App. No. 19/357,300

Strategy check: re-ranked from #2 — The only banked argument tagged to claim 2 is the Recentive/practical-application argument (rank 4), stress-rated fragile because Recentive is adverse and added data-content is not a technical operation; the inherited Prong 1 mental-process theory (rank 1) is stronger and should be extended to lead this dependent claim.

the data of an electronic medical record of a virtual patient is generated by using a generative model that generates data regarding treatment of the virtual patient from data regarding treatment of the patient (claim 9: the generative model is a large-scale language model)

The examiner's Step 2A Prong 1 conclusion depends on characterizing the claims as covering steps that 'may be practically performed in the human mind using observation, evaluation, judgment, and opinion' but for generic computer components (office action, Step 2A Prong 1). For counsel to weigh: the independent claims affirmatively require generation of the virtual-patient EMR 'by using a generative model,' and claim 9 narrows that generative model to 'a large-scale language model' — a limitation the examiner's mental-process framing does not engage. Under MPEP § 2106.04(a)(2)(III) a limitation falls in the mental-processes grouping only if it can PRACTICALLY be performed in the human mind; running a trained generative model / LLM to synthesize an electronic medical record is not an act a person can practically perform mentally. The examiner's assertion that this activity 'predating computers' can be done mentally does not address the specific generative-model step that the claim requires as a positive, non-severable limitation.

• —Office action, Step 2A Prong 1: 'covers performance of the limitation in the mind ... which may be practically performed in the human mind using observation, evaluation, judgment, and opinion (MPEP 2106.04(a)(2), subsection III), but for the recitation of generic computer components.'
• —Claim 1: 'the data of an electronic medical record of a virtual patient is generated by using a generative model that generates data regarding treatment of the virtual patient from data regarding treatment of the patient'
• —Claim 9: 'the generative model is a large-scale language model.'
• —Office action: 'applying machine learning to predicting disease progression, an activity predating computers, did not transform the abstract idea into a patent-eligible invention.'

Risk The examiner will likely respond that the generative model is recited only at a high level of generality and is treated as a generic tool applying the abstract idea (as stated in the Step 2A Prong 2 discussion). PHE caution: framing the invention as turning on the generative model / LLM step may narrow the claim scope in the file wrapper to embodiments that use such a model, which could limit later doctrine-of-equivalents reach.

generate time-series data regarding a medical condition of the virtual patient from the generated data regarding treatment of the virtual patient

The examiner alternatively places the claims in the 'mathematical concepts' grouping, stating that 'applying mathematical algorithms to produce synthetic data is mathematical concepts' and invoking Gottschalk v. Benson and Digitech (office action, Step 2A Prong 1). For counsel to weigh: under MPEP § 2106.04(a)(2)(I), a claim recites a mathematical concept only when it sets forth a mathematical relationship, formula/equation, or calculation in the claim itself; a step that merely can be implemented using math is not, by itself, a recited mathematical concept. The pending claims recite acquiring an EMR, generating a virtual-patient EMR via a generative model, generating time-series data, and outputting it — none of the claim limitations recites a mathematical formula, equation, or explicit calculation. Counsel may press that the mathematical-concepts characterization is unsupported by the claim language and that the examiner has described the invention 'at another level of abstraction' rather than by what the claims actually recite (cf. the office action's own Apple v. Ameranth discussion).

• —Office action, Step 2A Prong 1: 'applying mathematical algorithms to produce synthetic data is mathematical concepts.'
• —Claim 1 limitations recite 'acquire,' 'generate data of an electronic medical record,' 'generate time-series data,' and 'output' — no equation or formula appears in the claim text.
• —Office action: 'the claimed abstract idea could be described at different levels of abstraction' (quoting Apple v. Ameranth).

Risk The examiner will likely respond that the generative model inherently performs mathematical operations and that the mental-processes grouping independently sustains the rejection even if the mathematical-concepts label is dropped. PHE caution: none significant, but this argument does not by itself remove the mental-processes basis and should be paired with the Prong 1 mental-process argument.

generate data of an electronic medical record of a virtual patient ... by using a generative model

At Step 2B the examiner concludes the additional elements are 'well-understood, routine, conventional activity,' resting on MPEP 2106.05(d)(II) categories (receiving/transmitting data over a network, repetitive calculations, electronic recordkeeping, storing/retrieving information) and on the assertion that the invention 'utilizes conventional communication networks, generic processors and memory' (office action, Step 2B). For counsel to weigh: under Berkheimer v. HP and MPEP § 2106.05(d), a factual finding that a claim element is well-understood, routine, and conventional must be supported by one of the four evidentiary showings (a citation, a court holding, a publication, or an applicant admission). The office action's § 2106.05(d)(II) list addresses generic data receiving/storing/outputting but does not supply record evidence that using a generative model to synthesize an entire virtual-patient EMR and derive time-series 'patient journey' data was well-understood, routine, and conventional — the examiner appears to rely on the specification's general 'programmable processors executing ... computer programs' language, which speaks to the processor, not to the generative-modeling step. Counsel may press this as a Berkheimer evidentiary gap directed to the specific ordered combination as a whole.

• —Office action, Step 2B: 'a conclusion that the recited steps are well-understood, routine, conventional activity is supported under Berkheimer Option 2.'
• —Office action, Step 2B: 'the invention utilizes conventional communication networks, generic processors and memory, which can be found in mobile devices or desktop computers.'
• —Office action, Step 2B quotation of spec: 'the processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs.'
• —Office action, Step 2B: enumerated MPEP 2106.05(d)(II) categories (i–vi) addressing receiving/transmitting, repetitive calculations, electronic recordkeeping, storing/retrieving.

Risk The examiner will likely respond that the § 2106.05(d)(II) categories and the specification's own 'programmable processors' language satisfy Berkheimer Option 2 (specification admission), and that Step 2B need only address the additional elements beyond the abstract idea. PHE caution: arguing the generative-modeling step is unconventional is consistent with, and reinforces, later non-obviousness positions but should be kept consistent across the file wrapper.

generate time-series data regarding a medical condition of the virtual patient from the generated data regarding treatment of the virtual patient; and output the generated time-series data (claim 8: the time-series data is a patient journey)

The examiner's Step 2A Prong 2 analysis rests on the analogy to Recentive Analytics v. Fox Corp., stating the claims recite 'conventional machine learning models without specific improvements to the technology itself' and 'do not articulate how a technological improvement is achieved' (office action, Step 2A Prong 2). For counsel to weigh: whether the ordered combination — deriving a synthetic virtual-patient EMR covering 'each stage of a medical condition' and converting it into output time-series 'patient journey' data (claim 8) — reflects a specific technological application (for example, generating synthetic/privacy-preserving training or simulation data) rather than merely 'applying' an abstract idea. Counsel should confirm against the as-filed specification whether it describes a concrete technical problem (e.g., scarcity or privacy constraints of real EMR data) and a corresponding improvement, because a well-supported practical-application showing under MPEP § 2106.04(d)/(a) ends the analysis at Step 2A Prong 2. Absent such specification support, this lever is weaker than the Prong 1 and Berkheimer arguments above.

• —Office action, Step 2A Prong 2: 'Similar to Recentive Analytics, claim 11 recites conventional machine learning models without specific improvements to the technology itself.' and 'Claim 11 does not articulate how a technological improvement is achieved.'
• —Claim 1: 'the electronic medical record of the virtual patient includes data regarding treatment in each stage of a medical condition of the virtual patient.'
• —Claim 8: 'time-series data regarding a medical condition of the virtual patient is a patient journey.'

Risk The examiner will likely reiterate Recentive Analytics and Electric Power Group, responding that the claims merely produce additional data ('everything remains in the form of a code stored in the computer memory') without improving computer functionality. PHE caution: characterizing the improvement (e.g., synthetic-data generation for a stated technical problem) narrows the claimed purpose in the file wrapper; keep any such characterization tethered to the specification and consistent with positions taken elsewhere.