DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Notice to Applicant
Receipt of Applicant’s Amendment filed May 26, 2026 is acknowledged.
Response to Amendment
Claims 1-2, 7, 9, 14-15, and 18-20 have been amended. Claims 10-11 and 21 have not been modified. Claims 3-6, 8, 12-13, 16-17, and 22-27 have been cancelled. Claims 1-2, 7, 9-11, 14-15, and 18-21 are pending and are provided to be examined upon their merits.
Response to Arguments
Applicant’s arguments filed May 26, 2026 have been fully considered but they are not persuasive. A response is provided below.
Applicant argues 35 U.S.C. §103 Rejections, pg. 12 of Remarks:
Applicant argues that the prior cited art of record are insufficient to overcome the amended independent claims, specifically the “forward-in-time temporal relationship between the subsets of the set of data represented by two nodes”. Applicant argument is moot as new art is applied to address the amended claim limitations. Please see the modified 103 rejection below.
Applicant argues 35 U.S.C. §101 Rejections, pg. 13 of Remarks:
Regarding A and B, Applicant argues that independent claims 1 and 14 are not abstract and even if they were, they provide technological improvement to GNN technology. Examiner maintains that the claims are abstract, but agrees that they provide technological improvement to GNN technology in light of the amendments and paragraph [0066] of Applicant specification, which describes an improvement to GNN technology by improving the function of the GNN by “allow[ing] the GNN to learn from both the temporal dependency and knowledge about the type of observations”. Thus, the 35 U.S.C. 101 rejection is withdrawn from claims 1-2, 14-15, and 20.
Regarding C, Applicant argues that independent claims 7 and 18 are not abstract as there is no evidence that a veterinarian would perform “A computer implemented method of making an animal health prediction using a trained multi-layer graph neural network (GNN) model”. While a veterinarian may not regularly perform a method using a trained multi-layer graph neural network (GNN) model, the model is applied to perform animal health predictions, specifically “predict[ing] whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period”. As this prediction is an activity that is typically performed by veterinarians, the method is considered abstract for managing the personal behaviors of veterinarians. Regarding mathematical concepts, Examiner agrees and withdraws the grouping.
Regarding D, Applicant argues that independent claims 7 and 18 provide an improvement to GNN technologies. Examiner disagrees as the inputting a representation of a graph into a GNN model is simply inputting data into an already trained model. No specific, technical improvements are made to GNN technologies as they are only being applied to perform the abstract function of predicting animal health.
However, Examiner notes that claim 20 is not subject to the 35 U.S.C. 101 for the same reasons as described in the response to points A and B above.
Claim Interpretation
Examiner notes that the performance of the steps of claim 20 are not limited to performance prior to the steps of claim 18, as a second (and third) electronic database and second set of data is introduced by the preamble (“prior to obtaining, from an electronic database, a set of data: obtaining, from an electronic database,…”). Instead, under the broadest reasonable interpretation, the training of the graph neural network may occur at any time prior to obtaining any set of data from a second electronic database. However, the Examiner will interpret the claim as being performed prior to performance of the obtaining of set of data from the electronic database recited in claim 18 as claim 20 refers to the same multi-layer GNN model that is applied for making an animal health prediction in claim 18.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 9-11 and 19-21 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 9 and 19 recite “the graph adjacency graph”. There is insufficient antecedent basis for this limitation in the claim. Claims 10-11 and 21 are rejected by virtue of their dependency on claims 9 and 19.
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 7, 9-11, 18-19, and 21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
Subject Matter Eligibility Criteria – Step 1:
The claims recite subject matter within a statutory category as a process and a machine
(claims 7, 9-11, 18-19, and 21). Accordingly, claims 7, 9-11, 18-19, and 21 are all within at least one of the four statutory categories.
Subject Matter Eligibility Criteria – Step 2A – Prong One:
Regarding Prong One of Step 2A of the Alice/Mayo test, the claim limitations are to be analyzed to determine whether, under their broadest reasonable interpretation they “recite” a judicial exception or in other words whether a judicial exception is “set forth” or “described” in the claims. MPEP §2106.04(II)(A)(1). An “abstract idea” judicial exception is subject matter that falls within at least one of the following groupings: a) certain methods of organizing human activity, b) mental processes, and /or c) mathematical concepts. MPEP §2106.04(a).
The Examiner has identified method claim 7 as the claim that represents the claimed invention for analysis, and is similar to system claim 18.
Claim 7:
A computer implemented method of making an animal health prediction using a trained multi-layer graph neural network (GNN) model, the method comprising:
obtaining, from an electronic database, a set of data for an animal subject over a time period, the set of data including: (i) clinical observation data, (ii) body weight measurement data, (iii) outcome status data, (iv) veterinary treatment record data, or (v) any combination thereof;
encoding the set of data for the animal subject using a graph having nodes and directed edges between pairs of the nodes, wherein each of the nodes is associated with a respective timepoint in the time period and represents a subset of the set of data collected at the respective timepoint and wherein a directed edge between two nodes represents a forward-in-time temporal relationship between the subsets of the set of data represented by the two nodes;
generating a plurality of node embeddings by embedding the subsets of the set of data associated with nodes in the graph,
processing the plurality of node embeddings and a representation of the graph using a trained multi-layer graph neural network (GNN) model to predict whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period, wherein the trained multi-layer GNN model was previously trained to predict, from the plurality of node embeddings and a representation of the graph, whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period; and
outputting a classification based on the result for the animal subject.
These above limitations under their broadest reasonable interpretation, also cover performance of the limitation as certain methods of organizing human activity. The claim elements are directed towards obtaining sets of data for a plurality of animal subjects and training a multi-layer GNN model “to predict whether health of an animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period”. These claim elements are directed towards aiding in patient diagnoses. Diagnosing an animal subject falls under the abstract concept of managing personal behaviors of people, as it is a human activity regularly performed by veterinary doctors for their patients.
Accordingly, the claim recites at least one abstract idea.
Claim 18 is abstract for similar reasons.
Subject Matter Eligibility Criteria – Step 2A – Prong Two:
Regarding Prong Two of Step 2A of the Alice/Mayo test, it must be determined whether the claim as a whole integrates the idea into a practical application. As noted at MPEP §2106.04 (ID)(A)(2), it must be determined whether any additional elements in the claim beyond the abstract idea integrate the exception into a practical application in a manner that imposes a meaningful limit on the judicial exception. The courts have indicated that additional elements merely using a computer to implement an abstract idea, adding insignificant extra solution activity, or generally linking use of a judicial exception to a particular technological environment or field of use of a judicial exception to a particular technological environment or field of use do not integrate a judicial exception into a “practical application.” MPEP §2106.05(I)(A).
Additional elements cited in the claims:
Trained multi-layer graph neural network (GNN) model (7,18); electronic database (7,18); graphical user interface (11,21); one or more data processors (18,21); non-transitory computer readable storage medium (18)
Any computing devices and their components (data processors and non-transitory computer readable storage medium) that would be able to perform the method and the modules that are used within the computing environment are taught at a high level of generality such that the claim elements amounts to no more than mere instructions to apply the exception using any generic component capable of performing the claim limitations. [0104] of Applicant specification recites: “Systems, methods, and data structures described herein are operational with numerous other general purpose or special purpose computing system environments or configurations. Examples of known computing systems, environments, and/or configurations that may be suitable include, but are not limited to, personal computers, server computers, thin clients, thick clients, hand-held or laptop devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like.” [0079] further recites: “hardware such as one or more processors (e.g., a CPU, GPU, TPU, FPGA, the like, or any combination thereof), memory, and storage that operates software or computer program instructions (e.g., TensorFlow, PyTorch, Keras, and the like)” No specific, technical improvements are being made to computing devices as generic devices with software modules are simply being used to perform the abstract idea of animal health prediction.
Trained graph neural network models are also taught at a high level of generality. Claims 7 and 14 recite: “using a trained multi-layer graph neural network (GNN) model”. No specific, technical improvements are being made to GNN technologies as an already trained model is simply applied to perform the abstract idea of animal health prediction.
Databases are also taught at a high level of generality. [0065] recites: “Data are collected at a daily, sub-daily, or on- demand frequency by trained staff and typically stored in a relational database. In some instances, the observations made by the trained staff are entered into a database via a GUI where an entry is comprised a single observation made for a single animal subject. The database contains information about each animal subject and is delimited by a unique identifier (e.g., PRETEST_NUMBER in Table 1) such that trained staffed can enter clinical observations for that specific animal subject.” No specific, technical improvements are made to database technologies as they are only applied to perform an insignificant extra-solution activity of storing data.
Graphical user interfaces are also taught at a high level of generality. [0018] recites: “the computer-implement method further comprises providing the classification and/or the recommendation to a user through a graphical user interface (GUI).” [0097] further recites: “The set of data may be input into the machine-learning model via a graphical user interface (GUI).” No specific, technical improvements are being made to graphical user interfaces as they are applied to perform the insignificant extra-solution activities of simply presenting and receiving data from a user.
Thus, taken alone, the additional elements do not integrate the at least one abstract idea into a practical application.
Looking at the additional elements as an ordered combination adds nothing that is not already present when looking at the elements taken individually. For instance, there is no indication that the additional elements, when considered as a whole with the limitations reciting the at least one abstract idea, reflect an improvement in the functioning of a computer or an improvement to another technology or technical field, apply or use the above-noted judicial exception with a particular machine or manufacture that is integral to the claim, effect a transformation or reduction of a particular article to a different state or thing, or apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole does not integrate the abstract idea into a practical application of the abstract idea. MPEP §2106.05(I)(A) and §2106.04(IID)(A)(2).
The remaining dependent claim limitations not addressed above fail to integrate the abstract idea into a practical application as set forth below:
Claims 9 and 19: These claims recite wherein processing the plurality of node embeddings and a representation of the graph adjacency graph into using the trained multi- layer comprises determining a likelihood indicating whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period; and wherein outputting the classification comprises comparing the likelihood to a determined threshold and classifying, based on the comparison, the animal subject as having a normal health status, requiring veterinary attention in the upcoming time period, being likely to experience the unplanned death outcome in the upcoming time period, or likely to receive the treatment in the upcoming time period; which teaches an abstract idea of certain methods of organizing human activity by determining if an animal subject requires veterinary attention or is likely to receive a treatment, which is an activity typically performed by veterinarians.
Claim 10: This claim recites further comprising providing a recommendation based on the classification of the animal subject; which teaches an abstract idea of certain methods of organizing human activity, as providing a recommendation is analogous to teaching.
Claims 11 and 21: These claims recite the method further comprising providing the classification and/or the recommendation to a user through a graphical user interface (GUI); which teaches a graphical user interface at a high level of generality.
Subject Matter Eligibility Criteria – Step 2B:
Regarding Step 2B of the Alice/Mayo test, representative independent claims do not include additional elements (considered both individually and as an ordered combination) that are sufficient to amount to significantly more than the judicial exception for reasons the same as those discussed above with respect to determining that the claim does not integrate the abstract idea into a practical application.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to discussion of integration of the abstract idea into a practical application, the additional elements amount to no more than mere instructions to apply an exception, add insignificant extra-solution activity to the abstract idea, and generally link the abstract idea to a particular technological environment or field use. Additionally, the additional limitations, other than the abstract idea per se, amount to no more than limitations which:
Amount to elements that have been recognized as activities in particular fields (such as Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information), MPEP §2106.05(d)(II)(i);storing and retrieving information in memory, Versata Dev. Group, MPEP §2106.05(d)(II)(iv)).
Examiner notes multi-layer GNN models for disease prognosis are known. Thus, the application of a multi-layer GNN in the instant application is analogous to applying known elements to the abstract idea of predicting animal health. See the following examples:
Sanchez-Lengeling; Benjamin, A Gentle Introduction to Graph Neural Networks, 2 Sep 2021, Distill: Pg. 3, “A way of visualizing the connectivity of a graph is through its adjacency matrix. We order the nodes, in this case each of 25 pixels in a simple 5x5 image of a smiley face, and fill a matrix of n nodes × n nodes n nodes × n nodes with an entry if two nodes share an edge.”
Sun; Zhenchao, Disease Prediction via Graph Neural Networks, 22 Jun 2020, IEEE Journal of Biomedical and Health Informatics: Pg. 818, “we introduce an innovative model based on Graph Neural Networks (GNNs) for disease prediction, which utilizes external knowledge bases to augment the insufficient EMR data, and learns highly representative node embeddings for patients, diseases and symptoms from the medical concept graph and patient record graph respectively constructed from the medical knowledge base and EMRs.”
Lu; Haohui, Disease Prediction Using Graph Machine Learning Based on Electronic Health Data: A Review of Approaches and Trends, 4 Apr 2023, MDPI healthcare: Pg. 1, “Herein, a review of graph ML methods and their applications in the disease prediction domain based on electronic health data is presented in this study from two levels: node classification and link prediction. Commonly used graph ML approaches for these two levels are shallow embedding and graph neural networks (GNN). This study performs comprehensive research to identify articles that applied or proposed graph ML models on disease prediction using electronic health data.”
Dependent claims recite additional subject matter which, as discussed above with respect to integration of the abstract idea into a practical application, amount to invoking computers as a tool to perform the abstract idea. Dependent claims recite additional subject matter which amount to limitations consistent additional subject matter which amount to limitations consistent with the additional elements in the independent claims (such as claims 9-11, 19, and 21 additional limitations which amount to elements that have been recognized as activities in particular fields, claims 9-11, 19, and 21, e.g., performing repetitive calculations, Flook, MPEP §2106.05(d)(II)(ii); claims 9-11, 19, and 21, e.g., storing and retrieving information in memory, Versata Dev. Group, MPEP §2106.05(d)(II)(iv). Looking at the limitations as an ordered combination adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
Therefore, whether taken individually or as an ordered combination, claims 7, 9-11, 18-19, and 21 are nonetheless rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Regarding 35 U.S.C. 101 for claims 1-2, 14-15, and 20
The claims recite an abstract idea of mental processes as claims 1, 14, and 20 are directed towards training a multi-layer graph neural network (GNN) model by obtaining sets of data, generating training datasets, encoding the set of data, generating node embeddings, and training the model using the training datasets, similar to claim 2 of Example 47 (Step 2A, Prong One: Yes). However, the claims recite an improvement in GNN technology by means of training using forward-in-time temporal relationships between subsets of the set of data represented by the nodes, which improves the function of the GNN by “allow[ing] the GNN to learn from both the temporal dependency and knowledge about the type of observations” ([0066] of Applicant specification)(Step 2A, Prong Two: Yes).
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 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, 7, 14, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Baranzini (US 20240303544) in view of Liu (Liu; Chuanren, Temporal Phenotyping from Longitudinal Electronic Health Records: A Graph Based Framework, 10 Aug 2015, KDD '15: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Pages 705 – 714).
Regarding claim 1, Baranzini teaches computer implemented method of training a multi-layer graph neural network (GNN) model for animal health prediction ([0139], “model 1135 can be a neural network that comprises a number of neurons (e.g., Adaptive basis functions) organized in layers. The training of the neural network can iteratively search for the best configuration of the parameter of the neural network for feature recognition and classification performance. Various numbers of layers and nodes may be used.” [0078], “Process 700 can use a graph database (e.g., SPOKE) to obtain training vectors for training a machine learning model.” [0095], “The graph database may include M nodes, where M is greater than N. Entry nodes can correspond to any medically relevant concept, including diseases, genes, symptoms, or drugs that appear in fields of entity records that are to be classified.”) comprising:
obtaining, from an electronic database, sets of data for a plurality of animal subjects over a time period, wherein the sets of data comprise: (i) clinical observation data, (ii) body weight measurement data, (iii) outcome status data, (iv) veterinary treatment record data, or (v) any combination thereof ([0064], “the computer system may receive, for a plurality of entities, a plurality of entity records (e.g., EHRs), each with a plurality of fields and a known classification, as described herein. Example entities include patients, drugs, and animal subjects.” [0055], “an entity record (e.g., EHR) including a plurality of fields is received. Entity records may be records related to medical treatment and entity record fields can include information related to diagnosis and treatment, such as test results, drug prescriptions, symptoms, or diagnoses.”);
generating training datasets for the plurality of animal subjects ([0004], “The techniques include using a graph database (e.g., SPOKE) to generate training vectors (also referred to herein as SPOKEsigs) for each biological entity with a known classification.”), wherein generating a particular training dataset for a particular animal subject comprises processing the set of data for the particular animal subject by:
encoding the set of data for the particular animal subject using a graph having nodes and directed edges between pairs of the nodes ([0005], “As part of the training, a graph database is stored comprising (1) M nodes of a plurality of node types and (2) a plurality of edges of a plurality of edge types. A plurality of entity records are received for a plurality of entities, each with a plurality of fields and a known classification. N entry nodes of the M nodes that each match one of the plurality of fields are identified, wherein N is less than M. For each of the N entry nodes, a propagated entry vector having M entry values is generated, wherein each of the M entry values represents an importance of a corresponding node to the entry node. For each of the plurality of entities: each of the fields of the corresponding entity record that matches one of the N entry nodes is identified, thereby identifying K entity-specific entry nodes, wherein K is less than or equal to N. A set of K entity-specific entry vectors corresponding to the K entity-specific entry nodes are identified. A training vectors is generated by aggregating each of the K entity-specific entry vectors. Thereafter, a machine learning model can be trained using the training vectors and the known classifications.” [0025], “Graph elements (nodes) in SPOKE are linked to one another through edges that represent the node's relationships in the corresponding database”);
generating a plurality of node embeddings by embedding the subsets of the set of data associated with nodes in the graph, wherein the particular training dataset comprises the plurality of node embeddings, a representation of the graph, and a label indicative of whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period ([0046], “In generating a training vector, embodiments can identify overlapping concepts from the entity's records (EHRs) and the SPOKE graph database. Entry nodes associated with overlapping concepts are known as SPOKE entry points (SEPs) according to an embodiment of the disclosure. A propagated entry vector (PSEV), encoding the importance for each graph database node for a particular SEP (entry node), is created for each SEP.” [0051], “In order to convert structured EHR data within these individual snapshots into SPOKE embeddings, the population level interactions between different EHR concepts (e.g., fields) and SPOKE may be established. This is achieved through PSEVs which are machine-readable embeddings that quantify the significance of each node in SPOKE for a given cohort of patients.” [0055], “Patient Electronic Health Record (“EHR”) data is received. The EHR data includes standardized labels corresponding to the classifications of the patients (e.g., a disease diagnosis, a lab result, a therapeutic drug, etc.).” [0137], “A prediction stage 1030 can provide a predicted entity classification 1055 for a new entity's entity signature vector 1040 based on new entity records 1045.” [0082], “Example entities include patients, drugs, and animal subjects.”). Examiner interprets prediction of a disease in an animal to be a prediction wherein veterinary attention for the animal subject is likely required in an upcoming time period, as the average pet owner would be unable to treat or manage diseases such as cancer or chronic bowel inflammatory disease without veterinary oversight. Additionally, a prediction that an animal does not have a disease encompasses a prediction that the health of the animal subject is normal.
and training the multi-layer GNN model using the training datasets to predict whether health of an animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period ([0005], “For each of the plurality of entities: each of the fields of the corresponding entity record that matches one of the N entry nodes is identified, thereby identifying K entity-specific entry nodes, wherein K is less than or equal to N. A set of K entity-specific entry vectors corresponding to the K entity-specific entry nodes are identified. A training vectors is generated by aggregating each of the K entity-specific entry vectors. Thereafter, a machine learning model can be trained using the training vectors and the known classifications.” [0082], “Example entities include patients, drugs, and animal subjects.” [0126], “a machine learning model can be trained to diagnose a disease earlier than with conventional diagnostic methods (classify a biological entity).”).
Baranzini does not teach wherein each of the nodes is associated with a respective timepoint in the time period and represents a subset of the set of data collected at the respective timepoint and wherein a directed edge between two nodes represents a forward-in-time temporal relationship between the subsets of the set of data represented by the two nodes.
However, Liu does teach wherein each of the nodes is associated with a respective timepoint in the time period and represents a subset of the set of data collected at the respective timepoint and wherein a directed edge between two nodes represents a forward-in-time temporal relationship between the subsets of the set of data represented by the two nodes (pg. 707, “we can observe event xnl at time tnl in the sequence sn. We let the events xnl ∈ {1, …, M} and tnp ≤ tnq, for all p < q. We have one example of the medical event sequences of potential patients in Figure 1.With the observed event sequences, inspired by Liu et al. [17], we construct the following temporal graph for each sequence sn: Definition 1 (Temporal graph). The temporal graph Gn of sequence sn is a directed and weighted graph with our event set as its node set {1, …, M}… In Figure 2, we present the temporal graph of the event sequence in Figure 1. In the sequence, we have 6 observations of 4 unique events. We show the interval between event happening timestamps along the ordered edge between the observations.”). See Fig. 2, below, which provides a node for each data subset at a respective timepoint with the edges representing a forward-in-time temporal relationship via arrows between the subsets of the set of data between each node.
PNG
media_image1.png
600
593
media_image1.png
Greyscale
Baranzini in view of Liu are considered analogous to the claimed invention because they are in the field of machine learning for patient diagnosis. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Baranzini with Liu for the advantage of “captur[ing] temporal relationships of the medical events in each event sequence” (Liu; pg. 705).
Regarding claim 7, Baranzini teaches a computer implemented method of making an animal health prediction using a trained multi-layer graph neural network (GNN) model ([0050], “After the machine learning model is trained, the model can be used to classify new biological entities.” [0139], “model 1135 can be a neural network that comprises a number of neurons (e.g., Adaptive basis functions) organized in layers. The training of the neural network can iteratively search for the best configuration of the parameter of the neural network for feature recognition and classification performance. Various numbers of layers and nodes may be used.” [0078], “Process 700 can use a graph database (e.g., SPOKE) to obtain training vectors for training a machine learning model.” [0095], “The graph database may include M nodes, where M is greater than N. Entry nodes can correspond to any medically relevant concept, including diseases, genes, symptoms, or drugs that appear in fields of entity records that are to be classified.”), the method comprising:
obtaining, from an electronic database, a set of data for an animal subject over a time period, the set of data including: (i) clinical observation data, (ii) body weight measurement data, (iii) outcome status data, (iv) veterinary treatment record data, or (v) any combination thereof ([0131], “Once the model is trained, a new entity record (e.g., an EHR) can be classified, (e.g., whether a patient has a particular disease or whether a drug is suitable for a target).” [0064], “the computer system may receive, for a plurality of entities, a plurality of entity records (e.g., EHRs), each with a plurality of fields and a known classification, as described herein. Example entities include patients, drugs, and animal subjects.” [0055], “an entity record (e.g., EHR) including a plurality of fields is received. Entity records may be records related to medical treatment and entity record fields can include information related to diagnosis and treatment, such as test results, drug prescriptions, symptoms, or diagnoses.”);
encoding the set of data for the animal subject using a graph having nodes and directed edges between pairs of the nodes ([0132], “Using the new entity record, each SEP corresponding to the entity record is identified. Then, a PSEV is determined for the SEPs, and the entity signature is obtained by aggregating the corresponding PSEVs.” [0046], “embodiments can identify overlapping concepts from the entity's records (EHRs) and the SPOKE graph database. Entry nodes associated with overlapping concepts are known as SPOKE entry points (SEPs) according to an embodiment of the disclosure. A propagated entry vector (PSEV), encoding the importance for each graph database node for a particular SEP (entry node), is created for each SEP.”);
generating a plurality of node embeddings by embedding the subsets of the set of data associated with nodes in the graph ([0051], “In order to convert structured EHR data within these individual snapshots into SPOKE embeddings, the population level interactions between different EHR concepts (e.g., fields) and SPOKE may be established. This is achieved through PSEVs which are machine-readable embeddings that quantify the significance of each node in SPOKE for a given cohort of patients.”),
processing the plurality of node embeddings and a representation of the graph using a trained multi-layer graph neural network (GNN) model to predict whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period, wherein the trained multi-layer GNN model was previously trained to predict, from the plurality of node embeddings and a representation of the graph, whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period ([0050], “An entity signature vector can be generated for a new entity record and entered into the model to generate an entity classification. In this way an unknown biological entity can be classified using a machine learning model.” [0137], “A prediction stage 1030 can provide a predicted entity classification 1055 for a new entity's entity signature vector 1040 based on new entity records 1045.” [0055], “Patient Electronic Health Record (“EHR”) data is received. The EHR data includes standardized labels corresponding to the classifications of the patients (e.g., a disease diagnosis, a lab result, a therapeutic drug, etc.).” [0137], “A prediction stage 1030 can provide a predicted entity classification 1055 for a new entity's entity signature vector 1040 based on new entity records 1045.” [0082], “Example entities include patients, drugs, and animal subjects.”). Examiner interprets prediction of a disease in an animal to be a prediction wherein veterinary attention for the animal subject is likely required in an upcoming time period, as the average pet owner would be unable to treat or manage diseases such as cancer or chronic bowel inflammatory disease without veterinary oversight. Additionally, a prediction that an animal does not have a disease encompasses a prediction that the health of the animal subject is normal.
outputting a classification based on the result for the animal subject ([0101], “At block 870, an entity classification for the entity record is received as an output from the machine learning model.” [0082], “Example entities include patients, drugs, and animal subjects.”).
Baranzini does not teach wherein each of the nodes is associated with a respective timepoint in the time period and represents a subset of the set of data collected at the respective timepoint and wherein a directed edge between two nodes represents a forward-in-time temporal relationship between the subsets of the set of data represented by the two nodes
However, Liu does teach wherein each of the nodes is associated with a respective timepoint in the time period and represents a subset of the set of data collected at the respective timepoint and wherein a directed edge between two nodes represents a forward-in-time temporal relationship between the subsets of the set of data represented by the two nodes (pg. 707, “we can observe event xnl at time tnl in the sequence sn. We let the events xnl ∈ {1, …, M} and tnp ≤ tnq, for all p < q. We have one example of the medical event sequences of potential patients in Figure 1.With the observed event sequences, inspired by Liu et al. [17], we construct the following temporal graph for each sequence sn: Definition 1 (Temporal graph). The temporal graph Gn of sequence sn is a directed and weighted graph with our event set as its node set {1, …, M}… In Figure 2, we present the temporal graph of the event sequence in Figure 1. In the sequence, we have 6 observations of 4 unique events. We show the interval between event happening timestamps along the ordered edge between the observations.”). See Fig. 2, below, which provides a node for each data subset at a respective timepoint with the edges representing a forward-in-time temporal relationship via arrows between the subsets of the set of data between each node.
PNG
media_image1.png
600
593
media_image1.png
Greyscale
Baranzini in view of Liu are considered analogous to the claimed invention because they are in the field of machine learning for patient diagnosis. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Baranzini with Liu for the advantage of “captur[ing] temporal relationships of the medical events in each event sequence” (Liu; pg. 705).
Regarding claim 14, this claim is rejected for the same reasons as claim 1. Vogler further teaches one or more data processors; and a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform the method ([0145], “The software code may be stored as a series of instructions or commands on a computer readable medium for storage and/or transmission. A suitable non-transitory computer readable medium can include random access memory (RAM), a read only memory (ROM), a magnetic medium such as a hard-drive or a floppy disk, or an optical medium such as a compact disk (CD) or DVD (digital versatile disk) or Blu-ray disk, flash memory, and the like.”).
Regarding claims 18, this claim is rejected for the same reasons as claim 7. Vogler further teaches one or more data processors; and a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform the method ([0067], “the present disclosure provides a non-transitory computer readable medium, storing instructions that, when executed by a processor, cause a computer system to execute the steps of any one of the methods disclosed herein.”).
Regarding claim 20, this claim is rejected for the same reasons as claim 1, as described above.
Claims 2 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Baranzini (US 20240303544) in view of Liu (Liu; Chuanren, Temporal Phenotyping from Longitudinal Electronic Health Records: A Graph Based Framework, 10 Aug 2015, KDD '15: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Pages 705 – 714) further in view of Hirsch (US 20190096526).
Regarding claim 2, Baranzini in view of Liu teaches the method of claim 1, as described above. Baranzini in view of Liu does not teach wherein generating the plurality of node embeddings by embedding the subsets of the set of data associated with nodes in the graph comprises, generating a particular node embedding for a particular subset by: (i) (a) determining a free text entry in the sets of data; (b)applying an embedding model to the free text entry to generate a vector of the free text entry; (c) reducing a size of the vector using principal component analysis reduction method; and (d) including the vector in the particular node embedding; or (ii) (a) determining a categorical variable entry in the subset; (b) converting the categorical variable entry into a numerical value using a mapping between numerical values and categorical variable entries; and (c) including the numerical value in the particular node embedding.
However, Hirsch in view of Baranzini does teach wherein generating the plurality of node embeddings by embedding the subsets of the set of data associated with nodes in the graph comprises, generating a particular node embedding for a particular subset (Baranzini, [0046], “In generating a training vector, embodiments can identify overlapping concepts from the entity's records (EHRs) and the SPOKE graph database. Entry nodes associated with overlapping concepts are known as SPOKE entry points (SEPs) according to an embodiment of the disclosure. A propagated entry vector (PSEV), encoding the importance for each graph database node for a particular SEP (entry node), is created for each SEP.” [0051], “In order to convert structured EHR data within these individual snapshots into SPOKE embeddings, the population level interactions between different EHR concepts (e.g., fields) and SPOKE may be established. This is achieved through PSEVs which are machine-readable embeddings that quantify the significance of each node in SPOKE for a given cohort of patients.”) by:
(i) (a) determining a free text entry in the sets of data (Hirsch, [0052], “formulate the entity by performing preprocessing, text processing, data compression, data reduction, dimensional reduction, etc. from the retrieved dataset… Examples of data types can include, but are not limited to, text, alphanumeric, biological sequences,…”);
(b) applying an embedding model to the free text entry to generate a vector of the free text entry (Hirsch, [0086], “The system processor 128 can derive a set of conformed, that is mathematically well behaved, feature vectors in an N-dimensional Euclidean space, which can be viewed as a Hilbert Space, called the embedding space, from the input data in Tables 1 through 9, and any subset and/or combination thereof.” [0098], “the linear and/or nonlinear clustering can be mapped (that is, embedded) onto at least one of linear manifold and nonlinear manifold with the clustering modules 212 and 222.” [0084], “The linear manifold (LM) clustering can use locally linear and/or locally nonlinear high-dimensional spaces that are embedded on a linear manifold. The nonlinear manifold (NLM) clustering can use locally linear and/or locally nonlinear high-dimensional spaces that are embedded on a nonlinear manifold.”);
(c) reducing a size of the vector using principal component analysis reduction method (Hirsch, [0063], “Dimensionally of the vector space can be reduced/increased using information loss/gain as a controlling factor. The information loss using Principle Component Analysis (PCA), Singular Value Decomposition (SVD) and/or State Vector Machine (SVM).”); and
(d) including the vector in the particular node embedding (Hirsch, [Table 11], “5 Based on the expected loss and associated utility function and entity/attribute vector freeze the methods in the module under test and vary other modules at least one other module and observe the change in the expected loss. 6 Use global adjudication to select best processing branch. Repeat processes 1 to 5 as required continuing to observe the expect loss processing performed in the math model module 7 Rate of convergence/divergence, determination of cluster inclusion, bias, etc. as a tool to signal off ramp 8 Domain Expert can control training process, initial data set and dimensionality, maximum data set and dimensionality, method of expansion of data set and dimensionality, class of cost functions (e.g. cubic, quadratic, etc.), update coefficients for cost functions, thresholds, bias of data, weights and masking, collaborative filtering, etc.” Baranzini, [0046], “In generating a training vector, embodiments can identify overlapping concepts from the entity's records (EHRs) and the SPOKE graph database. Entry nodes associated with overlapping concepts are known as SPOKE entry points (SEPs) according to an embodiment of the disclosure. A propagated entry vector (PSEV), encoding the importance for each graph database node for a particular SEP (entry node), is created for each SEP.” [0051], “In order to convert structured EHR data within these individual snapshots into SPOKE embeddings, the population level interactions between different EHR concepts (e.g., fields) and SPOKE may be established. This is achieved through PSEVs which are machine-readable embeddings that quantify the significance of each node in SPOKE for a given cohort of patients.”);
or (ii) (a) determining a categorical variable entry in the subset; (b) converting the categorical variable entry into a numerical value using a mapping between numerical values and categorical variable entries; and (c) including the numerical value in the particular node embedding.
Baranzini in view of Liu further in view of Hirsch are considered analogous to the claimed invention because they are in the field of machine learning for patient diagnosis. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Baranzini in view of Liu with Hirsch for the advantage of “generat[ing] or modify[ing] term vectors formed in the data conditioning modules 210 and 220, with the TF.IDF being a metric that assigns numerical values to unstructured text” (Hirsch; [0065]).
Regarding claim 15, this claim is rejected for the same reasons as claim 2, as described above.
Claims 9-11, 19, and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Baranzini (US 20240303544) in view of Liu (Liu; Chuanren, Temporal Phenotyping from Longitudinal Electronic Health Records: A Graph Based Framework, 10 Aug 2015, KDD '15: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Pages 705 – 714) further in view of Bradley (US 20190096526).
Regarding claim 9, Baranzini in view of Liu teaches the method of claim 7, as described above. Baranzini further teaches wherein processing the plurality of node embeddings and a representation of the graph adjacency graph into using the trained multi- layer comprises determining a likelihood indicating whether: health of the animal subject is normal, veterinary attention for the animal subject is likely required in an upcoming time period, an unplanned death outcome for the animal subject is likely in the upcoming time period, or a treatment is likely to be administered to the animal subject in the upcoming time period ([0101], “At block 870, an entity classification for the entity record is received as an output from the machine learning model... In some implementations, the entity classification can be provided as a probability for a given classification.” [0055], “Patient Electronic Health Record (“EHR”) data is received. The EHR data includes standardized labels corresponding to the classifications of the patients (e.g., a disease diagnosis, a lab result, a therapeutic drug, etc.).” [0137], “A prediction stage 1030 can provide a predicted entity classification 1055 for a new entity's entity signature vector 1040 based on new entity records 1045.” [0082], “Example entities include patients, drugs, and animal subjects.”). Examiner interprets prediction of a disease in an animal to be a prediction wherein veterinary attention for the animal subject is likely required in an upcoming time period, as the average pet owner would be unable to treat or manage diseases such as cancer or chronic bowel inflammatory disease without veterinary oversight. Additionally, a prediction that an animal does not have a disease encompasses a prediction that the health of the animal subject is normal.
However, Baranzini in view of Liu does not teach wherein outputting the classification comprises comparing the likelihood to a determined threshold and classifying, based on the comparison, the animal subject as having a normal health status, requiring veterinary attention in the upcoming time period, being likely to experience the unplanned death outcome the upcoming time period, or likely to receive the treatment in the upcoming time period comparison.
However, Bradley does teach wherein the classification comprises comparing the result for the animal subject to a determined threshold and classifying, based on the comparison, the animal subject as having a normal health status, requiring veterinary attention in the upcoming time period, being likely to experience the unplanned death outcome the upcoming time period, or likely to receive the treatment in the upcoming time period comparison ([0024], “determining the risk of developing CKD by comparing the score with a threshold value; ” [0104], “A score of between 0 and 5 suggests that the cat will not likely develop CKD within the next 2 years. A score of between 6 and 25 suggests insufficient certainty to predict CKD in the cat, and a veterinary visit within 6 months is recommended. A score of between 26 and 49 suggests insufficient certainty to predict CKD in the cat, and a veterinary visit within 3 months is recommended.”).
Baranzini in view of Liu further in view of Bradley are considered analogous to the claimed invention because they are in the field of machine learning for patient diagnosis. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Baranzini in view of Liu with Bradley for the advantage of using a “decision threshold to classify an individual as predicted or non-ill patient” (Bradley; [0346]).
Regarding claim 10, Baranzini in view of Liu further in view of Bradley teaches the method of claims 7 and 9, as described above. Vogler in view of Baranzini does not teach the method further comprising providing a recommendation based on the classification of the animal subject.
However, Bradley does teach the method further comprising providing a recommendation based on the classification of the animal subject ([0009], “determining or categorizing, based on the output, whether the feline is at risk of developing CKD; and determining a customized recommendation based on the determining or categorizing.”).
Vogler in view of Baranzini further in view of Bradley are considered analogous to the claimed invention because they are in the field of machine learning for patient diagnosis. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vogler in view of Baranzini with Bradley for the advantage of “determining a customized recommendation based on the categorizing” (Bradley; [0049]).
Regarding claim 11, Baranzini in view of Liu further in view of Bradley teaches the method of claims 7 and 9-10, as described above. Baranzini further teaches the method further comprising providing the classification and/or the recommendation to a user through a graphical user interface (GUI) ([0146], “A computer system may include a monitor, printer, or other suitable display for providing any of the results mentioned herein to a user.” [0142], “monitor 1276 (e.g., a display screen, such as a light emitting diode (LED) display screen), which is coupled to display adapter 1282”).
Regarding claims 19 and 21, these claims are rejected for the same reasons as claims 9 and 11, respectively.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID CHOI whose telephone number is (571)272-3931. The examiner can normally be reached M-Th: 8:30-5:30 ET.
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, Shahid Merchant can be reached on (571)270-1360. 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.
/D.C./Examiner, Art Unit 3684
/Shahid Merchant/Supervisory Patent Examiner, Art Unit 3684
Prosecution Insights