fail to contain any additional element(s) or combination of additional elements that can be considered as significantly more and the claims are rejected under 35 U.S.C. 101 for lacking eligible subject matter.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-10 are rejected under 35 U.S.C. 103 as being unpatentable over Drake et al. (International Publication Number WO2019200410A1), referred to hereinafter as Drake, in view of Shi et al. (International Publication Number WO2022015700A1), referred to hereinafter as Shi, and Kotlov et al. (International Publication Number WO2022120256A2), referred to hereinafter as Kotlov.
Regarding claim 1, Drake teaches a cancer type prediction system, comprising: (Drake [0466] This example describes using perform predictive analytics using artificial intelligence based approaches to analyze acquired cfDNA data from a subject (to generate an output of diagnosis of the subject having a cancer (e.g., colorectal cancer, breast cancer or liver cancer or pancreatic cancer).”).
a first-level learning model establishment unit configured to: establish a first-level learning model, by using a machine learning technique, according to a first CNV (copy number variation), a first sample cancer type (Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”, Drake [0237] “As described above, a large set of features can be generated to provide a feature space from which a feature vector can be determined. This feature vector from each of a set of training samples can then be used for training a current version of the machine learning model. The types of features used can depend on the types of analytes used.” Drake [0238] “Examples of features can include variables related to structural variations (SVs), such as a copy number variation and translocations; fusions; mutations (e.g., SNPs or other single nucleotide variations (SNVs), or slightly larger sequence variations); telomere attrition; and nucleosome occupancy and distribution. These features can be calculated genomewide. Example classes (types) of features are provided below. When genetic sequence data is obtained from at least one of die analytes, example features can include aligned features (e.g., a comparison with one or more reference genomes) and non- aligned features. Example aligned features can include sequence variations and sequence counts in a genomic window. Example non-aligned features can include kmers from sequence reads and biological derived information from the reads.”, Drake [0306] “The methods and systems described herein are applicable to various cancer types, similar to grade and stage, and as such, is not limited to a single cancer disease type. Therefore, combinations of analytes and assays may be used in the present systems and methods to predict responsiveness of cancer therapeutics across different cancer types in different tissues and classifying individuals based on treatment responsiveness. In one example, the classifiers described herein are capable of stratifying a group of individuals into treatment responders and non-responders.”);
second CNV of each of a plurality of second learning sample as an input of the first-level learning model (Drake [0238] “Examples of features can include variables related to structural variations (SVs), such as a copy number variation and translocations; fusions; mutations (e.g., SNPs or other single nucleotide variations (SNVs), or slightly larger sequence variations); telomere attrition; and nucleosome occupancy and distribution. These features can be calculated genomewide. Example classes (types) of features are provided below. When genetic sequence data is obtained from at least one of die analytes, example features can include aligned features (e.g., a comparison with one or more reference genomes) and non- aligned features. Example aligned features can include sequence variations and sequence counts in a genomic window. Example non-aligned features can include kmers from sequence reads and biological derived information from the reads.”, Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”, Drake [0306] “The methods and systems described herein are applicable to various cancer types, similar to grade and stage, and as such, is not limited to a single cancer disease type. Therefore, combinations of analytes and assays may be used in the present systems and methods to predict responsiveness of cancer therapeutics across different cancer types in different tissues and classifying individuals based on treatment responsiveness. In one example, the classifiers described herein are capable of stratifying a group of individuals into treatment responders and non-responders.”); and
a second sample cancer type of each of the second learning samples (Drake [0306] “The methods and systems described herein are applicable to various cancer types, similar to grade and stage, and as such, is not limited to a single cancer disease type. Therefore, combinations of analytes and assays may be used in the present systems and methods to predict responsiveness of cancer therapeutics across different cancer types in different tissues and classifying individuals based on treatment responsiveness. In one example, the classifiers described herein are capable of stratifying a group of individuals into treatment responders and non-responders.”, and Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”).
Drake fails to explicitly teach a first gender of each of a plurality of first learning sample; use a second gender as an input, so that the first-level learning model outputs a plurality of first output cancer types of each second learning sample; and a second-level first learning model establishment unit configured to: establish a second-level first learning model, by using the machine learning technique, according to the first output cancer types.
Shi teaches a first gender of each of a plurality of first learning sample (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”);
use a second gender as an input, so that the first-level learning model outputs a plurality of first output cancer types of each second learning sample (Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”.); and
according to the first output cancer types (Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”.).
Kotlov teaches a second-level first learning model establishment unit configured to: establish a second-level first learning model, by using the machine learning technique (Kotlov, page 28, “In some embodiments, the hierarchy of machine learning classifiers (e.g., hierarchy of DNA-based machine learning classifiers or the hierarchy of RNA-based machine learning classifiers) may be stored in any suitable way. Each of the machine learning classifiers may comprise program code that, when executed, performs classification using the machine learning classifier’s inputs, the machine learning classifier’s parameters, the machine learning classifier’s hyperparameters, and/or any other suitable configuration information. The hierarchical relationships among the machine learning classifiers may be stored using one or more data structures having one or more fields storing information about the hierarchy. For example, the fields may store information indicating, for each machine learning classifier in the hierarchy, its relationship to one or more other machine learning classifiers in the hierarchy (e.g., indicating a parent machine learning classifier and/or one or more child machine learning classifiers), a respective category in the hierarchy of molecular categories to which the classifier corresponds, and/or any other suitable information, as aspects of the technology described herein are not limited in this respect.”, and Kotlov, page 26, “Another aspect of the approach developed by the inventors that contributes to its accuracy and robustness is the use of features (e.g., features derived from DNA and/or RNA expression data, which features may include the DNA and/or RNA expression data itself, in some embodiments) specified a priori for each molecular category to determine whether to identify the molecular category as a candidate molecular category for the biological sample. For example, RNA expression data for a specific set of genes for a particular molecular category may be processed using a machine learning classifier trained to predict whether a particular molecular category should be identified for the biological sample. The RNA expression data may be first processed to obtain a set of features specified a priori for the particular molecular category (e.g., gene rankings for a set of genes associated with the molecular category, the gene rankings obtained by numerically ranking the expression levels for genes in the set of genes) and this set of features may be provided as input to a specific machine learning classifier for that specific molecular category. As another example, DNA expression data may be used to obtain a specific set of DNA features (e.g., features indicating the presence of gene mutations, presence of genes, copy number alterations, loss of heterozygosity (LOH), ploidy, tumor mutational burden, presence of gene fusions, micro satellite instability (MSI) status, etc.) for a particular molecular category. Then these DNA features may be provided as input to and be processed using a machine learning classifier trained to predict whether the molecular category is a candidate molecular category for the biological sample. In some embodiments, the use of specific features tailored for each particular molecular category allows the techniques developed by the inventors to leverage domain- specific knowledge to distinguish among molecular categories, even when they share similar molecular features, contributing to the success of the techniques described herein. Examples of RNA and DNA features used by RNA-based and DNA-based machine learning classifiers, respectively, are provided herein.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to modify the cancer prediction machine learning system of Drake, as further informed by Shi, to employ the hierarchical machine learning classifier architecture taught by Ketlow. Drake teaches training machine learning models using biological sample data including copy number variation (CNV) features and cancer training samples to generate cancer classifications. Shi further teaches generating multiple cancer type predictions for a sample, including ranked cancer predictions based upon biomarker information, age, and gender. Ketlow teaches a hierarchy of machine learning classifiers comprising parent and child classifiers and further teaches machine learning classifiers configured to identify candidate molecular categories from DNA features, including copy number alterations. Collectively, these references teach generating initial cancer classifications and employing multiple levels of machine learning classification directed to candidate categories.
One of ordinary skill in the art would have recognized that utilizing Ketlow's hierarchical classifier framework within the cancer classification systems of Drake and Shi would have predictably improved classification accuracy by allowing an initial classifier to identify candidate cancer categories and a subsequent classifier to further refine those classifications. Ketlow teaches that the use of classifier hierarchies and category specific classifiers contributes to the accuracy of classification by leveraging domain specific knowledge to distinguish among categories that share similar molecular characteristics. Thus, it would have been obvious to apply a second level classifier to further evaluate outputs generated by a first level cancer classifier in order to improve predictive performance and reduce classification ambiguity among multiple possible cancer types.
Furthermore, one of ordinary skill in the art would have had a reasonable expectation of success in combining the references because all of the references are directed to machine learning cancer classification using biological and molecular features. Drake teaches machine learning cancer classification using genomic features including CNVs, Shi teaches generating multiple candidate cancer predictions using machine learning models that incorporate demographic and biomarker information, and Ketlow teaches hierarchical machine learning classifiers that process molecular category predictions using parent-child classifier relationships. Combining these known machine learning techniques would have amounted to the predictable use of prior art elements according to their established functions to obtain the expected benefit of improved cancer type classification accuracy and refined cancer predictions.
Regarding claim 2, Drake, Shi, and Kotlov teach the invention in claim 1, as discussed above, and further teach wherein the first genders of these persons to whom the first learning samples belongs are a combination of male and female (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”)., and
the second genders of these persons to whom all of the second learning samples belongs are male or female (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to utilize both mixed gender and gender specific training populations within the cancer prediction system in view of the teachings of Shi. Shi teaches generating machine learning classifier models using patient datasets that include age, gender, and biomarker information from patient populations and further teaches predicting multiple cancer types using such datasets. Shi additionally teaches that separate classifier models were developed for male and female patients and that such gender specific models demonstrated improved performance. A person of ordinary skill in the art would have understood that developing separate male and female models necessarily involves partitioning an overall population containing both male and female patients into gender specific subsets for model training. Therefore, it would have been obvious to use a first learning sample population comprising a combination of male and female patients while utilizing a second learning sample population consisting of only male patients or only female patients in order to account for gender differences in biomarker expression, cancer prevalence, and classifier performance, thereby predictably improving the accuracy of the cancer prediction models.
Regarding claim 3, Drake, Shi, and Kotlov teach the invention in claim 1, as discussed above, and further teach wherein the first-level learning model establishment unit further configured to: input a third gender and a third CNV of each of a plurality of third learning samples to the first-level learning model (Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”, Drake [0237] “As described above, a large set of features can be generated to provide a feature space from which a feature vector can be determined. This feature vector from each of a set of training samples can then be used for training a current version of the machine learning model. The types of features used can depend on the types of analytes used.” Drake [0238] “Examples of features can include variables related to structural variations (SVs), such as a copy number variation and translocations; fusions; mutations (e.g., SNPs or other single nucleotide variations (SNVs), or slightly larger sequence variations); telomere attrition; and nucleosome occupancy and distribution. These features can be calculated genomewide. Example classes (types) of features are provided below. When genetic sequence data is obtained from at least one of die analytes, example features can include aligned features (e.g., a comparison with one or more reference genomes) and non- aligned features. Example aligned features can include sequence variations and sequence counts in a genomic window. Example non-aligned features can include kmers from sequence reads and biological derived information from the reads.”, and
Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”,
so that the first-level learning model outputs a plurality of second output cancer types of each of the third learning samples (Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”.);
wherein the cancer type prediction system further comprises: a second-level second learning model establishment unit configured to: establish a second-level second learning model, by using machine learning technology, according to the second output cancer types and a third sample cancer type of each third learning sample (Kotlov, page 28, “In some embodiments, the hierarchy of machine learning classifiers (e.g., hierarchy of DNA-based machine learning classifiers or the hierarchy of RNA-based machine learning classifiers) may be stored in any suitable way. Each of the machine learning classifiers may comprise program code that, when executed, performs classification using the machine learning classifier’s inputs, the machine learning classifier’s parameters, the machine learning classifier’s hyperparameters, and/or any other suitable configuration information. The hierarchical relationships among the machine learning classifiers may be stored using one or more data structures having one or more fields storing information about the hierarchy. For example, the fields may store information indicating, for each machine learning classifier in the hierarchy, its relationship to one or more other machine learning classifiers in the hierarchy (e.g., indicating a parent machine learning classifier and/or one or more child machine learning classifiers), a respective category in the hierarchy of molecular categories to which the classifier corresponds, and/or any other suitable information, as aspects of the technology described herein are not limited in this respect.”, and Kotlov, page 26, “Another aspect of the approach developed by the inventors that contributes to its accuracy and robustness is the use of features (e.g., features derived from DNA and/or RNA expression data, which features may include the DNA and/or RNA expression data itself, in some embodiments) specified a priori for each molecular category to determine whether to identify the molecular category as a candidate molecular category for the biological sample. For example, RNA expression data for a specific set of genes for a particular molecular category may be processed using a machine learning classifier trained to predict whether a particular molecular category should be identified for the biological sample. The RNA expression data may be first processed to obtain a set of features specified a priori for the particular molecular category (e.g., gene rankings for a set of genes associated with the molecular category, the gene rankings obtained by numerically ranking the expression levels for genes in the set of genes) and this set of features may be provided as input to a specific machine learning classifier for that specific molecular category. As another example, DNA expression data may be used to obtain a specific set of DNA features (e.g., features indicating the presence of gene mutations, presence of genes, copy number alterations, loss of heterozygosity (LOH), ploidy, tumor mutational burden, presence of gene fusions, micro satellite instability (MSI) status, etc.) for a particular molecular category. Then these DNA features may be provided as input to and be processed using a machine learning classifier trained to predict whether the molecular category is a candidate molecular category for the biological sample. In some embodiments, the use of specific features tailored for each particular molecular category allows the techniques developed by the inventors to leverage domain- specific knowledge to distinguish among molecular categories, even when they share similar molecular features, contributing to the success of the techniques described herein. Examples of RNA and DNA features used by RNA-based and DNA-based machine learning classifiers, respectively, are provided herein.”)., and
Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”., and
Drake [0306] “The methods and systems described herein are applicable to various cancer types, similar to grade and stage, and as such, is not limited to a single cancer disease type. Therefore, combinations of analytes and assays may be used in the present systems and methods to predict responsiveness of cancer therapeutics across different cancer types in different tissues and classifying individuals based on treatment responsiveness. In one example, the classifiers described herein are capable of stratifying a group of individuals into treatment responders and non-responders.”, and Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”);
wherein the third genders of these persons to whom all of the second learning samples belongs are male or female (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to modify the machine learning cancer prediction system of Drake to include separate second-level learning models for different gender specific populations in view of the teachings of Shi and Ketlow. Drake teaches training machine learning models using cancer-related biological samples and CNV features to generate cancer classifications, while Shi teaches utilizing gender information in cancer prediction models, generating multiple candidate cancer type outputs, and developing separate machine learning models for male and female patients, which demonstrate the benefits of gender specific predictive modeling. Ketlow further teaches a hierarchical machine learning architecture comprising parent and child classifiers and category specific machine learning classifiers configured to refine classifications within a hierarchy. A person of ordinary skill in the art would have recognized that it would have been advantageous to extend the hierarchical classifier framework of Ketlow to the gender specific cancer prediction models of Shi and Drake by establishing an additional second-level learning model for another gender specific training population and training that model using cancer type outputs generated by a first-level classifier. Such a modification would have represented the predictable use of known machine learning stratification and hierarchical classification techniques to account for gender dependent differences in cancer characteristics, which improve the accuracy of cancer type predictions.
Regarding claim 4, Drake, Shi, and Kotlov teach the invention in claim 3, as discussed above, and further teach wherein the second-level second learning model establishment unit further configured to: establish the second-level second learning model according to an age of each third learning sample; wherein the second-level first learning model establishment unit further configured to: establish the second-level first learning model according to an age of each second learning sample (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to establish the second-level first learning model and the second-level second learning model according to the age of the corresponding learning samples in view of the teachings of Shi. Shi teaches obtaining patient datasets comprising age, gender, and biomarker features, selecting age as an input into a machine learning system, and generating classifier models based on training data that includes age information. Shi further teaches training classifier models, including neural network models, using age and gender as input values. A person of ordinary skill in the art would have recognized that age is a known predictive factor in cancer diagnosis and risk assessment and therefore would have found it obvious to incorporate the age of the second and third learning samples when training the respective second-level learning models in order to improve prediction accuracy, account for age differences in cancer characteristics, and enhance the overall performance of the machine learning classifiers. This use of age as an additional training parameter represents nothing more than the predictable use of a known patient attribute according to its established function in cancer prediction modeling.
Regarding claim 5, Drake, Shi, and Kotlov teach the invention in claim 1, as discussed above, and further teach a cancer type prediction system, comprising: a storage unit configured to: store the first-level learning model and the second-level first learning model (Drake [0223] Methods as described herein can be implemented by way of machine (e.g., computer processor) executable code stored on an electronic storage location of the computer system 101, such as, for example, on the memory 110 or electronic storage unit 115. The machine executable or machine-readable code can be provided in tire form of software. During use, the code can be executed by the CPU 105. In some cases, the code can be retrieved from the storage unit 115 and stored on the memory' 110 for ready access by the CPU 105. In some situations, the electronic storage unit 115 can be precluded, and machine- executable instructions are stored on memory 110. [0224] The code can be pre-compiled and configured for use with a machine having a processer adapted to execute the code or can be compiled during runtime. The code can be supplied in a programming language that can be selected to enable tire code to execute in a pre-compiled or as- compiled fashion.”, and
Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”, Drake [0237] “As described above, a large set of features can be generated to provide a feature space from which a feature vector can be determined. This feature vector from each of a set of training samples can then be used for training a current version of the machine learning model. The types of features used can depend on the types of analytes used.” and
Kotlov, page 28, “In some embodiments, the hierarchy of machine learning classifiers (e.g., hierarchy of DNA-based machine learning classifiers or the hierarchy of RNA-based machine learning classifiers) may be stored in any suitable way. Each of the machine learning classifiers may comprise program code that, when executed, performs classification using the machine learning classifier’s inputs, the machine learning classifier’s parameters, the machine learning classifier’s hyperparameters, and/or any other suitable configuration information. The hierarchical relationships among the machine learning classifiers may be stored using one or more data structures having one or more fields storing information about the hierarchy. For example, the fields may store information indicating, for each machine learning classifier in the hierarchy, its relationship to one or more other machine learning classifiers in the hierarchy (e.g., indicating a parent machine learning classifier and/or one or more child machine learning classifiers), a respective category in the hierarchy of molecular categories to which the classifier corresponds, and/or any other suitable information, as aspects of the technology described herein are not limited in this respect.”);
a prediction unit configured to: obtain a plurality of first prediction cancer types of the to-be-tested sample by inputting a sample CNV and a sample gender of a to-be-tested sample to the first-level learning model (Drake [0028] In a fourth aspect, the present disclosure provides a method of detecting presence of cancer in an individual comprising: a) assaying a plurality of classes of molecules in a biological sample obtained from the individual wherein the assaying provides a plurality of sets of measured values representative of the plurality of classes of molecules, b) identifying a set of features corresponding to properties of each of the plurality of classes of molecules to be input to a machine learning model, c) preparing a feature vector of feature values from each of the plurality of sets of measured values, each feature value corresponding to a feature of the set of features and including one or more measured values, wherein the feature vector includes at least one feature value obtained using each set of the plurality of sets of measured values, d) loading into a memory of a computer system a machine learning model that is trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified from individuals with cancer and a second subset of the training biological samples identified from individuals not having cancer, e) inputting the feature vector into the machine learning model to obtain an output classification of whether the biological sample is associated with the cancer, thereby detecting the presence of the cancer in the individual.”, Drake [0237] “As described above, a large set of features can be generated to provide a feature space from which a feature vector can be determined. This feature vector from each of a set of training samples can then be used for training a current version of the machine learning model. The types of features used can depend on the types of analytes used.” Drake [0238] “Examples of features can include variables related to structural variations (SVs), such as a copy number variation and translocations; fusions; mutations (e.g., SNPs or other single nucleotide variations (SNVs), or slightly larger sequence variations); telomere attrition; and nucleosome occupancy and distribution. These features can be calculated genomewide. Example classes (types) of features are provided below. When genetic sequence data is obtained from at least one of die analytes, example features can include aligned features (e.g., a comparison with one or more reference genomes) and non- aligned features. Example aligned features can include sequence variations and sequence counts in a genomic window. Example non-aligned features can include kmers from sequence reads and biological derived information from the reads.”, and
Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.” and Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”.);
determine whether a sample gender of the to-be-tested sample is the second gender (Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”); and when the sample gender of the to-be-tested sample is the second gender, obtain a second prediction cancer type of the to-be-tested sample by inputting the first prediction cancer type of the to-be-tested sample to the second-level first learning model (Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”. Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”, and
Kotlov, page 28, “In some embodiments, the hierarchy of machine learning classifiers (e.g., hierarchy of DNA-based machine learning classifiers or the hierarchy of RNA-based machine learning classifiers) may be stored in any suitable way. Each of the machine learning classifiers may comprise program code that, when executed, performs classification using the machine learning classifier’s inputs, the machine learning classifier’s parameters, the machine learning classifier’s hyperparameters, and/or any other suitable configuration information. The hierarchical relationships among the machine learning classifiers may be stored using one or more data structures having one or more fields storing information about the hierarchy. For example, the fields may store information indicating, for each machine learning classifier in the hierarchy, its relationship to one or more other machine learning classifiers in the hierarchy (e.g., indicating a parent machine learning classifier and/or one or more child machine learning classifiers), a respective category in the hierarchy of molecular categories to which the classifier corresponds, and/or any other suitable information, as aspects of the technology described herein are not limited in this respect.”, and Kotlov, page 26, “Another aspect of the approach developed by the inventors that contributes to its accuracy and robustness is the use of features (e.g., features derived from DNA and/or RNA expression data, which features may include the DNA and/or RNA expression data itself, in some embodiments) specified a priori for each molecular category to determine whether to identify the molecular category as a candidate molecular category for the biological sample. For example, RNA expression data for a specific set of genes for a particular molecular category may be processed using a machine learning classifier trained to predict whether a particular molecular category should be identified for the biological sample. The RNA expression data may be first processed to obtain a set of features specified a priori for the particular molecular category (e.g., gene rankings for a set of genes associated with the molecular category, the gene rankings obtained by numerically ranking the expression levels for genes in the set of genes) and this set of features may be provided as input to a specific machine learning classifier for that specific molecular category. As another example, DNA expression data may be used to obtain a specific set of DNA features (e.g., features indicating the presence of gene mutations, presence of genes, copy number alterations, loss of heterozygosity (LOH), ploidy, tumor mutational burden, presence of gene fusions, micro satellite instability (MSI) status, etc.) for a particular molecular category. Then these DNA features may be provided as input to and be processed using a machine learning classifier trained to predict whether the molecular category is a candidate molecular category for the biological sample. In some embodiments, the use of specific features tailored for each particular molecular category allows the techniques developed by the inventors to leverage domain- specific knowledge to distinguish among molecular categories, even when they share similar molecular features, contributing to the success of the techniques described herein. Examples of RNA and DNA features used by RNA-based and DNA-based machine learning classifiers, respectively, are provided herein.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to modify the cancer prediction system of Drake in view of the teachings of Shi and Ketlow to store multiple machine learning models, generate initial cancer type predictions using CNV and gender information, select a model based upon the gender of a patient, and further refine the prediction using a second-level classifier. Drake teaches storing machine learning models in computer memory and electronic storage units and further teaches generating cancer classifications using machine learning models trained on biological sample data including copy number variation features. Shi teaches utilizing gender as a predictive feature, developing separate models for male and female patients, and generating multiple candidate cancer type predictions for a sample. Ketlow teaches a hierarchical machine-learning architecture comprising parent and child classifiers and category specific classifiers arranged in a classifier hierarchy. A person of ordinary skill in the art would have recognized that combining Shi's gender specific cancer prediction models with Ketlow's hierarchical classifier framework would have predictably improved prediction accuracy by allowing an initial cancer type prediction generated by a first-level classifier to be further evaluated by a second-level classifier selected according to the patient's gender. This modification merely applies known machine learning techniques according to their established functions to refine candidate cancer classifications and improve the accuracy of cancer type prediction, yielding the predictable result of more accurate and specialized cancer classifications.
Regarding claim 6, Drake, Shi, and Kotlov teach the invention in claim 5, as discussed above, and further teach wherein the storage unit further configured to store a second-level second learning model, and the prediction unit configured to: determine whether the sample gender of the to-be-tested sample is a third gender (Drake [0223] Methods as described herein can be implemented by way of machine (e.g., computer processor) executable code stored on an electronic storage location of the computer system 101, such as, for example, on the memory 110 or electronic storage unit 115. The machine executable or machine-readable code can be provided in tire form of software. During use, the code can be executed by the CPU 105. In some cases, the code can be retrieved from the storage unit 115 and stored on the memory' 110 for ready access by the CPU 105. In some situations, the electronic storage unit 115 can be precluded, and machine- executable instructions are stored on memory 110. [0224] The code can be pre-compiled and configured for use with a machine having a processer adapted to execute the code or can be compiled during runtime. The code can be supplied in a programming language that can be selected to enable tire code to execute in a pre-compiled or as- compiled fashion.”, and
Shi [0014] “Disclosed herein are classifier models, machine learning systems, computer implemented systems and methods thereof. In some embodiments, this disclosure provides a computer-implemented method(s) for generating a classifier model comprising: a) obtaining, by one or more processors, a data set comprising, age, gender and biomarker features of a patient, wherein the biomarker features comprise a panel of pan and/or specific tumor biomarkers, wherein the biomarker features are from populations of patients, and wherein each population is labeled with a diagnostic indicator; b) selecting the panel of biomarker features, age, gender and diagnostic indicator as inputs into a machine learning system, wherein the input for each biomarker feature has a measured value or is absent for the population of patients; c) randomly partitioning the data set in training data and validation data; d) generating a first classifier model using a machine learning system based on the training data and the inputs, wherein each input has an associated weight, and wherein the classifier model provides binary outcomes selected from increased risk of having cancer or developing cancer above a pre-determined threshold or no increased risk of having or developing cancer below a pre-determined threshold; and, e) providing the classifier model to a user to predict an increased risk of having or developing cancer.”, and Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”); and
when the sample gender of the to-be-tested sample is the third gender, obtain the second prediction cancer type of the to-be-tested sample by inputting the first prediction cancer type of the to-be-tested sample to the second-level second learning model (Shi [00202] “The classifier model of Example 1 was trained using logistic regression (LR), with input data from each patient sample of age and a panel of 6 or 7 seven measured biomarkers, wherein separate models were developed for male and female patients. That model demonstrated a significant improvement as compared to a single marker measurement. See Example 4 and 5. However, a limitation of that model is that to use, a patient must have all of the same biomarkers measured as was used to train the classifier model. Some trained models were based on gender, which means gender was not an input value. The classifier model of this example was trained using a neural network (LSTM) with input values of age, gender and one or more measured biomarker values (see Tables 10 and 11 below). In this system, a biomarker that was not measured was assigned a value of zero and entered as an input. In this way, the new classifier model of this example can be used with a wide range of data, provided patient data of age, gender and at least one of the measured biomarkers is one that was used to train the classifier model, and for any marker not measured a value of zero is assigned as the input value.”. Shi [00157] “Using the entire cohort of cancer subjects (n=186) and the same six (or 5 for female individuals) biomarker measurements along with age and gender, we applied a model comprising a pattern recognition algorithm, and a k-Nearest Neighbors algorithm (kNN) employing a leave-one-out evaluation method to predict the top 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 cancers for each sample. The accuracies are reported in Table 5 and reflect the percentage of cases of each cancer type that were found in the top N (N=10 for Table 5) predicted cancers. Clearly, the accuracy of prediction varies based on both the cancer type and to some extent based on the number of cases of that type found in the dataset.”, and Shi [00158] “As such, it was decided to classify cancers more broadly based on organ system considering that would suggest the specialist to whom the patient should be referred. A similar analysis was performed, and the overall results interpreted. A balanced sensitivity and specificity are achieved when the Top three most likely affected organ systems are reported. To a large extent the accuracies/ sensitivities best reflect both the number of overall cases of a given cancer type in the dataset (i.e. Gastro-Intestinal (GI) and Genitourinary (GU) cancers vs. dermatological cancers) as well the nature of the biomarkers (e.g. PSA is specific for prostate and therefore GU”, and
Kotlov, page 28, “In some embodiments, the hierarchy of machine learning classifiers (e.g., hierarchy of DNA-based machine learning classifiers or the hierarchy of RNA-based machine learning classifiers) may be stored in any suitable way. Each of the machine learning classifiers may comprise program code that, when executed, performs classification using the machine learning classifier’s inputs, the machine learning classifier’s parameters, the machine learning classifier’s hyperparameters, and/or any other suitable configuration information. The hierarchical relationships among the machine learning classifiers may be stored using one or more data structures having one or more fields storing information about the hierarchy. For example, the fields may store information indicating, for each machine learning classifier in the hierarchy, its relationship to one or more other machine learning classifiers in the hierarchy (e.g., indicating a parent machine learning classifier and/or one or more child machine learning classifiers), a respective category in the hierarchy of molecular categories to which the classifier corresponds, and/or any other suitable information, as aspects of the technology described herein are not limited in this respect.”, and Kotlov, page 26, “Another aspect of the approach developed by the inventors that contributes to its accuracy and robustness is the use of features (e.g., features derived from DNA and/or RNA expression data, which features may include the DNA and/or RNA expression data itself, in some embodiments) specified a priori for each molecular category to determine whether to identify the molecular category as a candidate molecular category for the biological sample. For example, RNA expression data for a specific set of genes for a particular molecular category may be processed using a machine learning classifier trained to predict whether a particular molecular category should be identified for the biological sample. The RNA expression data may be first processed to obtain a set of features specified a priori for the particular molecular category (e.g., gene rankings for a set of genes associated with the molecular category, the gene rankings obtained by numerically ranking the expression levels for genes in the set of genes) and this set of features may be provided as input to a specific machine learning classifier for that specific molecular category. As another example, DNA expression data may be used to obtain a specific set of DNA features (e.g., features indicating the presence of gene mutations, presence of genes, copy number alterations, loss of heterozygosity (LOH), ploidy, tumor mutational burden, presence of gene fusions, micro satellite instability (MSI) status, etc.) for a particular molecular category. Then these DNA features may be provided as input to and be processed using a machine learning classifier trained to predict whether the molecular category is a candidate molecular category for the biological sample. In some embodiments, the use of specific features tailored for each particular molecular category allows the techniques developed by the inventors to leverage domain- specific knowledge to distinguish among molecular categories, even when they share similar molecular features, contributing to the success of the techniques described herein. Examples of RNA and DNA features used by RNA-based and DNA-based machine learning classifiers, respectively, are provided herein.”).
It would have been obvious to one of ordinary skill in the art at the time of the invention to modify the cancer prediction system of Drake in view of the teachings of Shi and Ketlow to store an additional second-level learning model, determine whether a patient belongs to a particular gender specific population, and utilize a corresponding second-level classifier to refine cancer predictions. Drake teaches storing machine learning models in memory and electronic storage units and generating cancer classifications from biological sample data using machine learning techniques. Shi teaches utilizing gender information in cancer prediction systems, developing separate machine learning models for male and female patients, and generating multiple candidate cancer type predictions for a patient sample. Ketlow further teaches a hierarchical machine learning architecture comprising parent and child classifiers and category specific classifiers configured to refine candidate classifications within a classifier hierarchy. A person of ordinary skill in the art would have recognized that once separate male and female prediction models were known, it would have been advantageous to store and employ an additional gender specific second-level classifier corresponding to another gender population and to invoke that classifier when the patient is determined to belong to that gender. Combining Shi's gender specific modeling with Ketlow's hierarchical classifier framework would have predictably allowed initial cancer-type predictions generated by a first-level classifier to be further refined through a second-level classifier tailored to a particular gender-specific population, thereby improving prediction accuracy and classification specificity. This modification represents the predictable use of known machine learning stratification and hierarchical classification techniques according to their established functions to obtain improved cancer type prediction results.
Claim 7 is analogous to claim 1, thus claim 7 is similarly analyzed and rejected in a manner consistent with the rejection of claim 1.
Claim 8 is analogous to claims 2 and 3, thus claim 8 is similarly analyzed and rejected in a manner consistent with the rejection of claims 2 and 3.
Claim 9 is analogous to claims 3 and 4, thus claim 9 is similarly analyzed and rejected in a manner consistent with the rejection of claims 3 and 4.
Claim 10 is analogous to claim 5, thus claim 10 is similarly analyzed and rejected in a manner consistent with the rejection of claim 5.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Cohen et al. (International Publication No. WO 2020/006547 A1) teaches a machine learning system and method for classifying patients according to their risk of having or developing cancer, and for higher risk patients, predicting the associated organ system or specific cancer type.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KYRA R LAGOY whose telephone number is (703)756-1773. The examiner can normally be reached Monday - Friday, 8:00 am - 5:00 pm EST.
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, Kambiz Abdi can be reached at (571)272-6702. 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.
/K.R.L./
Examiner, Art Unit 3685
/KAMBIZ ABDI/Supervisory Patent Examiner, Art Unit 3685
Prosecution Insights