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 .
DETAILED ACTION
In the response filed on 16 June 2025, the following has occurred: claims 1-20 have been amended.
Now claims 1-20 are pending.
Claims 4, 11 and 18 do not have prior art rejections.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-20 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1, 9 and 17 of Patent No. 12,424,338. Although, the claims at issue are not identical, they are not patentably distinct as both the instant application and the ‘338 patent are directed at using various machine learning models to make determinations about a service request to provide a recommendation to a user, although the instant application is not directed at a telehealth request these features would be obvious in view of the teachings of Crossen (20210407682): see below but at least paragraph [0022]; Moustafa (20230154596): see below but at least paragraph [0108].
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 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.
Claims 1, 8 and 15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite method, apparatus (i.e., system) and computer program product for determining a consultation recommendation score for service request. The limitations of:
Claim 1, which is representative of claims 8 and 15
generating, […], a recommendation score using a […] framework that comprises a first […] model, a hybrid classification […] model, and a recommendation scoring […] model, wherein the […] framework is [… built …], based at least in part on a comparison between (i) an inferred recommendation score corresponding to a service request data object and (ii) a ground-truth outcome corresponding to the service request data object, wherein generating the recommendation score comprises: generating, using the first […] model, an output comprising a probabilistic data object for the service request data object for the service request data object, wherein the probabilistic data object is generated based at least in part on input data associated with the service request data object, generating an updated output by updating the output based on historical location data for a user associated with the service request data object, generating, using the hybrid classification […] model and based at least in part on the updated output, a variable-length classification for the service request data object., performing [… organization of data …] by mapping the updated output of the first […] model to a variable-length subset of a plurality of candidate classes based on the variable-length classification, wherein the plurality of candidate classes comprises one or more first classes, one or more second classes, and one or more hybrid classes, and generating, using the recommendation scoring […] model, the recommendation score for the service request data object based at least in part on the mapping and one or more scores for the variable-length subset; and initiating, […], one or more prediction-based actions based at least in part on the recommendation score.
, as drafted, is a system which under its broadest reasonable interpretation, covers a method of organizing human activity (i.e., managing personal behavior including following rules or instructions) via human interaction with generic computer components. That is, by a human user interacting with a computer using one or more processors (claim 1), a processor and memory (claim 8), a non-transitory computer readable medium (CRM) (claim 15), the claimed invention amounts to managing personal behavior or interaction between people, the Examiner notes as stated in 2106.04(a)(2), “certain activity between a person and a computer… may fall within the “certain methods of organizing human activity” grouping”. For example, by a human user interacting with a computer using one or more processors (claim 1), a processor and memory (claim 8), a non-transitory computer readable medium (CRM) (claim 15), the claim encompasses organization of collected data into a score using various models that make a determination about a service request for a human user to determine an action a human user should perform based on a result of the organized result. If a claim limitation, under its broadest reasonable interpretation, covers managing personal behavior or interactions between people but for the recitation of generic computer components, then it falls within the “method of organizing human activity” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of a computer using one or more processors (claim 1), a processor and memory (claim 8), a non-transitory computer readable medium (CRM) (claim 15), which implements the abstract idea. The computer using one or more processors (claim 1), a processor and memory (claim 8), a non-transitory computer readable medium (CRM) (claim 15) are recited at a high-level of generality (i.e., a general-purpose computers/ computer component implementing generic computer functions; see Applicant’s specification Figures 1-2, paragraphs [0090]-[0092]) such that it amounts no more than mere instructions to apply the exception using generic computer components. Accordingly, these additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
The claim recites the additional elements of “a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model… wherein the machine learning framework is trained, end-to-end… using the first machine learning model… using the hybrid classification machine learning model… using the recommendation scoring machine learning model…” and “performing transfer learning by mapping…”. The “a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model… wherein the machine learning framework is trained, end-to-end… using the first machine learning model… using the hybrid classification machine learning model… using the recommendation scoring machine learning model…” are recited at a high-level of generality (i.e., as use of a trained generic model that comprises a plurality of off-the shelf models) and amounts to generally linking the abstract idea to a particular technological environment. The “performing transfer learning by mapping…” is recited at a high-level of generality. (i.e., as training an off the shelf machine learning model in a generic manner) and amounts to generally linking the abstract idea to particular technological environment. Accordingly, even in combination, these additional elements do not integrate the abstract idea into a practical application. The claim is directed to an abstract idea.
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements of a computer using one or more processors (claim 1), a processor and memory (claim 8), a non-transitory computer readable medium (CRM) (claim 15), to perform the noted steps amounts to no more than mere instructions to apply the exception using generic hardware components. Mere instructions to apply an exception using a generic hardware component cannot provide an inventive concept (“significantly more”).
Also, as discussed above with respect to integration of the abstract idea into a practical application, the additional elements of “a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model… wherein the machine learning framework is trained, end-to-end… using the first machine learning model… using the hybrid classification machine learning model… using the recommendation scoring machine learning model…” and “performing transfer learning by mapping…” were considered generally linking the abstract idea to particular technological environment. The “a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model… wherein the machine learning framework is trained, end-to-end… using the first machine learning model… using the hybrid classification machine learning model… using the recommendation scoring machine learning model…” has been re-evaluated under the “significantly more” analysis and determined to amount to be well-understood, routine, and conventional elements/functions. As described in Rao (2019/0110754): see below but at least paragraph [0127]; Crossen (20210407682): paragraph [0046]; Moustafa (20230154596): paragraph [0075]; Loghmani (20120185275): paragraph [0018]; use of a trained machine learning model that comprises various models is well-understood, routine and conventional. The “performing transfer learning by mapping…” been re-evaluated under the "significantly more" analysis and determined to amount to be well- understood, routine, and conventional elements/functions. As described in Gunsola (20230154608): paragraph [0057]; Morgan (10,943,407): Column 3; Franchitti (2019/0171438): paragraph [0191]; Gilutz (2022/0230731): paragraph [0069]; Das (20190295703): paragraph [0039]; Zhang (20190110753): paragraph [0005]; training a model using transfer learning is well-understood routine and conventional. Well-understood, routine, and conventional elements/functions cannot provide “significantly more.” As such the claim is not patent eligible.
Claims 2-7, 9-14 and 16-20 are similarly rejected because either further define the abstract idea and/or do not further limit the claim to a practical application or provide as inventive concept such that the claims are subject matter eligible.
Claims 2, 9 and 16 further describe the fields of input data, however does not recite any additional elements are therefore cannot provide a practical application and/or significantly more.
Claims 3, 10 and 17 further describes determination for of conditions for a plurality of classes and use of a threshold, however does not recite any additional elements are therefore cannot provide a practical application and/or significantly more.
claims 4-6, 11-13 and 18-20 further describe various classes, however does not recite any additional elements are therefore cannot provide a practical application and/or significantly more.
Claims 7 and 14 recite the use of a calendar API, however does not recite any additional elements are therefore cannot provide a practical application and/or significantly more. \
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 1-3, 5-6, 8-10, 12-13, 15-17 and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Pub. No. 20210407682 (hereafter “Crossen”), in view of U.S. Patent Pub. No. 20230154596 (hereafter “Moustafa”), in view of U.S. Patent Pub. No. 20120185275 (hereafter “Loghmani”), in view of U.S. Patent Pub. No. 20230154608 (hereafter “Gunsola”).
Regarding (Currently Amended) claim 1, Crossen teaches a computer-implemented method (Crossen: Claim 1, “A method”, paragraph [0232], “The various techniques described herein may be implemented in the context of computer”) comprising:
generating, by one or more processors, a recommendation score using a machine learning framework that comprises a first machine learning model, a hybrid […] machine learning model, and a recommendation scoring machine learning model, wherein the machine learning framework is trained, end-to-end, […] (Crossen: paragraph [0018], “a machine learning model to determine whether a medical procedure may be automatically approved”, paragraph [0033], “the service provider may determine which of the potential diagnoses, medications, gaps in care, and/or clinical recommendations to present to the medical provider based on a score associated therewith and/or other ranking structure. In some examples, the score and/or ranking structure may be determined based on severity of the potential diagnoses, medications, gaps in care, and/or clinical recommendations”, paragraphs [0060]-[0061], “the learning component 130 may train one or more machine learned models associated with the recommendation component 122 to determine one or more clinical recommendations”. The Examiner notes that a claim may be rendered obvious where the limiting function is that of making a set of prior-known elements contiguous, i.e., bringing them together. However, the opposite is also true. In this case, the limiting function is that of splitting prior-known elements and or functionality into discrete elements: (a) a first machine learning model (b) a hybrid […] machine learning model, and (c) a recommendation scoring machine learning model, the one or more machine learning models, of the learning components as taught by Crossen in combination with the teachings of Moustaffa for the hybrid classification machine learning model teach the functionality of the claimed elements respectively (see mapping below). As such, this claim would be obvious to one of ordinary skill in the art at the time of the invention to make the learning component of Crossen separable without undue experimentation or risk of unexpected results, see In re Dulberg, 289 F.2d 522, 523, 129 USPQ 348, 349 (CCPA 1961). MPEP 2144.04), wherein generating the recommendation score comprises:
generating, using the first machine learning model, an output comprising a probabilistic data object for the service request data object for the service request data object (Crossen: paragraphs [0020]-[0022], “the service provider may determine the one or more potential diagnoses for the member based on the member data. For example, the service provider may determine, based on a medical history and series of lab results over a period of time, that the member may have type 2 diabetes”, paragraphs [0040]-[0044], “a single computing device for managing patient services, such as to prepare for a clinical visit, prepare a clinical visit, submit referrals, submit insurance claims, and the like… a medical provider 108 may submit a request for a clinical assessment 116 to the service provider computing device(s) 104 via the first instance of the application 102(1)”, paragraph [0046], “a diagnosis component 120 of the service provider computing device(s) 104 may be configured to determine the one or more potential diagnoses associated with the patient. In some examples, the diagnosis component 120 may utilize one or more machine learning techniques to determine the one or more diagnoses associated with the patient. In such examples, a machine learning model may be trained to identify one or more diagnoses and/or probabilities associated therewith”, paragraph [0232], “executed by the processor(s) of one or more computing devices”. The Examiner notes the provider submitting a request for service with patient data reads on a “service request data object”, under the broadest reasonable interpretation),
wherein the probabilistic data object is generated based at least in part on input data associated with the service request data object (Crossen: paragraphs [0020]-[0021], “The service provider may receive the information and generate the interface to assist a medical provider during the clinical visit… The member data may include demographic information, medical history… laboratory results… medical test results”, paragraphs [0032]-[0034], “the service provider may identify multiple potential diagnoses, medications, gaps in care, and/or clinical recommendations… the medical provider may access the interface to review member data, process medications, submit claims for service, submit referrals, or the like”, paragraph [0040], “a single computing device for managing patient services, such as to prepare for a clinical visit, prepare a clinical visit, submit referrals, submit insurance claims, and the like”, paragraph [0046], “the diagnosis component 120 may utilize one or more machine learning techniques to determine the one or more diagnoses associated with the patient. In such examples, a machine learning model may be trained to identify one or more diagnoses and/or probabilities associated therewith”),
generating an updated output by updating the output based on historical location data for a user associated with the service request data object (Crossen: paragraphs [0025]-[0027], “update the member data and/or process an insurance approval for additional procedures and/or testing based on the input… responsive to receiving the input, the service provider may update the member data, and/or process approval”, paragraph [0055], “the locations and/or medical providers may be determined based in part on a location associated therewith, a member location, a quality metric associated with the location and/or medical provider, a cost associated with the service at various locations and/or by various providers, insurance coverage approval, association with network (e.g., in network or out of network), and the like”, paragraph [0129]-[0132], “”, paragraph [0135], “”. The current location is used to update the score based on known historical data association with locations which teaches what is required under the broadest reasonable interpretation),
generating, using the hybrid […] machine learning model and based at least in part on the updated output, a variable-length [… determination …] for the service request data object, […] by mapping the updated output of the first machine learning model to a variable-length subset of a plurality of candidate [… results …] based on the variable-length [… determination …], wherein the plurality of candidate [… results …] comprises one or more first classes, one or more second classes, and one or more hybrid classes (Crossen: paragraph [0032]-[0033], “the service provider may identify multiple potential diagnoses, medications, gaps in care, and/or clinical recommendations and may determine a number of each to surface via the interface… the score and/or ranking structure may be determined based on severity of the potential diagnoses, medications, gaps in care, and/or clinical recommendations”, paragraph [0043], “The medical provider data may include… insurance billing history, procedural history, procedural and/or practice specialties, provider quality metric (e.g., based on quality of service (e.g., based on feedback from members, professional organizations, awards earned, etc.)”, paragraph [0048], “ranking may be based on a determined level of severity of the diagnosis (e.g., terminal, minor condition, etc.), a probability that the member 112 suffers from the associated condition (e.g., determined based on member data by the service provider 104, the machine learning model, etc.)”, paragraphs [0060]-[0061], “the learning component 130 may train one or more machine learned models associated with the diagnosis component 120, the recommendation component 122, and/or other components of the service provider computing device(s) 104… the learning component 130 may train one or more machine learned models associated with the recommendation component 122 to determine one or more clinical recommendations”, paragraph [0133], “the service provider may determine the quality metric based on the medical provider data. The quality metric may be based on a quality of service”, paragraph [0219], “provider experience with a procedure”. The Examiner notes severity is a probabilistic-based class, a quality metric is a provider-based class, and provider experience with a procedure is a hybrid-based class under the broadest reasonable interpretation), and
generating, using the recommendation scoring machine learning model, the recommendation score for the service request data object based at least in part on the mapping and one or more scores […] (Crossen: paragraph [0018], “a machine learning model to determine whether a medical procedure may be automatically approved”, paragraph [0033], “the service provider may determine which of the potential diagnoses, medications, gaps in care, and/or clinical recommendations to present to the medical provider based on a score associated therewith and/or other ranking structure. In some examples, the score and/or ranking structure may be determined based on severity of the potential diagnoses, medications, gaps in care, and/or clinical recommendations”, paragraphs [0060]-[0061], “the learning component 130 may train one or more machine learned models associated with the recommendation component 122 to determine one or more clinical recommendations”); and
initiating, by the one or more processors, one or more prediction-based actions based at least in part on the recommendation score (Crossen: paragraphs [0030]-[0033], “The clinical recommendations may include information to inform medical decisions, such as information associated with a member diagnosis (e.g., known diagnosis), a potential treatment for the member, studies published regarding a diagnosis or medical condition associated with the member, and the like… the service provider may determine which of the potential diagnoses, medications, gaps in care, and/or clinical recommendations to present to the medical provider based on a score associated therewith and/or other ranking structure”, paragraph [0057], “the recommendation component 122 may determine a number of clinical recommendations to include in the clinical assessment 118, such as based on a threshold”, paragraphs [0060]-[0061], “train a machine learning model of the recommendation component 122 to determine one or more locations and/or a provider that may be pre-approved by the service provider 104. In such examples, the determination for pre-approval may be provided to the medical provider 108 to inform a decision”. The Examiner notes approval for a service request teaches what is required of the claim under the broadest reasonable interpretation).
Crossen may not explicitly teach (underlined below for clarity):
generating, by one or more processors, a recommendation score using a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model, wherein the machine learning framework is trained, end-to-end, […], generating, using the hybrid classification machine learning model and based at least in part on the updated output, a variable-length classification for the service request data object, […] by mapping the updated output of the first machine learning model to a variable-length subset of a plurality of candidate classes based on the variable-length classification, wherein the plurality of candidate classes comprises one or more first classes, one or more second classes, and one or more hybrid classes,
the recommendation score for the service request data object based at least in part on the mapping and one or more scores for the variable-length subset;
Moustafa teaches generating, by one or more processors, a recommendation score using a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model, wherein the machine learning framework is trained, end-to-end, […], generating, using the hybrid classification machine learning model and based at least in part on the updated output, a variable-length classification for the service request data object, […] by mapping the updated output of the first machine learning model to a variable-length subset of a plurality of candidate classes based on the variable-length classification, wherein the plurality of candidate classes comprises one or more first classes, one or more second classes, and one or more hybrid classes (Moustafa: paragraphs [0002]-[0004], “generating, for each of a plurality of provider entities, a compliance profile data object”, paragraphs [0075]-[0078], “a classification machine learning model… configured and used to parse tokens (e.g., each word) of text of the reference data 510 and are configured to at least identify and output criteria”, paragraph [0085], “grouping procedural record data object 610 associated with the provider entity 602 according to applicable patient cohorts described by guideline data objects 505… guideline data objects 505 associated with the given service need condition are associated with some number of distinct or unique patient cohorts (e.g., a distinct or unique combination of patient-specific criteria), and the procedural record data objects 610 are grouped according to the distinct or unique patient cohorts”, paragraph [0095], “a classification machine learning model that may be configured (e.g., using supervised training) to output a binary prediction of whether a provider entity 602 is guilty of malpractice based at least in part on the compliance profile data object 605”),
the recommendation score for the service request data object based at least in part on the mapping and one or more scores for the variable-length subset (Moustafa: paragraphs [0002]-[0004], “performing predictive recommendations… provider entities (e.g., healthcare providers, groups or cohorts of healthcare providers, hospitals, organizations) are evaluated for compliance with respect to each of a set of service need conditions based at least in part on procedures and treatments performed by the provider entities… generating a plurality of guideline data objects for a plurality of service need conditions… generating, for each of a plurality of provider entities, a compliance profile data object including a plurality of compliance scores for the provider entity with respect to the plurality of service need conditions”, paragraphs [0027]-[0029], “the compliance profile data object is a vector of compliance scores… output a probabilistic value of the associated service need condition being present”, paragraph [0037], “evaluating compliance of a provider entity with clinical guidelines for treatment of service need conditions… recommendation of provider entities at least according to an evaluated or determined compliance of the provider entities with clinical guidelines for treatment of service need conditions”, paragraphs [0092]-[0095], “the compliance score with respect to the given service need condition is then generated based at least in part on a combination (e.g., an averaging, a sum)… the compliance profile data object 605 generated for a provider entity 602 is evaluated for malpractice detection. For instance, if a provider entity 602 has unsatisfactory compliance scores across a significant number of service need conditions, the provider entity 602 may be willfully or ignorantly guilty of malpractice”, paragraphs [0103]-[0105], “an overall score R may be assigned to each provider entity 602 by factoring in patient-provider match score M in combination with the compliance profile data object 605… compliance scores for each of a plurality of different service need conditions… an average of the compliance scores not satisfying a malpractice threshold may cause a malpractice-based action 820 to be performed”);
One of ordinary skill in the art before the effective filing would have found it obvious to include using a classifier machine learning model with scores associated with classes and using these class scores in part to make a recommendation as taught by Moustafa within the machine learning for scoring and recommending approval/denial of service requests as taught by Crossen with the motivation of “improving computational efficiency, operational reliability, and operational throughput of predictive recommendation systems” (Moustafa: paragraph [0017]).
Crossen and Moustafa may not explicitly teach (underlined below for clarity):
[…], wherein the machine learning framework is trained, end-to-end, based at least in part on a comparison between (i) an inferred recommendation score corresponding to a service request data object and (ii) a ground-truth outcome corresponding to the service request data object,
Lohgmani teaches […], wherein the machine learning framework is trained, end-to-end, based at least in part on a comparison between (i) an inferred recommendation score corresponding to a service request data object and (ii) a ground-truth outcome corresponding to the service request data object (Lohgmani: Figure 12, paragraph [0017], “using at least one of machine learning and pattern recognition”, paragraphs [0088]-[0092], “An anomaly detection algorithm (49) is shown in FIG. 12 in accordance with an illustrative embodiment of the present invention. The algorithm (49) can be implemented, for example, via the main server (9). The system uses a previously discovered mapping relationship (e.g., map m1 (74)) that predicts the relationship between a given set of input data… and an expected output or distribution or outputs (e.g., potential types of diagnoses) to compare the expected result obtained from the given map (i.e., map m1 (74)) with the actual output as indicated at (84)… the system can examine the data about a given medical invoice, or a series of medical invoices, for a given subscriber and compare the actual claimed data (such as claimed expenses, treatment provided, tests performed) with what the expected data would be using formulae obtained from various regressions methods based on the combination of actual input data (such as patient age, symptoms, prior ailments, or season) to determine whether the actual data varies from the expected data… the system runs multiple comparisons with multiple mapping relationships (using different mapping relationships previously discovered for different combinations of input and output data), and in each case, compares the expected output result with the actual output result (86) and (87)”),
One of ordinary skill in the art before the effective filing date would have found it obvious to include using comparison to ground truth data as taught by Lohgmani with the training of a machine learning framework as taught by Crossen and Mostafa with the motivation of “improve analyses (e.g., automated analysis of medical bills and health insurance documents)” (Lohgmani: paragraph [0003]).
Crossen, Moustafa and Loghmani may not explicitly teach (underlined below for clarity):
performing transfer learning by mapping the updated output of the first machine learning model to a variable-length subset of a plurality of candidate classes based on the variable-length classification, wherein the plurality of candidate classes comprises one or more first classes, one or more second classes, and one or more hybrid classes,
Gunsola teaches performing transfer learning by mapping the updated output of the first machine learning model to a variable-length subset of a plurality of candidate classes based on the variable-length classification, wherein the plurality of candidate classes comprises one or more first classes, one or more second classes, and one or more hybrid classes (Gunsola: Figure 4, paragraph [0057], “performing transfer learning by mapping output of a diagnostic classification machine learning model to the output of a recommendation classification machine learning model”),
One of ordinary skill in the art before the effective filing date would have found it obvious to include transfer learning as taught by Gunsola with the training of machine learning models as taught by Crossen, Mostafa and Lohgmani with the motivation of “reduce the storage-wise complexity and thus improve storage-wise efficiency of predictive data analysis systems that are configured to train and use diagnosis classification machine learning models and/or recommendation classification machine learning models by reducing the amount of training data those systems need to store to train diagnosis classification machine learning models and/or recommendation classification machine learning models” (Gunsola: paragraph [0014]).
Regarding (Currently Amended) claim 2, Crossen, Moustafa, Lohgmani and Gunsola teach the limitations of claim 1, and further teach wherein the input data comprise one or more symptom fields and one or more medical history fields (Crossen: paragraphs [0021]-[0022], “the service provider may generate the interface based on member data… The member data may include demographic information, medical history (e.g., previous diagnoses, medical procedures, surgeries, etc.), laboratory results (e.g., glucose, cholesterol, etc.), medical test results (e.g., Echo stress test result, EKG, etc.), member location information (e.g., home address, work address, etc.), pharmacological information (e.g., prescriptions, prescription fill information (e.g., last fill, expirations, etc.), preferred pharmacy, etc.)... the interface may include evidence to support the one or more potential diagnoses… symptoms, evidence, and/or health trends”, paragraphs [0071]-[0072], “responsive to selecting the selectable option 222, a window 224 may surface via the interface 200. The window 224 may include data fields 226, 228, 230, and 232 for the medical provider 218 to input relevant data about an upcoming clinical visit… a first data field 226… the second data field 228”. The Examiner notes there are various fields for input of symptoms and medical history and teaches what is required of the claim under the broadest reasonable interpretation).
The motivation to combine is the same as in claim 1, incorporated herein.
Regarding (Currently Amended) claim 3, Crossen, Moustafa, Lohgmani and Gunsola teach the limitations of claim 1, and further teach wherein: the probabilistic data object describes a plurality of conditions and a plurality of probabilities for the plurality of diagnosed conditions (Crossen: paragraphs [0032]-[0033], “identify multiple potential diagnoses, medications, gaps in care, and/or clinical recommendations and may determine a number of each to surface”, paragraph [0046], “the diagnosis component 120 may utilize one or more machine learning techniques to determine the one or more diagnoses associated with the patient. In such examples, a machine learning model may be trained to identify one or more diagnoses and/or probabilities associated therewith”);
a fist class of the one or more classes is associated with: (i) a related subset of the plurality of conditions that is associated with the fist class, and (ii) a classification score that is determined based at least in part on a probability of the plurality of probabilities for the related subset; and the service request data object is mapped to the first class responsive to the classification score for a candidate class of the plurality of candidate classes satisfies a classification score threshold (Crossen: paragraph [0022], “include evidence to support the one or more potential diagnoses. In such examples, the medical provider may access specific reasoning for a potential diagnosis, such as to discuss the symptoms, evidence, and/or health trends with the member”, paragraphs [0032]-[0033], “identify multiple potential diagnoses, medications, gaps in care, and/or clinical recommendations and may determine a number of each to surface… the score and/or ranking structure may be determined based on severity of the potential diagnoses, medications, gaps in care, and/or clinical recommendations”, paragraph [0048], “the inclusion of a potential diagnosis in the clinical assessment may be based on a probability associated therewith meeting or exceeding a threshold probability”, paragraph [0155], “the supporting evidence 810 includes a relevant member history, including a diagnosis, medications, lab results, and clinical guidelines 812”; Also see, Moustafa: paragraphs [0085]. The Examiner notes use of probability as a classification score teaches what is required under the broadest reasonable interpretation).
The motivation to combine is the same as in claim 1, incorporated herein.
Regarding (Currently Amended) claim 5, Crossen, Moustafa, Lohgmani and Gunsola teach the limitations of claim 1, and further teach wherein the one or more second classes comprise at least an availability class and a past interaction class (Crossen: paragraph [0020], “determine that a patient (e.g., member) associated with a service provider is scheduled for an appointment (e.g., clinical visit)”, paragraph [0024], “the medical provider may determine that a most recent glucose test was conducted too long in the past to effectively diagnose type 2 diabetes at the clinical visit”, paragraph [0055], “identify one or more locations and/or providers available to provide a service associated with the referral”, paragraph [0127], “The medical provider data may include… clinical visit times (e.g., average time, preferred (target) time, longest visit, shortest visit, etc.), insurance billing history, procedural history, procedural and/or practice specialties, provider quality metric (e.g., based on quality of service (e.g., based on feedback from members, professional organizations, awards earned, etc.)”. Also see, paragraph [0043]. The Examiner notes availability to provide a service is an availability class, and having a test recent enough to use is a past interaction class under the broadest reasonable interpretation).
The motivation to combine is the same as in claim 1, incorporated herein.
Regarding (Currently Amended) claim 6, Crossen, Moustafa, Lohgmani and Gunsola teach the limitations of claim 1, and further teach wherein the one or more hybrid classes comprise at least an individual past service class, a facility-based past service class, and a suboptimal past service class. (Crossen: paragraph [0055], “a quality metric associated with the location and/or medical provider”, paragraph [0127], “The medical provider data may include… clinical visit times (e.g., average time, preferred (target) time, longest visit, shortest visit, etc.), insurance billing history, procedural history, procedural and/or practice specialties, provider quality metric (e.g., based on quality of service (e.g., based on feedback from members, professional organizations, awards earned, etc.)”. Also see, paragraph [0043]; Moustafa: paragraph [0004], “generating, for each of a plurality of provider entities, a compliance profile data object including a plurality of compliance scores for the provider entity with respect to the plurality of service need conditions”, paragraph [0095], “the provider entity 602 may be willfully or ignorantly guilty of malpractice… output a binary prediction of whether a provider entity 602 is guilty of malpractice”. The Examiner notes a compliance profile for providers and facilities and a classification of malpractice teach the various required classed under the broadest reasonable interpretation).
The motivation to combine is the same as in claim 1, incorporated herein.
Regarding (Currently Amended) claim 7, Crossen, Moustafa, Lohgmani and Gunsola teach the limitations of claim 1, and further teach in response to determining that the recommendation score satisfies a recommendation score threshold, scheduling an entry on a calendar data object associated with a provider data object via interacting with an integrated calendar application programming interface (API) (Crossen: paragraph [0029], “processing the referral may include processing an approval for insurance payment, scheduling an appointment”; Moustafa: paragraph [0074], “an identifier may be associated with a predicted risk threshold”, paragraph [0108], “examples of prediction-based actions based at least in part on one or more output data items of various embodiments of the present invention include automatic appointment scheduling”; Lohgmani: Figure 20, paragraph [0113], “the user or subscriber (145) may use an application residing on the users' computing device (146) which provides an interface for the user to access various services such as… "Health Calendar" (149), e.g. to retrieve and update a subscriber's health related records using a calendar schema”).
The motivation to combine is the same as in claim 1, incorporated herein.
REGARDING CLAIM(S) 8 and 15
Claim(s) 8 and 15 is/are analogous to Claim(s) 1, thus Claim(s) 8 and 15 is/are similarly analyzed and rejected in a manner consistent with the rejection of Claim(s) 1.
REGARDING CLAIM(S) 9-10, 12-14, 16-17 and 19-20
Claim(s) 9-10, 12-14, 16-17 and 19-20 is/are analogous to Claim(s) 2-3 and 5-7, thus Claim(s) 9-10, 12-14, 16-17 and 19-20 is/are similarly analyzed and rejected in a manner consistent with the rejection of Claim(s) 2-3 and 5-7.
Response to Arguments
Applicant's arguments filed on 16 June 2025 have been fully considered but they are not persuasive. Applicant's arguments will be addressed below in the order in which they appear in the response filed on 16 June 2025.
Rejections under 35 U.S.C. § 103
Regarding the rejection of claims 1-20, the Examiner has considered the Applicant's arguments but does not find them persuasive. The Examiner has attempted to address all of the arguments presented by the Applicant; however, any arguments inadvertently not addressed are not persuasive for at least the following reasons:
Applicant argues:
As discussed during the interview, the cited references fail to teach or suggest at least "generating a recommendation score using a machine learning framework that comprises a first machine learning model, a hybrid classification machine learning model, and a recommendation scoring machine learning model, wherein the machine learning framework is trained, end-to-end, based at least in part on a comparison between (i) an inferred recommendation score corresponding to a service request data object and (ii) a ground-truth outcome corresponding to the service request data object" as recited by claim 1… The Office Action cites a combination of Crossen and Moustafa for allegedly disclosing the claim features of the independent claims… However, Crossen and Moustafa, individually or in combination, fail to teach or suggest
The Examiner respectfully disagrees.
It is respectfully submitted, in view of the amendments to the claims Lohgmani and newly applied Gunsola have been newly added to the independent claims (see above for mapping), and in combination with the motivations of “improve analyses (e.g., automated analysis of medical bills and health insurance documents)” (Lohgmani: paragraph [0003]) and “reduce the storage-wise complexity and thus improve storage-wise efficiency of predictive data analysis systems” (Gunsola: paragraph [0014]), would be prima facie obvious to include within the teachings of Crossen and Moustafa to one of ordinary skill in the art. Therefore, in view of the new grounds of rejection as necessitated by amendment the argument is not persuasive.
Rejections under 35 U.S.C. § 101
Regarding the rejection of claims 1-20, the Examiner has considered the Applicant's arguments but does not find them persuasive. The Examiner has attempted to address all of the arguments presented by the Applicant; however, any arguments inadvertently not addressed are not persuasive for at least the following reasons:
Applicant argues:
As stated above, the Office Action rejects claims 1-20 under 35 U.S.C. § 101 as allegedly being directed to "an abstract idea without significantly more." Office Action, p. 4. As amended, the claims recite an improved machine learning framework that connects, and trains end-to-end, a set of different machine learning models to improve their performance through joint predictions. See, e.g., Specification, as published [0015], [0017], [0019], [0021 ], [0067], [0070], and [0090]… In line with the court's requirements for patent-el