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
Status of the Application
Claims 1-20 are currently pending in this case and have been examined and addressed below. This communication is a Non-Final Rejection in response to the Claims filed on 07/28/2025.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 03/04/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Claims 1 and 2-19 fall within the statutory category of a process. Claim 20 falls within the statutory category of an apparatus or system.
Step 2A, Prong One
As per Claims 1 and 20, the limitations of preprocessing the structured data to convert raw event requests into a structured token sequence including: performing a sorting algorithm to organize codes within each event request in the structured data into a clinically logical sequence, including chronologically ordering event requests for a respective individual to form a temporally sequenced dataset thereby enabling a machine learning model to learn chronological order of events; inserting one or more delimiter tokens into the structured data for concatenating intra-event request codes, inter-event requests codes for the respective individual, and data for different individuals, thereby enabling batch data processing; and tokenizing the structured data to obtain a sequence of tokens, wherein the tokenizer is trained on event request data with a predetermined vocabulary size; and training a causal language model using the structured token sequence to predict an outcome, wherein training the causal language model comprises predicting a next code in the structured token sequence based on prior codes, thereby generating a sequence of codes for each event request in a causally coherent manner, wherein predicting the next code is modeled as a probability distribution over possible codes, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The steps of performing a sorting algorithm to organize codes into a clinically logical sequence, inserting delimiter tokens into the structured data, tokenizing the structured data to obtain a sequence of tokens, are concepts performed including observation, evaluation, judgement and opinion in the human mind. If a claim limitation, under its broadest reasonable interpretation, covers the performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. The steps of the tokenizer trained on event request data with a predetermined vocabulary size and training a causal language model using the structured token sequence to predict an outcome describes the training as being performed by predicting a next code in the sequence which is modeled as a probability distribution over possible codes. The training of the model is considered to be part of the abstract idea because they fall under data manipulations that humans perform and thus are part of the mental process. The steps of performing a sorting algorithm to organize codes, including chronologically ordering event requests and tokenizing the structured data to obtain a sequence of tokens can also fall into the grouping of mathematical concepts as mathematical relationships and calculations can be used to carry out the steps. As per MPEP 2106.04(a), a claim can recite more than one judicial exception and claims which recite a series of steps that recite mental steps which are also mathematical calculations are identified as both. Accordingly, the claims recite an abstract idea.
As per Claim 2, the limitations of preprocessing the structured data to convert raw event requests into a structured token sequence and training a causal language model using the structured token sequence to predict an outcome, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The steps of preprocessing the structured data to convert raw event requests into structured token sequence and training a causal language model, are concepts performed including observation, evaluation, judgement and opinion in the human mind. If a claim limitation, under its broadest reasonable interpretation, covers the performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. The type of training utilized by the claimed invention is not described by the Applicant. As such the Examiner is required to analyze the training step given the broadest reasonable interpretation. The training of the model is considered to be part of the abstract idea because they fall under data manipulations that humans perform and thus are part of the mental process. Accordingly, the claims recite an abstract idea.
Step 2A, Prong Two
The judicial exception is not integrated into a practical application because the additional elements and combination of additional elements do not impose meaningful limits on the judicial exception. In particular, Claim 20 recites the additional elements – a computer system comprising one or more processors, a display, and a memory that stores programs for execution by the processors. The computer system in these steps is recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using a generic computer component. Accordingly, this 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. Claims 1 and 20 recite preserving the delimiter tokens to maintain context of event request data, which is a function of storing data. Storing data amounts to mere instructions to apply the exception, as per MPEP 2106.05(f)(2). Claims 1 and 20 also recite the use of a tokenizer to carry out the steps of the abstract idea. The tokenizer is recited at a high-level of generality to execute the abstract idea, which amounts to mere instructions to apply the exception. As per MPEP 2106.05(f)(2), the use of a mathematical algorithm applied on a general purpose computer has been found by the courts to do no more than invoke computers as a tool and amount to mere instructions to apply the exception. The claims also recite the additional element of obtaining a training dataset that includes structured data including codes, which amounts to insignificant extra-solution activity, as in MPEP 2106.05(g), because the steps of obtaining a training dataset is mere data gathering in conjunction with the abstract idea where the limitation amounts to necessary data gathering and outputting, (i.e., all uses of the recited judicial exception require such data gathering or data output). See Mayo, 566 U.S. at 79, 101 USPQ2d at 1968; OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1092-93 (Fed. Cir. 2015) (presenting offers and gathering statistics amounted to mere data gathering). Because the additional elements do not impose meaningful limitations on the judicial exception, the claim is directed to an abstract idea.
Step 2B
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements when considered both individually and as an ordered combination do not amount to significantly more than the abstract idea. As discussed above with the respect to integration of the abstract idea into a practical application, the additional element of a computer system to perform the method of the invention amounts to no more than mere instructions to apply the exception using a generic computing component. The system including the processor, display, and memory are recited at a high level of generality and are recited as generic computer components where the processor, display, and memory are not specified or limited by the specification, and therefore do not add meaningful limitations to the abstract idea beyond mere instructions to apply an exception. The claims also recite preserving the delimiter tokens to maintain context of event request data and use of a tokenizer to carry out the steps of the abstract idea which amount to mere instructions to apply the exception. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claims also include the additional elements of obtaining a training dataset that includes structured data including codes which is an element that is well-understood, routine and conventional computer functions in the field of data management because it is claimed at a high level of generality and include receiving or transmitting data as well as storing and retrieving information from memory, which have been found to be well-understood, routine and conventional computer functions by the Court (MPEP 2106.05(d)(II)(i) Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) (using a telephone for image transmission); OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015) (sending messages over a network); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014) (computer receives and sends information over a network); but see DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245, 1258, 113 USPQ2d 1097, 1106 (Fed. Cir. 2014) ("Unlike the claims in Ultramercial, the claims at issue here specify how interactions with the Internet are manipulated to yield a desired result‐‐a result that overrides the routine and conventional sequence of events ordinarily triggered by the click of a hyperlink." (emphasis added) and (iv) Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93). Looking at the limitations as an ordered combination adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of the computer or improves another technology. The claims do not amount to significantly more than the underlying abstract idea.
Dependent Claims
Dependent Claims 3-19 add further limitations which are also directed to an abstract idea. For example, Claims 3-4, 7, 11-12 further limit or specify the limitations of independent claim 2 and are therefore directed to the same abstract idea. Claim 5 includes performing a sorting algorithm to organize codes within each event request in the structured data into a clinically logical sequence, including chronologically ordering event requests for a respective individual to form a temporally sequenced dataset thereby enabling a machine learning model to learn chronological order of events which can be performed by mental evaluation, observation, judgment, and opinion which falls into the abstract grouping of a mental process. This can also fall into a mathematical concept as performing a sorting algorithm is a mathematical relationship or calculation. Claim 6 provides details of the mathematical equation used in the sorting algorithm and therefore falls into the abstract grouping of mathematical concepts. Claim 8 includes tokenizing the structured data to obtain a sequence of tokens, wherein the tokenizer is trained on event requests data with a predetermined vocabular size, which falls into the abstract idea of mental processes and mathematical concepts similar to Claim 1. Claim 8 also includes a tokenizer and preserving delimiter tokens which is analogous to storing data, which amount to mere instructions to apply the exception. Claim 9 provides details of the mathematical relationships used by the tokenizer to create a fixed-size vocabulary balancing medical language specificity and therefore falls into the abstract grouping of mathematical concepts. Claim 10, 13-15 provides details of the mathematical relationships used to train the causal language model and therefore falls into the abstract grouping of mathematical concepts. Claims 16-17 includes the use of zero-shot prompting for forecasting outcomes which falls into the abstract grouping of mathematical concepts. Claim 18 includes fine-tuning the trained causal language model to generate a synthetic dataset which falls into the abstract grouping of mathematical concepts. Claim 19 provides details of the mathematical relationships used to fine-tune the trained causal language model and therefore falls into the abstract grouping of mathematical concepts. Because the additional elements do not impose meaningful limitations on the judicial exception and the additional elements are well-understood, routine and conventional functionalities in the art, the claims are directed to an abstract idea and are not patent eligible.
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.
Claims 1-3, 5, 8, 13-14, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Rose et al. (US 2026/0058015 A1), hereinafter Rose, in view of Wullschleger (P. Wullschleger, S. Lionetti, D. Daly, F. Volpe and G. Caro, "Auto-Regressive Self-Attention Models for Diagnosis Prediction on Electronic Health Records," 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 2022, pp. 1950-1956), hereinafter Wullschleger.
As per Claims 1 and 20, Rose discloses a computer system for predicting outcomes using large language models, comprising:
one or more processors ([0065]);
a display (see fig. 2); and
memory, wherein the memory stores one or more programs configured for execution by the one or more processors ([0064]), and the one or more programs comprising instructions for:
obtaining a training dataset that includes structured data including codes ([0012] structured health care record data comprises clinical codes which are obtained by accessing the database);
preprocessing the structured data to convert raw event requests into a structured token sequence ([0014] converting the electronic health record data into a text sequence, where the text is represented by a sequence of text tokens; [0118] sequence of text/codes processed into tokens), including:
performing a sorting algorithm to organize codes within each event request in the structured data into a clinically logical sequence, including chronologically ordering event requests for a respective individual to form a temporally sequenced dataset thereby enabling a machine learning model to learn chronological order of events ([0159] dataset of each patient which is converted to a sequence ordered in time, organized by time stamps of the data);
inserting one or more delimiter tokens into the structured data for concatenating intra-event request codes, inter-event request codes for the respective individual, and data for different individuals, thereby enabling batch data processing ([0112-0113] insert tokens into the structured data and concatenate the word embeddings which maintain the positional information of the data, [0114] sum the embeddings to form an input representation); and
tokenizing the structured data using a tokenizer to obtain a sequence of tokens, wherein the tokenizer preserves the one or more delimiter tokens to maintain context of event request data ([0112] prepare text by tokenization, convert text sequence into words/tokens with separator tokens in between), wherein the tokenizer is trained on event request data with a predetermined vocabulary size ([0118] tokens are formed using a fixed size vocabulary); and
training a causal language model using the structured token sequence to predict an outcome ([0010] training a machine learning model to predict patient condition, i.e. outcome, based on text sequence).
However, Rose may not explicitly disclose the following which is taught by Wullschleger: wherein training the causal language model comprises predicting a next code in the structured token sequence based on prior codes (Page 1951 Col. 1 next-diagnosis prediction is causal language modelling when the data is text, Page 1954 Col. 1 B. Training use elements in a sequence to train the model to predict the next element/code), thereby generating a sequence of codes for each event request in a causally coherent manner, wherein predicting the next code is modeled as a probability distribution over possible codes (Page 1950 Col. 2 next-diagnosis prediction is based on calculating the probability of observing code).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of generating a sequence of codes including predicting the next code using causal language modelling from Wullschleger with the method of predicting an outcome from a tokenized sequence of medical codes from Rose in order to provide diagnostic granularity to provide more accurate results and help to diagnosis rare conditions (Wullschleger Page 1590, Col. 1 Abstract).
As per Claim 2, Rose discloses a method of training a causal language model for predicting clinical outcomes, the method comprising:
obtaining a training dataset that includes structured data including codes ([0012] structured health care record data comprises clinical codes which are obtained by accessing the database);
preprocessing the structured data to convert raw event requests into a structured token sequence ([0014] converting the electronic health record data into a text sequence, where the text is represented by a sequence of text tokens); and
training a causal language model using the structured token sequence to predict an outcome ([0010] training a machine learning model to predict patient condition, i.e. outcome, based on text sequence).
However, Rose may not explicitly disclose the following which is taught by Wullschleger: the model is a causal language model (Page 1951 Col. 1 next-diagnosis prediction is causal language modelling when the data is text).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of generating a sequence of codes including predicting the next code using causal language modelling from Wullschleger with the method of predicting an outcome from a tokenized sequence of medical codes from Rose in order to provide diagnostic granularity to provide more accurate results and help to diagnosis rare conditions (Wullschleger Page 1590, Col. 1 Abstract).
As per Claim 3, Rose and Wullschleger discloses the method of Claim 2. Rose also teaches the structured data includes a respective dataset for a plurality of individuals ([0008] data is made of structured electronic health records for a plurality of patients), each dataset including a plurality of event requests, wherein each individual has a corresponding set of event requests, each event request comprising a set of codes, wherein each code is either a diagnosis code or a procedural code ([0008] electronic health records for each patient include event requests such as a plurality of clinical observations associated with a description/code, [0012] electronic health record data comprises clinical codes for the clinical observations, including code for disease diagnosis or a clinical procedure).
As per Claim 5, Rose and Wullschleger discloses the method of Claim 2. Rose also teaches performing a sorting algorithm to organize codes within each event request in the structured data into a clinically logical sequence, including chronologically ordering event requests for a respective individual to form a temporally sequenced dataset thereby enabling a machine learning model to learn chronological order of events ([0018] combine text sequences in the order of associated time stamp, [0159] dataset of each patient which is converted to a sequence ordered in time, organized by time stamps of the data).
As per Claim 8, Rose and Wullschleger discloses the method of Claim 2. Rose also teaches tokenizing the structured data using a tokenizer to obtain a sequence of tokens, wherein the tokenizer preserves one or more delimiter tokens to maintain context of data ([0112] prepare text by tokenization, convert text sequence into words/tokens with separator tokens in between), wherein the tokenizer is trained on event requests data with a predetermined vocabulary size ([0118] tokens are formed using a fixed size vocabulary).
As per Claim 13, Rose and Wullschleger discloses the method of Claim 2. Wullschleger also teaches training the causal language model using the structured token sequence comprises predicting a next code in the structured token sequence based on prior codes (Page 1951 Col. 1 next-diagnosis prediction is causal language modelling when the data is text, Page 1954 Col. 1 B. Training use elements in a sequence to train the model to predict the next element/code), thereby generating a sequence of codes for each claim in a causally coherent manner (Page 1950 Col. 2 next-diagnosis prediction is based on calculating the probability of observing code).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of generating a sequence of codes including predicting the next code using causal language modelling from Wullschleger with the method of predicting an outcome from a tokenized sequence of medical codes from Rose in order to provide diagnostic granularity to provide more accurate results and help to diagnosis rare conditions (Wullschleger Page 1590, Col. 1 Abstract).
As per Claim 14, Rose and Wullschleger discloses the method of Claim 13. Wullschleger also teaches predicting the next code is modeled as a probability distribution over possible codes (Page 1950 Col. 2 next-diagnosis prediction is based on calculating the probability of observing code).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of generating a sequence of codes including predicting the next code using causal language modelling from Wullschleger with the method of predicting an outcome from a tokenized sequence of medical codes from Rose in order to provide diagnostic granularity to provide more accurate results and help to diagnosis rare conditions (Wullschleger Page 1590, Col. 1 Abstract).
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Rose (US 2026/0058015 A1), in view of Wullschleger (P. Wullschleger, S. Lionetti, D. Daly, F. Volpe and G. Caro, "Auto-Regressive Self-Attention Models for Diagnosis Prediction on Electronic Health Records," 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 2022, pp. 1950-1956), in view of Troeger et al. (US 2024/0363237 A1), hereinafter Troeger.
As per Claim 4, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Troeger: each event request in the structured data corresponds to an individual-provider encounter, aggregates medical codes in a non-sequential order (Troeger [0019] medical record datasets are records associated with a healthcare provider, i.e. encounters, which are aggregated based on non-sequential order such as based on insurance provider).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of aggregating codes from patient data of patient encounters in non-sequential order from Troeger with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to support data driven decision making such as insight into health outcomes (Troeger [0007]).
Claims 7, 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over Rose (US 2026/0058015 A1), in view of Wullschleger (P. Wullschleger, S. Lionetti, D. Daly, F. Volpe and G. Caro, "Auto-Regressive Self-Attention Models for Diagnosis Prediction on Electronic Health Records," 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 2022, pp. 1950-1956), in view of Oren et al. (US 2019/0034590 A1), hereinafter Oren.
As per Claim 7, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Oren: preprocessing the structured data comprises: intra-event request codes, inter-event request codes for the respective individual, and data for different individuals, thereby enabling batch data processing ([0127] data from the EHR includes series of events, i.e. inter-event, and also details of the events, i.e. intra-event, [0011] health records are from a multitude of patients to be aggregated).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of process structured health data from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to enable healthcare provider to allocate attention efficiently with the large amount of information available to analyze and consider (Oren [0004]).
As per Claim 11, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Oren: the training dataset includes event requests data that covers a plurality of individual demographics and conditions from a plurality of care settings (Abstract electronic health records from a multitude of patients from different demographics, health conditions).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of process structured health data from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to enable healthcare provider to allocate attention efficiently with the large amount of information available to analyze and consider (Oren [0004]).
As per Claim 12, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Oren: the training dataset comprises billions of event requests corresponding to millions of individuals, tens of thousands of diagnosis codes, and tens of thousands of unique procedure codes, and wherein the training dataset excludes invalid codes resulting from intake or ingestion errors ([0011] health records are from multitude of patients up to even millions of patients from plurality of different institutions which are in different formats, Examiner interprets different formats to include different code types, [0105] exclude encounters without ICD-9 diagnosis, i.e. invalid).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of process structured health data from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to enable healthcare provider to allocate attention efficiently with the large amount of information available to analyze and consider (Oren [0004]).
Claims 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over Rose (US 2026/0058015 A1), in view of Wullschleger (P. Wullschleger, S. Lionetti, D. Daly, F. Volpe and G. Caro, "Auto-Regressive Self-Attention Models for Diagnosis Prediction on Electronic Health Records," 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 2022, pp. 1950-1956), in view of Frayling (E. Frayling, J. Lever, G. McDonald, "Zero-shot and Few-shot Generation Strategies for Artificial Clinical Records", SIGIR '24, Washington, DC,14 March 2024, p. 1-5), hereinafter Frayling.
As per Claim 16, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Frayling: using zero-shot prompting for forecasting outcomes (Page One, Abstract use of zero-shot model to achieve results).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of zero-shot prompting from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to generate more accurate and contextually relevant results (Frayling Page One Abstract).
As per Claim 17, Rose, Wullschleger, and Frayling discloses the method of Claim 16. Rose also teaches analyzing output generated by the language model for event occurrence ([0021] output of model predicts the clinical outcome based on the output representation, i.e. analyze the output representation for clinical outcome).
However, Rose and Wullschleger may not explicitly disclose the following which is taught by Frayling: wherein using zero-shot prompting comprises inputting, to the causal language model, an individual's event request history for an observation period (Page Two, Col. 2, Prompting Strategies input into the model the patient history and determine the output).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of zero-shot prompting from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to generate more accurate and contextually relevant results (Frayling Page One Abstract).
As per Claim 18, Rose and Wullschleger discloses the method of Claim 2. However, Rose and Wullschleger may not explicitly disclose the following which is taught by Frayling: generating a synthetic dataset based on fine-tuning the trained causal language model on an evaluation dataset (Page One, Col. 2, generating synthetic clinical EHR data based on a fine-tuning learning strategy).
Therefore, it would have been obvious to a person of ordinary skill in the art prior to the filing of the present invention to combine the known concept of zero-shot prompting from Oren with the method of predicting an outcome from a tokenized sequence of medical codes from Rose and Wullschleger in order to generate more accurate and contextually relevant results (Frayling Page One Abstract).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Gonzalez-Hernandez (US 2025/0053790 A1) teaches large language models to receive medical text, analyze the text to make predictions using the LLM model.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Evangeline Barr whose telephone number is (571)272-0369. The examiner can normally be reached Monday to Friday 8:00 am to 4:00 pm.
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, Fonya Long can be reached at 571-270-5096. 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.
/EVANGELINE BARR/Primary Examiner, Art Unit 3682
