DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA and is in response to communications filed on 3/09/2026 in which claims 1-6, 8-21 are presented for examination.
Continued Examination Under 37 CFR 1.114
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 3/09/2026 has been entered.
Claim Rejections - 35 USC § 103
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 of this title, 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, 5, 8, 12, 14-15, 19, and 21 are rejected under 35 U.S.C. 103 as being unpatentable over James et al. US 20220019914 A1 (hereinafter referred to as “James”) in view of Jarvis et al. US 20220217135 A1 (hereinafter referred to as “Jarvis”) in view of Solmer et al. US 12141732 B1 (hereinafter referred to as “Solmer”) and further in view of Barnes et al. US 20170076046 A1 (hereinafter referred as “Barnes”).
As per claim 1, James teaches:
A computer-implemented method for generating a cross-temporal search result prediction for a predictive entity (James, [0027] – Various embodiments of the present invention provide robust predictive anomaly detection solutions that both enable inter-period cross-temporal inferences between the event records of various event periods as well as intra-period cross-temporal inferences between the various event records of a single event period, wherein anomaly detection is interpreted as a search which results in predictions of predictive anomalies. [0066] – The predictive data analysis computing entity 106 identifies event records associated with a predictive entity (e.g., with a patient profile), where each event record may be associated with an event period, an event date, and an event code),
the computer-implemented method comprising:
identifying, using one or more processors, a current input document (James, [0019] – Input data is interpreted as input documents. The embodiments also support confirmation of existing conditions contained within the group, when compared to external documentation (e.g. medical chart) for cross document discrepancies (e.g. conditions contained within claims and present amongst neighbors, but missing from the chart)) and
a plurality of historical input documents associated with the predictive entity (James, [0022] – Ingests input data from input data sources and creates clinical sentences [0025] – The term “event record” can be a data object that describes properties of a medical services delivery report, such as a generated medical claim. An event record may include event codes),
wherein a historical input document of the plurality of historical input documents comprises at least one per-modality segment for a plurality of historical input modalities (James, [0028]-[0030] – Event records are interpreted as historical input documents. Event codes as described for medical claims are interpreted as historical input modalities. Each medical claim is associated with a DOS day of a DOS (date of service) year, the event record profile for a DOS year that is associated with a first medical claim that occurs prior to a second medical claim may include a listing of the service codes for the first medical claim followed by the service codes of the second medical claim);
generating, using the one or more processors, a historical input embedding for the predictive entity based at least in part on the plurality of historical input documents (James, [0031] – The term “anomaly detection machine learning model” may refer to a data object that describes parameters and/or hyper-parameters of a machine learning model that uses mappings of a group of event record profiles to a multi-dimensional embedding space to determine predicted subject-matter correlations between the event record profiles, where the noted subject-matter correlations may then in turn be used to determine anomaly predictions for the mapped event record profiles), wherein:
(i) the historical input embedding is generated based at least in part on a plurality of per-document historical input embeddings for the plurality of historical input documents (James, [0028] – The term “event code” may refer to a data object that describes a property of an event record that can be used to generate an event record profile that can in turn be used to map the event record to a multi-dimensional space), and
(ii) generating a respective per-document historical input embedding for a particular historical input document comprises:
generating, based at least in part on each modality representation using the one or more processors and a cross-modality attention machine learning model, the respective per-document historical input embedding (James, [0089] – Historical cross-profile distance measure values across various recorded time periods is interpreted as per-document historical input embeddings);
generating, using the one or more processors and based at least in part on the historical input embedding, a current input embedding for the current input document, and a plurality of referential embeddings for a plurality of reference documents, the cross-temporal search result prediction (James, [0072] – Following determining an initial temporally-related subset of event records for a particular event record, the initial temporally-related subset of event records may be filtered to exclude any event records that do not have all of the one or more non-temporal properties in common with the particular event record, wherein initial temporally-related subset of event records is interpreted as a set of referential embeddings. [0090]-[0091] – The predictive data analysis computing entity 106 generates the anomaly predictions based on each inclusion ratio for a target code that is associated with at least one event record profile in the neighboring subset); and
Although James teaches predictive data analysis using event code documents and vector representations with word embeddings in [0081], James doesn’t explicitly teach cross verification of data in conjunction with machine learning, however, Jarvis teaches:
generating one or more modality representations for each historical input modality of the plurality of historical input modalities based at least in part on one or more input tokens of the particular historical input document associated with the historical input modality of a per-modality cross-token attention machine learning model (Jarvis, [0043] – A security token may also be dynamically generated and transmitted to the user via different modalities for cross-verification in real time, alleviating the user from memorizing the pre-configured security token associated with the app. [0044] – The content and/or format of OTP message may be customized using various machine learning techniques such as an AI model established based on the user's historical and habitual data, and a group of users' historical and habitual data); and
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify James’s invention in view of Jarvis in order to include cross-verifications of data; this is advantageous because it improves the security of the system (Jarvis, paragraph [0043]).
Although James as modified with Rossetto doesn’t explicitly teach that an input document is rendered, Solmer teaches:
generating, … the current input document (Solmer, column 23, lines 63-67 and column 24, lines 1-2 – A document may be uploaded by a user for modeling and indexing, and the criteria for splitting the document may be adjusted via user interface, in other embodiments such criteria for splitting may be solely based on automated inferences based on, at least in part, the aforementioned criteria for division for modeling or known equivalents. Column 66, lines 4-6 – Conceptual summarization may be controlled by the same slider as highlighting (as depicted)) and
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify James’s invention as modified in view of Solmer in order to include the input document as rendered on an interface; this is advantageous because it allows the system to provide analysis of an uploaded document (Solmer, column 66, lines 4-6).
James as modified doesn’t explicitly teach that the documents are viewable simultaneously, however, Barnes teaches:
generating, using the one or more processors, a prediction output user interface configured to simultaneously render the current input document and one or more reference document sections, of the plurality of reference documents, associated with the cross-temporal search result prediction (Barnes, [0012] – Search engine to search for patients with specific attributes; a graphing tool that can display disparate clinical variables in a single chart; a virtual PinBoard for users to identify relevant patient information for board meetings; an image viewing application that can provide for “side by side” comparison of images from different information systems; structured reporting functionality that incorporates system aggregated patient information and board recommendations. [0097] – The graphs not only include the visualization of numerical data, but also provide access to qualitative images and reports that correspond to specific dates and times, wherein this is interpreted as temporal data. [0119] – The user may select one of the patients in the results section 82 and be provided with the medical file of the selected patient. In another embodiment, the user may select several patients in the results section 82 and be provided with a comparison of the attributes of the selected patients. The attribute selection section 84 can provide an interface to select clinical attributes to be used to initiate an updated search for similar patients).
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify James’s invention as modified in view of Barnes in order to include the input document as rendered on an interface; this is advantageous because it enables medical personnel to search the database 35, which includes information from EMR system 20 and information systems 22, such as radiology RIS/PACS, digital pathology and LIS, using both automated and manual queries (Barnes, [0118]).
As per claim 5, James as modified teaches:
The computer-implemented method of claim 1, wherein the plurality of historical input modalities are defined by a modality taxonomy that is shared across the plurality of historical input documents (James, [0028]-[0030] – Event records are interpreted as historical input documents. Event codes as described for medical claims are interpreted as historical input modalities. [0033] – Ground-truth HCC codes may be determined based on historical medical chart review result data, and may be used to determine HCCs of new event record profiles by utilizing an anomaly detection machine learning model as described above).
Claims 8 and 12 are directed to an apparatus performing steps recited in claims 1 and 5 with substantially the same limitations. Therefore, the rejections made to claims 1 and 5 are applied to claims 8 and 12.
As per claim 14, James as modified teaches:
The system of claim 8, wherein the cross-temporal search result prediction describes a ranked list of reference document sections from a selected subset of the plurality of referential embeddings (James, [0024] – Charts may then be ranked and targeted according to the chart metadata and risk anomaly propensity. The highest charts may then be prioritized for retrieval and coding).
Claims 15 and 19 are directed to an apparatus performing steps recited in claims 1 and 5 with substantially the same limitations. Therefore, the rejections made to claims 1 and 5 are applied to claims 15 and 19.
As per claim 21, James as modified teaches:
The computer-implemented method of claim 1, wherein the per-modality cross-token attention machine learning model used to generate the one or more modality representations is selected from a plurality of per-modality cross-token attention machine learning models, and wherein each per-modality cross-token attention machine learning model is associated with a unique input modality type of a plurality of input modality types (Jarvis, [0043] – A security token may also be dynamically generated and transmitted to the user via different modalities for cross-verification in real time, alleviating the user from memorizing the pre-configured security token associated with the app).
Claims 2-5, 9-11, 13, 16-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over James in view of Jarvis in view of Solmer in view of Barnes and further in view of Rossetto et al. US 20240004911 A1 (hereinafter referred to as “Rossetto”).
As per claim 2, although James as modified teaches predictive data analysis using event code documents and vector representations with word embeddings in [0081], James doesn’t explicitly teach cross-token attention machine learning to generate representations of documents, and generating a display based on the reference sections however, Rossetto teaches:
The computer-implemented method of claim 1, wherein the plurality of referential embeddings comprises, for each reference document, a per-document referential embedding and a plurality of per-section referential embeddings for a plurality of reference document sections of the reference document (Rossetto, [0015] – The first machine-learning model segments the input document into sentences and generates a set of embeddings corresponding to the sentences to generate an initial or rough topic segmentation of the document).
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify James’s invention in view of Rossetto in order to include cross-segment positional data; this is advantageous because it improves readability of the document (Rossetto, paragraph [0014]).
As per claim 3, James as modified with Rossetto teaches:
The computer-implemented method of claim 2, wherein the plurality of per-section referential embeddings are generated based at least in part on the plurality of reference document sections and using a cross-section attention machine learning model (Rossetto, [0015] – The first machine-learning model segments the input document into sentences and generates a set of embeddings corresponding to the sentences to generate an initial or rough topic segmentation of the document [0022] – The sentence segmentation manager 110, embeddings generator 120, and transformation manager 130 co-operatively execute Cross-segment Bidirectional Encoder Representations from Transformers (BERT)).
As per claim 4, James as modified with Rossetto teaches:
The computer-implemented method of claim 2, wherein generating the cross-temporal search result prediction comprises:
generating a cross-temporal input embedding based at least in part on the historical input embedding and the current input embedding (James, [0027] – Various embodiments of the present invention provide robust predictive anomaly detection solutions that both enable inter-period cross-temporal inferences between the event records of various event periods as well as intra-period cross-temporal inferences between the various event records of a single event period, wherein anomaly detection is interpreted as a search which results in predictions of predictive anomalies. [0066] – The predictive data analysis computing entity 106 identifies event records associated with a predictive entity (e.g., with a patient profile), where each event record may be associated with an event period, an event date, and an event code);
generating, based at least in part on the cross-temporal input embedding and a plurality of per-document referential embeddings for the plurality of reference documents, a defined-size related reference document subset of the plurality of reference documents (Rossetto, [0015] – A sentence is interpreted as a defined-size related reference document subset); and
generating, based at least in part on the cross-temporal input embedding and each per- section referential embedding for reference document sections that are associated with the defined-size related reference document subset, the cross-temporal search result prediction (Rossetto, [0016] – The final topic segmented document associated with the merchant system generated by the second phase of machine-learning processing can be stored in a knowledge graph associated with the merchant system or provisioned to an end-user system (e.g., in response to a search query for which the topic segmented document is part of the search results) or publisher system (e.g., in response to a search query initiated via the publisher system or for storage by the publisher system for future processing)).
As per claim 6, James as modified with Rossetto teaches:
The computer-implemented method of claim 1, wherein the cross-modality attention machine learning model is a bidirectional transformer model (Rossetto, [0022] – Embeddings generator 120, and transformation manager co-operatively execute Cross-segment Bidirectional Encoder Representations from Transformers (BERT) processing).
Claims 9-11, and 13 are directed to an apparatus performing steps recited in claims 2-4, and 6 with substantially the same limitations. Therefore, the rejections made to claims 2-4, and 6 are applied to claims 9-11, and 13.
Claims 16-18, and 20 are directed to a computer program product performing steps recited in claims 2-4, and 6 with substantially the same limitations. Therefore, the rejections made to claims 2-4, and 6 are applied to claims 16-18, and 20.
Response to Arguments
Applicant’s arguments filed 3/09/2026 have been fully considered but they are not persuasive. Applicant’s arguments begin on page 8 of Remarks which is addressed below:
Argument: Applicant argues in Remarks on page 8 that Applicant disagrees with the general rejection based on the prior art of record because James as modified specifically with Rossetto (used as the reference for mapping the specific limitations of the rejection) makes no mention of processing different modalities through different models and then fusing them.
Response: This argument as well as other arguments pertaining to the limitations regarding processing different modalities through different modals have been fully considered. However, the arguments are generally moot because a new reference has been found which teaches these limitations: namely, Jarvis et al.
As per claim 21, James as modified with the prior art of record teaches the limitations of the claim. See rejection above.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Shukla et al. US 20230316345 A1 teaches predictive data analysis operations by generating integrative predicted scores based at least in part on at least one of: within-cluster consistency scores determined for clusters determined using a first clustering scheme (e.g., a service clustering scheme), within-cluster consistency scores determined for clusters determined using a second clustering scheme (e.g., a recipient clustering scheme), cross-cluster consistency scores, and cross-temporal consistency scores (abstract).
Sharma et al. US 20240070479 A1 teaches an attention-based classifier machine learning model with both intra-token and cross-token positional indicator data, an approach which increases training speed of the attention-based classifier machine learning model given a constant target predictive accuracy in [0066].
GOPINATH,Divyaetal."Fast,StructuredClinicalDocumentationViaContextualAutocomplete,"ProceedingsofMachineLearningResearch,Vol.106,pp.1-28,September18,2020,(Year:2020).
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Matthew Ellis whose telephone number is (571)270-3443. The examiner can normally be reached on Monday-Friday 8AM-5PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Neveen Abel-Jalil can be reached on (571)270-0474. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
March 19, 2026
/MATTHEW J ELLIS/Primary Examiner, Art Unit 2152