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 .
Claim Status
Claims 1-20 are currently pending and under examination herein.
Priority
The present application, filed on May 19, 2022 claims benefit to provisional application 63/223,029, filed on July 18, 2021. The instant application under current examination has filing date of July 18, 2021.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 19 May 2022 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Drawings
The drawings, submitted on 19 May 2022, have been accepted by the examiner.
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.
Claims 6, 7, 14, 15, and 20 are rejected under 35 U.S.C. 112(b) 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 6, 14, and 20 recite the equation ”
y
=
f
θ
(
M
,
D
,
C
)
” wherein variables M, D, and C are not defined. To overcome this rejection, define variables within the claims, consistent with the current disclosure.
Claim 7 and 15 recite …“to encode molecular graphs representing the drug molecules and to average over embeddings of multiple drugs molecules”. The claims are rejected for the following reasons:
The limitation “the drug molecules” lacks antecedent basis. If molecular graphs representing drug molecules was obtained from drug molecule data, please clarify or provide proper antecedent basis for the limitation.
“embeddings of multiple drugs molecules” is unclear and/or lacks antecedent basis. Clarification is required.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101.
Eligibility Step 1: Subject matter eligibility evaluation in accordance with MPEP § 2106:
Claims 1-8 are directed to a statutory category (machine).
Claims 9-16 are directed to a statutory category (method).
Claims 17-20 are directed to a statutory category (machine).
Therefore, in accordance with MPEP § 2106.03 claims 1-20 have patent eligible subject matter.
[Eligibility Step 1: YES]
Eligibility Step 2A: This step determines whether a claim is directed to a judicial exception in accordance with MPEP § 2106.
Eligibility Step 2A -- Prong One: Limitations are analyzed to determine if the claims recite any concepts that could equate to a judicial exception (i.e. abstract idea, law of nature, or natural phenomenon). Possible judicial exceptions are explored below.
Independent Claim 1:
parse the CT data to derive drug molecules data, disease information data, and trial protocols data,
Independent Claim 9:
parsing, by the trial prediction (TP) node, the CT data to derive drug molecules data, disease information data, and trial protocols data;
Parsing relevant data from a dataset exemplifies a mental observation and analysis that can be performed within the human mind or while utilizing pen and paper. As such claims 1 and 9 recite limitations that may draw to an abstract idea that fall within the grouping of “mental processes”.
The following limitations contain judicial exceptions in the form of abstract ideas that can be categorized as mathematical concepts:
Independent Claim 1:
encode the drug molecules data, the disease information data, and the trial protocols data into corresponding embeddings,
generate knowledge pre-trained embeddings using external knowledge data
Claim 3:
The system of claim 2, wherein the instructions further cause the processor to generate the knowledge pre-trained embeddings based on the drug pharmaco-kinetics data and the disease risk data.
Claim 4:
The system of claim 3, wherein the instructions further cause the processor to generate a disease risk embedding to pre-train prediction models for the disease risk.
Claim 6:
The system of claim 1, wherein the instructions further cause the processor to train a deep neural network
y
=
f
θ
M
,
D
,
C
to predict the CT outcome based on model parameters θ.
Claim 7:
The system of claim 1, wherein the instructions further cause the processor to use a message passing network to encode molecular graphs representing the drug molecules and to average over embeddings of multiple drugs molecules.
Claim 9:
encoding, by the trial prediction (TP) node, the drug molecules data, the disease information data, and the trial protocols data into corresponding embeddings;
generating, by the trial prediction (TP) node, knowledge pre-trained embeddings using external knowledge data;
Claim 11:
The method of claim 10, further comprising generating the knowledge pre-trained embeddings based on the drug pharmaco-kinetics data and the disease risk data.
Claim 12:
The method of claim 11, further comprising generating a disease risk embedding to pre-train prediction models for the disease risk.
Claim 14:
The method of claim 9, further comprising training a deep neural network
y
=
f
θ
M
,
D
,
C
to predict the CT outcome based on model parameters
θ
.
Claim 15:
The method of claim 9, further comprising using a message passing network to encode molecular graphs representing the drug molecules and to average over embeddings of multiple drugs molecules.
Claim 20:
The non-transitory computer readable medium of claim 17, further comprising instructions, that when read by the processor, cause the processor to train a deep neural network
y
=
f
θ
M
,
D
,
C
to predict the CT outcome based on model parameters θ.
Encoding data into embeddings, when viewed in the broadest reasonable interpretation performs mathematical calculations (ie. Euclidean difference, dot product, cosine similarity) on input data to transform it into secondary data. The limitation of “averaging over embeddings” recites performing a mathematical calculation (mean) on the secondary data.
Additionally, though training a neural network itself is not an abstract idea, claims that recite a mathematical formula, including
y
=
f
θ
M
,
D
,
C
, demonstrate mathematical concepts. As such claims 1, 3, 4, 6, 7, 9, 11, 12, 14, 15, and 20 contain limitations that fall within the mathematical concepts grouping.
Therefore, all of the independent claims and some of the dependent claims appear to read on judicial exceptions (abstract ideas).
[Eligibility Step 2A – Prong One: YES]
Eligibility Step 2A – Prong Two: A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception. If the claim contains no additional claim elements beyond the abstract idea, the claim fails to integrate the abstract idea into a practical application (MPEP 2106.04(d)).
The following limitations are additional elements that are analyzed to determine if they integrate the judicial exceptions into practical applications.
Independent Claim 1:
A system, comprising: a processor of a trial prediction (TP) node connected to at least one cloud server node over a network configured to host a machine learning (ML) module;
a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to:
Independent Claim 17:
A non-transitory computer readable medium comprising instructions, that when read by a processor, cause the processor to perform:
The additional elements above represent computer components that specify the technological environment. When viewed in the context of the claimed invention as a whole, they are necessary parts for a generic computing system to receive and transmit data. This category of components does not integrate a judicial exception into practical application, as established in Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016).
Independent Claim 1:
receive a clinical trial (CT) data,
provide the knowledge pre-trained embeddings to the ML module for prediction of the CT outcome.
Independent Claim 9:
A method, comprising: receiving, by a trial prediction (TP) node, a clinical trial (CT) data
providing, by the trial prediction (TP) node, the knowledge pre-trained embeddings to an ML module for prediction of the CT outcome.
Independent Claim 17:
receiving a clinical trial (CT) data;
providing the knowledge pre-trained embeddings to an ML module for prediction of the CT outcome.
Dependent Claim 2:
The system of claim 1, wherein the instructions further cause the processor to query a database for drug pharmaco-kinetics data and disease risk data.
Dependent claim 10:
The method of claim 9, further comprising querying a database for drug pharmaco- kinetics data and disease risk data
Dependent claim 18:
The non transitory computer readable medium of claim 17 further comprising instructions that when read by a processor, cause the processor to query a database for drug pharmaco-kinetics data and disease risk data
Receiving data, querying a database, and providing information to a machine learning module represent mere data gathering activities that the courts established as insignificant, extra solution activity via Mayo, 566 U.S. at 79, 101 USPQ2d at 1968; OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1092-93 (Fed. Cir. 2015), which also fail to integrate the recited judicial exceptions into practical application.
Dependent Claim 5:
The system of claim 1, wherein the instructions further cause the processor to train a dynamic attentive graph neural network to predict the CT outcome.
Dependent Claim 8:
The system of claim 1, wherein the instructions further cause the processor to pre-train prediction models for absorption, distribution, metabolism, excretion, and toxicity based on the drug molecules data.
Dependent Claim 13:
The method of claim 9, further comprising training a dynamic attentive graph neural network to predict the CT outcome.
Dependent Claim 16:
The method of claim 9, further comprising pre-training prediction models for absorption, distribution, metabolism, excretion, and toxicity based on the drug molecules data.
The limitations of claims 5 and 13 only recite the active step of training a dynamic attentive neural network (DAGNN). The limitation would cover every mode of implementing a DAGNN using the general data gathering steps and abstract ideas (mathematical encoding process) present in the independent claim, while the computer merely acts as a tool to apply them. Claims 8 and 16 describe pre-training prediction models based on a particular dataset, with no meaningful limitations on how the models operate. None of the claims actively describe how the model solves a technical problem and instead only recite the idea of a solution or outcome. These processes also do not improve computer functionality individually or when examined within the whole claimed invention and can therefore can be viewed as nothing more than an attempt to generally link the judicial exception to a particular field of use or technological environment.
After evaluation of the additional elements individually and in context of their claimed invention as a whole, they fail to integrate the judicial exceptions (identified in Step 2A) into practical application. As such claims 1-20 are directed to judicial exceptions, in accordance with MPEP § 2106.04.
[Eligibility Step 2A – Prong Two: YES]
Eligibility Step 2B: Claim elements are probed for inventive concept equating to significantly more than the judicial exception (MPEP 2106.04(II)).
Elements that described computer functions such as storing and retrieving information in memory and receiving or transmitting data over a network are considered to be well-understood, routine, and conventional. These findings were established by the courts in Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015) and buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014), respectively. Furthermore, the elements, in the context of their invention as a whole merely specify the computing environment as a tool to perform an existing process more quickly.
When examined in the context of their claimed inventions, the elements do not show a clear improvement to technology, thus maintaining no more than their conventional function.
Elements in the claims that described pre-training/training a neural network were also found to lack inventive concept. Sun et al (Briefings in Bioinformatics; Vol. 21(3); 2019) reviews graph convolutional networks that use different methods of encoding, evaluating, and training prediction models that aid the drug development process. Skarding et al (IEEE Access; Vol. 9; 2021) further shows conventionality via a survey of dynamic graph neural networks. Both establish that the additional elements, as described within the claims and when considered as a whole invention, are routine and conventional within the art.
As such, claims 1-20 do not show inventive concept in the form of significantly more.
[Eligibility Step 2B: NO]
Evaluation in accordance with Alice/Mayo, MPEP 2143
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.
Claims 1-4, 6, 8, 9-12, 14, 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Association for Computing Machinery; 2020) in view of Gal et al (Briefings in Bioinformatics; Vol. 19(6); 2018).
Claims 1, 9, and 17 are directed to a system, method, and computer readable medium that receives, parses, and encodes clinical trial data pertaining the drug molecules, disease, and protocol. It also generates knowledge pre-trained embeddings using external data and provides these embeddings to a ML (machine learning) module.
Zhang et al. describes Deep Enroll, a system of predicting patient-trial matching and entailment using knowledge pre-trained embeddings based on eligibility criteria(EC) and patient data derived from electronic health records (EHR) (abstract).
Zhang et al. shows a system that includes a trial EC embedding module, a hierarchical patient representation module, and the alignment and entailment prediction module (page 1030, column 2).
Zhang et al. shows the method of receiving data from ClinicalTrials.gov and a large-scale patient-trial matching dataset (page 1033, column 1). Zhang et al. shows the clinical trial data being parsed by selecting the inclusive and exclusive criteria, then extracting EC statements, rare disease data, and patient health information (page 1033, column 1). The parsed information includes patient characteristics such as: age, gender (1031, column 2), medical history (page 1036, column 1), target disease conditions (page 1035, column 2), and current health status (page 1031, column 1), in line with trial protocol data according to the specification.
Zhang et al. shows encoding enrollment criteria and patient records into a shared latent space for matching inference (page 1029, column 1), by encoding clinical trial information into sentence embedding U and patient health data into embedding V (page 1030, figure 1).
Zhang et al. shows generating knowledge pre-trained embeddings by applying a Bidirectional Encoder Representations from Transformers (BERT) and hierarchical embedding module (page 1029, column 1). Zhang et al. shows the Clinical BERT, was pretrained using external medical corpora data, (page 1031, column 1) and the hierarchical embedding module used external patient, visit, and demographic information (page 1031, column 1). Zhang et al. shows jointly considering interactions between single EHR and EC as well as for the original EC embeddings and EHR embeddings by providing their concatenation to a machine learning module to compute the entailment relationship between given patient and trial (page 1032, column 1).
With respect to claims 2, 18, and 10, Zhang et al. shows querying databases for EHR (page 1037, column 1) and EC data (page 1031, column 1). With respect to claims 3, 11, and 19, Zhang et al. shows generating knowledge-pre trained embeddings based on EC data (page 1031, column 1). With respect to Claims 4 and 12, Zhang et al. shows further generating a dense representation embedding of patient health data (page 1031, column 2).
Claims 6, 14, and 20 draw to training a deep neural network prediction model based on the relationship between drug molecules, diseases, and patient characteristics.
Zhang et al. shows training a deep neural network by calculating the relationship between variables within the triplet {E, H, y}, where E represents the Eligibility Criteria sentence embedding, H represents the hierarchal patient embedding, and y is the predicted entailment outcome (page 1033, algorithm 1). Figure 1 further illustrates the processes drawn to the claims above (page 1030, figure 1).
Zhang et al. does not explicitly show:
a processor of a trial prediction (TP) node connected to at least one cloud server node over a network configured to host a machine learning (ML) module; (Claim 1)
a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to… (Claim 1)
A non-transitory computer readable medium comprising instructions, that when read by a processor, cause the processor to perform… (Claim 17)
However, Zhang et al. teaches an analogous system that embodies the utilization of inherent computer components such as machine-readable instructions that cause a processor to perform the aforementioned functions. It would be obvious to one having ordinary skill in the art to automate this system further by connecting its prediction module to at least one cloud server node.
Zhang et al. also does not show any use of drug molecule data to predict clinical trial outcome. Instead, Zhang et al. uses EC and EHR data to predict clinical trial entailment and patient matching (abstract).
Gal et al. presents a review of data and developmental models required to accurately simulate clinical trials (page 1, abstract). Figure 3 summarizes that mathematical models representing information about the drug, disease, and patient derived from clinical trial data are required for accurate simulation/predictive models (page 1212, figure 3). Gal et al. further shows the necessity of trial design strategies (page 1210, column 1) and protocol incorporated within the patient data (page 1208, column 1).
Therefore, it would be obvious of one of ordinary skill in the art to adapt the method of Zhang et al. to predict clinical trial outcome by supplementing EC and EHR data with drug molecule information in view of Gal et al’s teaching that the combination of all three are the most suitable components for modelling clinical trials.
With respect to claims 8, 16, Zhang et al. also does not show training prediction models for absorption, distribution, metabolism, excretion, and toxicity based on the drug molecules data.
Gal et al. however teaches that it is crucial to take molecular classifications into account when generating clinical trial simulations (page 1209, column 1).
Gal et al. further teaches that use of computational tool admetSAR (absorption, distribution, metabolism, excretion, toxicity structure-activity relationship) which includes 22 qualitative classification models, and five quantity regression models developed using support vector machine algorithm that provide a more accurate prediction of the biodegradability of 27 novel compounds compared with models that evaluated different molecular qualities (page 1209, column 1).
Therefore, it is obvious for one of ordinary skill in to further adapt the system with ADMET prediction models to best account for molecular classifications within the clinical trial drug, as suggested by Gal et al.
Claims 5, 7, 13, and 15 are rejected under 35 USC 103 as being unpatentable over Zhang et al. in view of Gal et al. as applied to claims 1-4, 6, 8, 9-12, 14, 16-20 above, and in further view of Hamilton (Graph Representation Learning, 2020), Ji et al (Association for Computational Linguistics; Vol. 1, 2015), and Skarding (IEEE Access; Vol. 9, 2021).
Claims 5 and 13 are directed to training a dynamic attentive graph neural network. Claims 7 and 15 are directed to using a message passing network to encode molecular graphs representing drug molecules and obtaining a singular representation of the molecules by computing the mean of their embeddings.
Zhang et al. shows training an attentive neural network to predict whether the hypothesis is entailed in the premise i.e. given a trial EC, where a particular patient is a match for the trial (page 1032, column 1). Zhang et al. shows the match prediction module is updated by the result of the numerical information entailment (page 1031, Figure 1), which changes the result over time.
Zhang et al. does not show that the dynamic attentive neural network is a graph-based model, nor the encoding of molecular graphs of drug molecules.
Hamilton (Graph Representation Learning, 2020) is a released pre-publication of graph representation learning textbook that explains the basics of graph neural network models.
Hamilton teaches that graph neural networks generate representations that depend on the structure of the graph, without topological restrictions (page 47, column 1) and are defined by their message passing network feature (page 48, column 1). Hamilton further teaches adding attention increases the representational capacity of a GNN model (page 57, column 1) and updating the results of the prediction model is useful when the prediction task requires complex reasoning over the global structure of the graph (page 61, column 1).
Hamilton further teaches graph-based neural networks are particularly useful when analyzing and predicting the structural motifs, features (page 50, column 1), and properties of graph-based representations of molecules (page 68, column 1). Hamilton further teaches averaging node embeddings is sufficient to obtain singular representations of these small graphs (page 64, column 1) and attention based GNNs are popular architecture when incorporating natural language processing transformers, such as BERT (page 57, column 1).
Ji et al (Association for Computational Linguistics; Vol. 1, 2015) teaches Knowledge Graph Embedding via Dynamic Mapping Matrix. Ji et al. teaches dynamically mapped matrices consider the diversity of entities and relations simultaneously and provide a flexible style to project entity representations to relation vector spaces (page 688, column 2). Ji et al. further teaches this improved method of mapping can be easily applied on large-scale knowledge graphs to enhance link prediction, triplet classification tasks (page 688, column 2), and serve as useful resource for practical applications where data is incomplete (page 695, column 1).
Skarding et al (IEEE Access; Vol. 9; 2021) teaches dynamic graph neural networks add a new dimension, time, to network modelling and prediction which radically influences network properties and enables a more powerful representation of network data, which in turn increases predictive capabilities (page 79143, column 1).
As stated previously Zhang et al. does not manipulate drug molecules data, so there is little motivation for a graph-based neural network. But if the method was adapted for a different function, in view of Gal et al as previously applied, one of ordinary skill in the art would be motivated to further apply the suggested graph-based techniques (i.e. message passing network, mean pooling), taught by Hamilton to gain an accurate representation of the structural data properties of the drug molecules during the encoding process.
Hamilton, Ji et al, and Skarding et al. provide motivation to further enhance predictions by training a dynamic attentive neural network that will fully account for the complex relationship between the crucial data entities, including BERT trained knowledge graphs, without reducing features of the drug molecules.
Conclusion
No claims are currently allowed.
Correspondence
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Milana Thompson whose telephone number is (571)272-8740. The examiner can normally be reached Monday - Friday, 9:00-6:00 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, Karlheinz Skowronek can be reached at (571) 272-1113. 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.
/M.K.T./Examiner, Art Unit 1687
/Karlheinz R. Skowronek/Supervisory Patent Examiner, Art Unit 1687