Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
2. A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office Action has been withdrawn pursuant to 37 CFR 1.114.
Detailed Action
3. This Non-Final Office Action is responsive to Applicants’ RCE submission dated 1/16/26, which asserts the amendments and arguments as received by way of Applicants’ After-Final Reply dated 12/15/25. Claims 1-12 remain pending, of which claims 1 and 7 are independent.
Claim Rejections - 35 USC § 103
4. 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.
5. 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.
6. 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.
7. Claims 1-3 and 7-9 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication No. 2017/0116390 (“Fokoue-Nkoutche”) in view of Non-Patent Literature “Developing a New Reference Standard... Is Validation Necessary?” (“Gold”) and further in view of U.S. Patent No. 10685293 (“Heimann”).
Regarding claim 1, FOKOUE-NKOUTCHE teaches A processor-implemented adverse drug reaction detection method ([0002]: “The present disclosure relates to prediction of adverse drug events and more specifically, to methods, systems and computer program products for analysis of data to provide personalized and detailed adverse drug events.”, and [0017]: “In accordance with exemplary embodiments of the disclosure, methods, systems and computer program products for predicting adverse drug reactions are provided”, and where [0067] and [0072] make clear the computer and processor implemented aspect of the taught invention), the method comprising:
receiving, by one or more processors, raw data including information on adverse events of a plurality of patients with regard to a target drug (FIG. 5 step 402 and FIG. 6 step 502, as preliminary steps to collect drug and adverse drug event data in advance of machine learning techniques that result in making an active adverse event prediction for a candidate drug or drug pair (FIG. 5 step 412, FIG. 6 step 516) for scenarios with or without a particular patient in mind (FIG. 5 verses FIG. 6, where FIG. 6 steps 508-512 consider a particular patient’s information/history to better assess/predict the candidate drug’s potential for adverse event));
classifying, by one or more processors, the received raw data into first data corresponding to adverse drug reactions of the target drug, second data corresponding to no adverse reactions of the target drug and drugs similar to the target drug ... ([0043] teaching the creation of known adverse event feature tables or a similar repository, to associate adverse event features with known drugs or drug pairs corresponding to those adverse event features, where per [0045] and [0064], tables / data structures such as those are a basis for a classifier to predict adverse drug events, such that it would be obvious for the classifier to classify particular data into predicting yes or no for an adverse drug event);
learning a machine learning model by implementing a gold standard dataset including data corresponding to the first data and the second data among the received raw data (FIG. 8 step 718 teaching the construction of an adverse drug reaction classifier, clarified per [0060] and [0064], where such a classifier using a training repository ([0060], [0063]-[0064]) would necessarily be trained using data that had instances of positive or negative indications of adverse drug reactions, so that it could successfully function and discern as described, and those training data examples that inform the classifier’s resulting model/implementation are what would be considered “ground truth” or as recited here “gold standard”);
wherein the received raw data is classified based on a database that is separate from and not generated by the machine learning model, prior to the learning of the machine learning model (as discussed above per the receiving limitation: FIG. 5 step 402 and FIG. 6 step 502, as preliminary steps to collect drug and adverse drug event data in advance of machine learning techniques that result in making an active adverse event prediction for a candidate drug or drug pair (FIG. 5 step 412, FIG. 6 step 516) for scenarios with or without a particular patient in mind (FIG. 5 verses FIG. 6, where FIG. 6 steps 508-512 consider a particular patient’s information/history to better assess/predict the candidate drug’s potential for adverse event), where the “known drug data” subject to the receiving per FIG. 5 step 402 is is further clarified in [0038]: “Known drug data can include structured data, unstructured data, or both structured and unstructured data. As used herein, structured data includes data that is categorized or grouped in accordance with a system of defined rules. As used herein, unstructured data includes data that is not categorized or grouped in accordance with a system of defined rules. For example, unstructured data includes, but is not limited to, data published in journal articles in a narrative format. In exemplary embodiments, known drug data includes data from databases generally known to persons of ordinary skill in the art. For example, known drug data can include data from the DrugBank database, UniProt, Unified Medical Language System TM, PubMed, and/or various scientific journals, including, but not limited to, the Journal of Clinical Oncology, JAMA, BJC, and Clinical Infectious Diseases.”).
Applicants’ claims further recite the additional limitation for classifying the received raw data into ... third data, that is separate and apart from first and second data instances as discussed above, such that determining, by one or more processors, a possibility of adverse reactions for a prediction dataset including the data corresponding to the third data among the received raw data by implementing the machine learning model which Fokoue-Nkoutche alone does not teach.
At best, the aforementioned reference teaches, with FIG. 5 steps 412-414 and FIG. 6 step 516, that inference modes can be applied to perform adverse drug reaction prediction framework, e.g., based on the framework’s model being trained using the gold set of known data points for adverse and non-adverse reactions. The Examiner does not believe this reference alone sufficiently encompasses the consideration of a “third data” as recited which appears to be neither data known to be adverse or non-adverse in terms of associated drug reaction to a target drug.
Rather, the Examiner relies upon GOLD and then HEIMANN to teach what Fokoue-Nkoutche alone otherwise lacks:
Gold’s Abstract and Introduction on page 1, discussing the development of a new standard that shifts from an existing gold standard, intuitively causing the shift by accounting for and validating data that does not fit into the gold standard, thereby creating a more shifting or more permissive new standard. Validation considerations and measures are discussed on pages 4-5 that seek to ensure that the new data that is operative in shifting the standard is worthy.
The Examiner understands this to suggest that not all data that underlies a model is of the same trust/fidelity measure, e.g. even gold standard data is subject to scrutiny and skepticism. However, for modeling of a problem/solution to persist, new data must continue to be evaluated and sometimes with the result that the standard shifts or changes to encompass new data that is suitably apt.
Hence, data in a gold standard, such as the one taught by Fokoue-Nkoutche, is eligible for scrutiny, and data in the sort of databases, repositories, and publications that Fokoue-Nkoutche obtains data from to use and/or rely upon in developing a standard then, by extension, is also eligible for scrutiny. If this is a reasonable understanding of Fokoue-Nkoutche as clarified by Gold, then such data that is less than certain in what it stands for, per Gold for example, would constitute and read on Applicants’ recitation of “third data” that is decidedly not either of the recited first and second data.
Both Fokoue-Nkoutche and Gold relate to the development and use of gold set standards to model solutions/predictions based on underlying data, e.g. medical/patient data. Hence, the aforementioned references are similarly directed and therefore analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to assess data points in existing gold standard sets or new data under consideration as Gold teaches in a framework such as Fokoue-Nkoutche, with a reasonable expectation of success, to make determinations as to whether the gold standard remains sufficient, is in need of adjustment to account for a model shift, and so forth, as Gold discusses.
Heimann’s classification of data as “unknown” or “decision boundary”, per column 26 lines 45-62 and further per column 27 lines 11-59, where such data as subjected to this particular classification feature an uncertainty that requires further processing and/or intervention, such that over time, as the training data and framework evolves to know or learn that particular data, the overall learning aspect has improved and become more competent.
The Examiner understands Heimann’s modeling aspect to account for uncertain data and selectively validate it such that an existing model can continue to develop and be useful. In the Examiner’s view, such measures per Heimann can be applicable to maintaining a model, such as Fokoue-Nkoutche’s, in view of data that is questionable in or outside of a gold set, as Fokoue-Nkoutche clarified in view of Gold suggests.
Both Fokoue-Nkoutche and Heimann relate to classifier frameworks and are therefore analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Heimann’s combined active-learning and self-learning aspects with Fokoue-Nkoutche’s framework, with a reasonable expectation of success, such that the classification is improved over time and repetitions to handle more data with more certainty.
Regarding claim 2, Fokoue-Nkoutche in view of Gold and further in view of Heimann teach the method of claim 1, as discussed above. The aforementioned references further teach the additional limitations wherein the first data, the second data and the third data are classified based on the database including information about the adverse reactions of the target drug and the drugs similar to the target drug based on a predetermined standard (Fokoue-Nkoutche’s [0004] discussing drug information database instances, with varying levels of completeness to their substance/contents, and see also [0017] and [0038] for further clarification, such that the basis of this information in this type of resource is used by the classifier per [0045] and [0064] to predict adverse drug events, e.g. for a candidate drug or for a particular drug-drug pair ([0017], [0035]-[0036]), and where the data involved in the modelling must meet a standard as Gold contemplates, e.g. either for inclusion in a gold set or possibly to adjust a gold set standard thereby allowing the standard to shift. Such a certainty standard is suggested by Gold but also exists per Heimann’s column 26 lines 45-62 and further per column 27 lines 11-59, where a particular uncertainty threshold is at least implied to result in a classification indicating “uncertainty” or “decision boundary” for example, which would be a “predetermined standard” as recited that governs classification in either Heimann’s and also Fokoue-Nkoutche’s as modified in view of Heimann). The motivation for combining the references is as discussed above in relation to claim 1.
Regarding claim 3, Fokoue-Nkoutche in view of Gold and further in view of Heimann teach the method of claim 1, as discussed above. The aforementioned references further teach the additional limitations wherein the machine learning model is a learning model that implements one of a gradient boosting machine and a random forest algorithm (Heimann’s classifier using random forest, per column 26 lines 33-44). The motivation for combining the references is as discussed above in relation to claim 1.
Regarding claim 7, the claim includes the same or similar limitations as discussed above in relation to claim 1, and is therefore rejected under the same rationale.
Regarding claim 8, the claim includes the same or similar limitations as discussed above in relation to claim 2, and is therefore rejected under the same rationale.
Regarding claim 9, the claim includes the same or similar limitations as discussed above in relation to claim 3, and is therefore rejected under the same rationale.
9. Claims 4 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Fokoue-Nkoutche in view of Gold and Heimann and further in view of Non-Patent Literature “About Train, Validation and Test Sets in Machine Learning” (“Shah”).
Regarding claim 4, Fokoue-Nkoutche in view of Gold and Heimann teach the method of claim 1, as discussed above. The aforementioned references teach the limitation wherein the learning of the machine learning model comprises learning the gold standard dataset (as discussed in relation to claim 1: Fokoue-Nkoutche’s FIG. 8 step 718 teaching the construction of an adverse drug reaction classifier, clarified per [0060] and [0064], where such a classifier using a training repository ([0060], [0063]-[0064]) would necessarily be trained using data that had instances of positive or negative indications of adverse drug reactions, so that it could successfully function and discern as described, and those training data examples that inform the classifier’s resulting model/implementation are what would be considered “ground truth” or as recited here “gold standard”, e.g. training data that is classified with certainty as a modification with Heimann might suggest) but not specifically as further limited to be randomly divided into a training dataset and an evaluation dataset according to a predetermined ratio. Rather, the Examiner relies upon SHAH to teach what Fokoue-Nkoutche etc. otherwise lack, see e.g., Shah’s pages 3-4 discussing (i) the “dataset split ratio” which may be determined with a user’s discretion as to the number of samples and the actual model being used (i.e., the recited “predetermined ratio”) and (ii) that the splitting of the dataset (per Shah’s italicized “Note on Cross Validation” on its page 4) may involve a step to “randomly choose” a particular percentage of the dataset to inform one of the split subsets, which the Examiner equates with the recited “randomly” dividing.
Both Fokoue-Nkoutche and Heimann relate to classifier frameworks and are therefore analogous. Shah similarly contemplates discussion of how to train such frameworks, and is therefore similarly directed and certainly relevant. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Shah’s discretion to split datasets with Fokoue-Nkoutche’s modified framework, with a reasonable expectation of success, such that the a user can with deliberation choose a ratio that suits their sample quantity at hand as well as the model they are working with.
Regarding claim 10, the claim includes the same or similar limitations as discussed above in relation to claim 4, and is therefore rejected under the same rationale.
10. Claims 5-6 and 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over Fokoue-Nkoutche in view of Gold and Heimann and Shah, and now further in view of Non-Patent Literature “Deep Learning Prediction of Adverse Drug Reactions Using Open TG-GATEs and FAERS Databases” (“Mohsen”).
Regarding claim 5, Fokoue-Nkoutche in view of Gold and Heimann and Shah teach the method of claim 4, as discussed above. The aforementioned references teach the additional limitation wherein the learning of the machine learning model comprises: first learning the machine learning model by implementing the training dataset (Shah’s setting of a training dataset, e.g. as discussed per claim 4, as incorporated into Fokoue-Nkoutche’s modified framework having a classifier and its own training scenario with training data that can be partitioned and used in accordance with Shah’s teachings) but not the further limitation for setting a threshold to have a maximum area under the curve (AUC) of a receiver operating characteristics (ROC) curve with the evaluation dataset for the first learned machine learning model. Rather, the Examiner relies upon MOHSEN to teach what Fokoue-Nkoutche etc. otherwise lack, see e.g., Mohsen’s pages 6-7, section 2.4, and page 8, sections 3.3-3.4, discussing accuracy estimation for a model being evaluated using “area under the ROC curve”, where true positive and false positive rates are evaluated per Figure 5 on page 12 to max out the area under the curve.
Like Fokoue-Nkoutche, Mohsen is directed to classification of drug data to predict adverse drug reactions using databases. Hence, they are similarly directed and therefore analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Mohsen’s accuracy evaluation methods as discussed here with Fokoue-Nkoutche’s modified framework, with a reasonable expectation of success, such that improvement methodologies known in the state of the art for the same type of framework can be extended to benefit Fokoue-Nkoutche’s.
Regarding claim 6, Fokoue-Nkoutche in view of Gold and Heimann and Shah, and now further in view of Mohsen teach the method of claim 5, as discussed above. The aforementioned references further teach the additional limitations wherein the determining of the possibility of adverse reactions for the prediction dataset comprises further determining whether there are adverse reactions based on the possibility of adverse reactions of the prediction dataset and the set threshold (as discussed per claim 5, incorporating Mohsen’s accuracy estimation methodologies into Fokoue-Nkoutche’s modified framework to obtain the sort of outcomes/results/benefits discussed per that reference in relation to claim 1 for example). The motivation for combining the references is as discussed above in relation to claim 5.
Regarding claim 11, the claim includes the same or similar limitations as discussed above in relation to claim 5, and is therefore rejected under the same rationale.
Regarding claim 12, the claim includes the same or similar limitations as discussed above in relation to claim 6, and is therefore rejected under the same rationale.
Response to Arguments
11. Applicants’ arguments filed 12/15/25, and reasserted by way of the RCE submission dated 1/16/26, have been fully considered, and will be addressed here below:
Regarding argument 1 on page 6 of Applicants’ Reply dated 12/15/25, the Examiner disagrees. See Fokoue-Nkoutche’s [0038], discussing what constitutes “known drug data”, inclusive of database content that is clearly apart and separate from the framework’s developed model. This paragraph was cited in the mappings for claim 2, as even previously rejected, and the Examiner emphasizes it here. In other words, Applicants have not fully considered the reference and rather have characterized it unfairly and too narrowly, and on this basis, the argument against the reference’s applicability is not persuasive. The Examiner has addressed the new limitation to the independent claims in the 103 rejection, citing this same rationale.
Regarding Applicants’ argument 2, the Examiner disagrees, essentially for the same reason give above. Applicants have misunderstood the reference and have interpreted it too narrowly. The same misreading noted above underpins this same argument addressed here.
Regarding Applicants’ argument 3, the Examiner agrees, and has updated the prior art search to better address the distinction being argued here. The throughline between the three references Fokoue-Nkoutche, Gold, and Heimann is the treatment and consideration of data, as obtained, for its use in modelling, and specifically uncertainties relating to the data. Even gold set data, as the primary reference Fokoue-Nkoutche discusses is subject to some measure of scrutiny and reconsideration, when read in view of Gold, for example. This is particularly compelling given the breadth of external sources that may constitute this data, as Fokoue-Nkoutche details in its [0038]. Given that this data is external and beyond the control of the model developer, it would prudent to assess, scrutinize, and validate it, as Gold and Heimann suggest.
Regaridng Applicants’ argument 4, please see the remarks provided just above in reply to argument 3, which the Examiner believes provides a concrete motivation for why the three references addressed above are readily combinable to one of ordinary skill in the art before the effective filing date of the claimed invention.
Conclusion
12. The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
CN 105046284 A (Zou)
WO 2020021857 A1 (Sato)
Non-Patent Literature “Fuzzy gold standards: Approaches to handling an imperfect reference standard” (Walsh)
Non-Patent Literature “Adaptive Validation Design: A Bayesian Approach to Validation Substudy Design With Prospective Data Collection” (Collin)
13. Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHOURJO DASGUPTA whose telephone number is (571)272-7207. The examiner can normally be reached M-F 8am-5pm CST.
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, Tamara Kyle can be reached at 571 272 4241. 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.
/SHOURJO DASGUPTA/Primary Examiner, Art Unit 2144