DETAILED ACTION
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 .
Status of Claims
This Final Office Action is in response to the Amendment and Remarks filed 07/24/2026. Claims 1-20 are currently pending and considered herein.
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.
Claim 1 recites, wherein the abstract elements are not emboldened:
A computer-implemented method for generating a patient-specific clinical assessment, the method comprising: receiving, by a processor, health data corresponding to a patient, wherein the health data comprises one or more of: biometric data, historical medical data, and patient-reported symptoms; analyzing, by the processor, the health data using a trained large language model (LLM), wherein the analyzing comprises contextually processing the health data based at least in part on the historical medical data corresponding to the patient, to generate a set of insights comprising potential diagnoses, treatment recommendations, and prioritized patient conditions; generating, by the processor, a set of interactive prompts for a patient interface, wherein the set of interactive prompts is generated based on the set of insights and configured to obtain additional information associated with at least one of the potential diagnoses, treatment recommendations, and prioritized patient conditions; dynamically adapting, by the processor, the set of interactive prompts based on a set of patient responses using a decision-tree algorithm implemented by the trained LLM, wherein the dynamically adapting comprises modifying subsequent prompts based on analysis of previous patient responses to obtain context-specific clinical information; receiving, by the processor, the set of patient responses responsive to the dynamically adapted set of interactive prompts; updating, by the processor, the set of insights by processing the set of patient responses using the trained LLM, wherein the updating comprises modifying at least one of the potential diagnoses, treatment recommendations, or prioritized patient conditions based on the set of patient responses, wherein the updating is based on the context-specific clinical information obtained using the decision-tree algorithm, to generate an enriched patient clinical profile that reflects iterative refinement of the insights through the decision-tree algorithm; evaluating, by the processor, at least one of the updated insights in the enriched patient clinical profile against a pre-configured clinical threshold; in response to the at least one updated clinical insight satisfying the pre-configured clinical threshold: automatically triggering a clinical protocol; re-prioritizing the prioritized patient conditions based on the updated insights and the triggered clinical protocol; routing the enriched patient clinical profile to a designated care team member based on the re-prioritized patient conditions to generate a routed care team assignment; re-generating the enriched patient clinical profile to reflect the triggered clinical protocol, the re-prioritized patient conditions, and the routed care team assignment; and causing display of, by the processor, the re-generated enriched patient clinical profile on a clinician interface by transmitting the re-generated enriched patient clinical profile to the clinician interface, the re-generated enriched patient clinical profile comprising the updated insights and the set of patient responses.
Independent claims 10 and 17 recite substantially similar limitations. The claimed invention is directed to the abstract idea of collecting patient and health information and responses, analyzing the information, and generating updated clinical insights based on the analyses.
The limitations of “receiving health data corresponding to a patient, wherein the health data comprises one or more of: biometric data, historical medical data, and patient-reported symptoms; analyzing the health data, wherein the analyzing comprises contextually processing the health data based at least in part on the historical medical data corresponding to the patient, to generate a set of clinical insights comprising potential diagnoses, treatment recommendations, and prioritized patient conditions; generating a set of interactive prompts for a patient interface, wherein the set of interactive prompts is generated based on the set of clinical insights and configured to obtain additional information associated with at least one of the potential diagnoses, treatment recommendations, and prioritized patient conditions; dynamically adapting the set of interactive prompts based on a set of patient responses, wherein the dynamically adapting comprises modifying subsequent prompts based on analysis of previous patient responses to obtain context-specific clinical information; receiving the set of patient responses responsive to the dynamically adapted set of interactive prompts; updating the set of clinical insights by processing the set of patient responses, wherein the updating comprises modifying at least one of the potential diagnoses, treatment recommendations, or prioritized patient conditions based on the set of patient responses, wherein the updating is based on the context-specific clinical information obtained, to generate an enriched patient clinical profile that reflects iterative refinement of the insights; evaluating at least one of the updated insights in the enriched patient clinical profile against a pre-configured clinical threshold; in response to the at least one updated clinical insight satisfying the pre-configured clinical threshold: automatically triggering a clinical protocol; re-prioritizing the prioritized patient conditions based on the updated insights and the triggered clinical protocol; routing the enriched patient clinical profile to a designated care team member based on the re-prioritized patient conditions to generate a routed care team assignment; re-generating the enriched patient clinical profile to reflect the triggered clinical protocol, the re-prioritized patient conditions, and the routed care team assignment; the re-generated enriched patient clinical profile comprising the updated insights and the set of patient responses,” as drafted is a process that, under its broadest reasonable interpretation, is an abstract idea that covers performance of the limitation as organizing human activity. For example, but for the generic computer system reciting a processor, trained LLM, decision-tree algorithm and display interface (and generic memory of claim 10 and non-transitory computer-readable storage medium and executable program instructions of claim 17), the claim recites an abstract idea that covers performance of the limitation as organizing human activity including following rules or instructions. The claim recites as a whole a method of organizing human activity because the limitations include a method that allows users to access myriad patient data and responses, analyze those data and determine whether certain conditions are met and recommendations made based on the analyses. This is a method of managing interactions between people. The nominal recitation of a generic computer processor and LLM with decision tree algorithm and a display interface does not take the claims out of the method of organizing human interactions grouping. The additional limitations amount to computer methods for further implementing the abstract idea of organizing human activity. Thus, the claims recite an abstract idea.
The claims also recited an abstract idea including mental processes. But for the generic computer components and LLM decision tree algorithm and interface display, nothing in the claims is precluded from being performed in the mind. For example, a caregiver can collect the patient health data and responses and analyze those data and then formulate clinical insights and appropriate recommendations/treatments. Thus, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of the generic computer components, LLM decision tree algorithm and interface display. The computer and LLM and decision tree algorithm and interface are recited at a high-level of generality (i.e., as a generic processor/server/storage/display performing a generic computer function of receiving inputs, analyzing the inputs, and displaying selected information) such that it amounts no more than mere instructions to apply the exception using a generic computer component. Further, the iterative process of the LLM and decision tree are not the focus of the claim limitations (there is no underlying technological improvement to these computer components and LLM found in the claims or Specification), but rather are leveraged as tools to employ the judicial exception. Accordingly, these additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The limitations seem to monopolize the abstract idea of patient analysis and diagnoses and general techniques between a physician and her patient. Furthermore, there is no clear improvement to the underlying computer technology in the claim. The claim is thus directed to an abstract idea.
The claims do 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 the generic computer components, LLM decision tree algorithm and interface display amounts to no more than mere instructions to apply the exception using a computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The additional elements when considered separately and as an ordered combination do not amount to add significantly more as these limitations provide nothing more than to simply apply the exception in a generic computing environment.
The dependent claims do not remedy the deficiencies of the independent claims with respect to patent eligible subject matter. The dependent claims further limit the abstract idea. Claims 2, 11 and 18 describe using natural language processing and receiving inputs from wearable devices, which are recited at a high level of generality such that it amounts to no more than mere instructions to apply the judicial exception using a generic computer component and cannot provide an inventive concept. Even in combination, the NLP and wearable device do not integrate the abstract idea into a practical application and does not amount to significantly more than the abstract idea itself. Claims 3, 12 and 19 detail a mapping of data and further limit the abstract idea. Claims 4, 13 and 20 describe update an LLM and synchronize data, which further limits the abstract idea. Claims 5 and 14 detail contextually processing data and a wearable device, which are recited at a high level of generality such that it amounts to no more than mere instructions to apply the judicial exception using a generic computer component and cannot provide an inventive concept. Even in combination, the contextual processing and wearable device do not integrate the abstract idea into a practical application and does not amount to significantly more than the abstract idea itself. Claims 6 and 15 detail access controls and decryption permission and limits the abstract idea. Claims 7 and 16 describe augmenting the LLM using a RAG technique and a database, which are recited at a high level of generality such that it amounts to no more than mere instructions to apply the judicial exception using a generic computer component and cannot provide an inventive concept. Even in combination, the augmenting and database do not integrate the abstract idea into a practical application and does not amount to significantly more than the abstract idea itself. Claim 8 recites an encryption protocol, which is recited at a high level of generality such that it amounts to no more than mere instructions to apply the judicial exception using a generic computer component and cannot provide an inventive concept. Even in combination, the encryption protocol does not integrate the abstract idea into a practical application and does not amount to significantly more than the abstract idea itself. Claim 9 describes generating a notification when an insight is beyond a threshold and further limits the abstract idea. Therefore, the claims 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 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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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 1-6, 8-15 and 17-20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. 2021/0050089 A1 to Mohammad et al., hereinafter “Mohammad,” in view of U.S. 2018/0001184 A1 to Tran et al., hereinafter “Tran.”
Regarding claim 1, Mohammad discloses A computer-implemented method for generating a patient-specific clinical assessment, the method comprising: receiving, by a processor, health data corresponding to a patient, wherein the health data comprises one or more of: biometric data, historical medical data, and patient-reported symptoms (See Mohammad at least at Abstract; Paras. [0006]-[0008], [0044]-[0045] (patient reported); Figs. 1-14); analyzing, by the processor, the health data using a trained model, wherein the analyzing comprises contextually processing the health data based at least in part on the historical medical data corresponding to the patient, to generate a set of insights comprising potential diagnoses, treatment recommendations, and prioritized patient conditions (See id. at least at Abstract; Paras. [0006]-[0008] (machine learning model trained using patient data), [0044]-[0045], [0059]-[0063], [0075]-[0079], [0119] (training), [0123]-[0133] (clinical insights), [0140]; Figs. 1-14); generating, by the processor, a set of interactive prompts for a patient interface, wherein the set of interactive prompts is generated based on the set of insights and configured to obtain additional information associated with at least one of the potential diagnoses, treatment recommendations, and prioritized patient conditions (See id. at least at Paras. [0034]-[0036] (interactive patient interface), [0053], [0063] (interactive prompts), [0075]-[0079], [0189]-[0194]; Claims 2, 3; Figs. 1-14); updating, by the processor, the set of insights by processing the set of patient responses using the trained LLM, wherein the updating comprises modifying at least one of the potential diagnoses, treatment recommendations, or prioritized patient conditions based on the set of patient responses, to generate an enriched patient clinical profile (See id. at least at Paras. [0006] (establish personal profile for a patient), [0028]-[0031], [0086]-[0088] (updated responses and recommendations, re-generating profile), [0113]-[0122], [0128]-[0129]); and causing display of, by the processor, the re-generated enriched patient clinical profile on a clinician interface […] enriched patient clinical profile comprising the updated insights and the set of patient responses (See id. at least at Paras. [0006]-[0009] (profiles and updating), [0030]-[0034] (updating profiles, i.e., re-generated enriched patient profile and displaying), [0052], [0059], [0063]-[0064]; Figs. 1-4).
Mohammad may not specifically describe but Tran teaches wherein the updating is based on the context-specific clinical information obtained using the decision-tree algorithm, to generate an enriched patient clinical profile that reflects iterative refinement of the insights through the decision-tree algorithm (See Tran at least at Paras. [0220]-[0222] (updating probabilities and beliefs defined based on clinical relevance), [0297] (algorithms updated), [0300] (“The patient profile stored on the database may thus continually update with each added metric and generate a notification.”), [0393], [0398]-[0413] (“The prediction engine effectively operates as an analysis database in which data mining predictions are performed. In operation, the received or “target” patient-procedure data is processed through each of the models, and given a score. A comparator module then analyzes each score from the three models to determine if the predictions are in agreement. For instance, the comparator module can be programmed or otherwise configured to determine if each score is above a predetermined threshold (after normalization of scores) […] the machine learning system can be a supervised or unsupervised learning method such as support vector machine, random forest, nearest neighbor analysis, linear regression, binary decision tree, discriminant analyses, logistic classifier and cluster analysis, among others.”); Fig. 15); evaluating, by the processor, at least one of the updated insights in the enriched patient clinical profile against a pre-configured clinical threshold (See id. at least at Paras. [0398]-[0413] (“The data is provided to a learning module to create medical models for each patient/subject. A classifier is generated by the learning module, and the classifier can receive new EHR and fitness data and generate output for a prediction engine. The prediction engine in turn provides patient health predictions based on data mining, classification and prediction […] The prediction engine effectively operates as an analysis database in which data mining predictions are performed. In operation, the received or “target” patient-procedure data is processed through each of the models, and given a score. A comparator module then analyzes each score from the three models to determine if the predictions are in agreement. For instance, the comparator module can be programmed or otherwise configured to determine if each score is above a predetermined threshold (after normalization of scores) […] When the patient visits the doctor, the progress or failure in the treatment can be reviewed, and the system can make recommendations on ways to improve the treatment based on the totality of data received for the patient and from a population of similar patients.”); Fig. 15); in response to the at least one updated clinical insight satisfying the pre-configured clinical threshold: automatically triggering a clinical protocol; re-prioritizing the prioritized patient conditions based on the updated insights and the triggered clinical protocol (See id. at least at Paras. [0398]-[0413] (“[T]he comparator module can be programmed or otherwise configured to determine if each score is above a predetermined threshold (after normalization of scores). If two or three of the three predictions agree based on this determination, then a corresponding set of procedure parameters (e.g., procedure location, procedure equipment, and procedure personnel/technician) from the matched models are provided to the scheduling application, and the patient is predictively scheduled for that procedure.”); Figs. 15-16); routing the enriched patient clinical profile to a designated care team member based on the re-prioritized patient conditions to generate a routed care team assignment (See id. at least at Paras. [0393], [0398]-[0413] (“If two or three of the three predictions agree based on this determination, then a corresponding set of procedure parameters […] the patient is predictively scheduled for that procedure. Otherwise, if the predictions disagree, then not prediction is made. In such a case, a manual scheduling procedure can be used.”); re-generating the enriched patient clinical profile to reflect the triggered clinical protocol, the re-prioritized patient conditions, and the routed care team assignment (See id. at least at Paras. [0220]-[0222] (updating probabilities and beliefs defined based on clinical relevance), [0297] (algorithms updated), [0300] (“The patient profile stored on the database may thus continually update with each added metric and generate a notification.”), [0393], [0398]-[0413] (“[M]ine the clinical data repository and health data repository for patients sharing similarity with the subject, including one or more similar biomarkers associated with health conditions [0411] l. identify at least one similar health conditions and identifying one or more corrective actions recorded in the repository and the result of each action for the one or more health conditions; [0412] m. present the corrective action and result to the subject and recommending an action to reduce risk from the predicted health condition [0413] n. monitor the health condition using updates in the clinical data repository and health data repository.”); Claim 2 (“Presenting the corrective action and result to a doctor and recommending an action to reduce risk from the predicted health condition, and monitoring the health condition using updates in the clinical data repository and health data repository.”); Fig. 15); and causing display of, by the processor, the re-generated enriched patient clinical profile on a clinician interface by transmitting the re-generated enriched patient clinical profile to the clinician interface, the re-generated enriched patient clinical profile comprising the updated insights (See id. at least at Paras. [0003], [0392]-[0393] (“[T]he web interface 202 can be configured to provide social functions between users. In some embodiments, the web interface can limit sharing to practice groups, within treatment facilities, or within medical institutions (e.g., hospitals). According to one aspect, sharing of test results and associated genomic information on patients can create a strong community of physicians, and foster discussion about treatment or even specific patients.”), [0398]-[0415]; Claim 2; Figs. 9, 15).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad to incorporate the teachings of Tran and provide a trained machine learning model and dynamically adapting clinical insights and clinical protocols. Tran is directed to a smart device that includes a decision support system for a patient and treatment. Incorporating the dynamically adapting clinical insights and clinical protocols of Tran with the machine learning and precision treatment platform and techniques of Mohammad would thereby increase the applicability, utility, and efficacy of an AI model for clinical recommendations.
Regarding claim 2, Mohammad as modified by Tran discloses the limitations of claim 1 and Mohammad further discloses processing, by the processor, the health data by normalizing the health data using a natural language processing technique and a data standardization technique (See Mohammad at least at Paras. [0036], [0150], [0165]; Claim 3).
While Tran further teaches augmenting, by the processor, the normalized data with metadata corresponding to the health data, wherein the metadata is received from one or more input sources comprising: electronic health records, wearable devices, or patient self-reports, and wherein the metadata comprises temporal attributes, locational attributes, and categorical attributes corresponding to the health data (See Tran at least at Paras. [0003], [0275] (“[S]ituation-aware metadata description framework meaning that context specific information for all possible contexts specified in the meta data during system design. This process may be done for every cause, thereby specifying what remedial actions are more appropriate for it and for what context.”), [0393], [0398]-[0400]; Claims 2, 19; Fig. 15).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad to incorporate the teachings of Tran and provide incorporating metadata into a patient profile. Tran is directed to a smart device that includes a decision support system for a patient and treatment. Incorporating the dynamically adapting clinical insights and clinical protocols of Tran with the machine learning and precision treatment platform and techniques of Mohammad would thereby improve the recommendations for a clinical support system.
Regarding claim 3, Mohammad as modified by Tran discloses the limitations of claim 2 and Mohammad further discloses wherein the processing of the received health data further comprises: mapping, by the processor, the normalized data and the augmented data to a standardized medical coding system (See Mohammad at least at Paras. [0048], [0121]-[0122], [0145]-[0147]; Figs. 1-5).
Regarding claim 4, Mohammad as modified by Tran discloses the limitations of claim 1 and Mohammad further discloses dynamically updating, by the processor, the trained LLM with the health data to improve diagnostic accuracy and treatment recommendations over time; and synchronizing, by the processor, the set of clinical insights, and clinician-reviewed data with an electronic medical record (EMR) system (See id. at least at Paras. [0030]-[0034], [0044], [0048]-[0050] (model updated over time), [0063]-[0064], [0067], [0114], [0128], [0133], [0168]-[0169] (synchronizing medical data); Claim 9; Figs. 1-5).
Regarding claim 5, Mohammad as modified by Tran discloses the limitations of claim 1 and Mohammad further discloses wherein the analysis further comprises: contextually processing the health data to personalize the set of clinical insights using the trained LLM, the contextual processing being based at least in part on historical health data and real-time updates from at least one wearable device associated with the patient (See id. at least at Paras. [0006]-[0008], [0030]-[0034], [0168]-[0172]; Claim 3; Figs. 1-14).
Regarding claim 6, Mohammad as modified by Tran discloses the limitations of claim 1 and Tran further teaches enforcing role-based access controls for decryption permissions (See Tran at least at Paras. [0393] (permission information), [0398]-[0413] (“If two or three of the three predictions agree based on this determination, then a corresponding set of procedure parameters […] the patient is predictively scheduled for that procedure. Otherwise, if the predictions disagree, then not prediction is made. In such a case, a manual scheduling procedure can be used.”), [0561]-[0562] (“[T]he communications between the base station 20 and the server 200 are encrypted to protect patient identifiable information and other private details of the person …] Upon receiving the request from the doctor through the server 200, the base station 20 “tailors” the request to suit the proper device capability before relaying it the appropriate device.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad to incorporate the teachings of Tran and provide permissions for access and encrypting data. Tran is directed to a smart device that includes a decision support system for a patient and treatment. Incorporating the dynamically adapting clinical insights and clinical protocols of Tran with the machine learning and precision treatment platform and techniques of Mohammad would thereby improve the privacy and protection of data for a clinical recommendation system.
Regarding claim 9, Mohammad as modified by Tran disclose the limitations of claim 1 and Tran further teaches evaluating, by the processor, at least one of the updated insights against a pre-configured clinical threshold; and in response to the at least one updated clinical insight satisfying the pre-configured clinical threshold, and automatically generating a clinical notification directed to a designated care team member identified based on the triggered clinical protocol or a patient response indicating an urgent clinical condition (See Tran at least at Paras. [0070], [0081], [0282] (identified patterns notifications to doctor), [0289] (outside threshold send alerts to doctor), [0398]-[0413] (“The data is provided to a learning module to create medical models for each patient/subject. A classifier is generated by the learning module, and the classifier can receive new EHR and fitness data and generate output for a prediction engine. The prediction engine in turn provides patient health predictions based on data mining, classification and prediction […] The prediction engine effectively operates as an analysis database in which data mining predictions are performed. In operation, the received or “target” patient-procedure data is processed through each of the models, and given a score. A comparator module then analyzes each score from the three models to determine if the predictions are in agreement. For instance, the comparator module can be programmed or otherwise configured to determine if each score is above a predetermined threshold (after normalization of scores) […] When the patient visits the doctor, the progress or failure in the treatment can be reviewed, and the system can make recommendations on ways to improve the treatment based on the totality of data received for the patient and from a population of similar patients.”), [0441] (“[D]epending on selected rules triggered by the patient response, the system sends notifications to third-party devices such as devices owned by family members or caregivers. The system can also send notifications to doctors, doctor's staff, or other authorized service providers who then send in response results that are automatically processed by the system to alter the behavior of some rules.”), [0744]-[0759] (i.e., Claim 19); Figs. 1-6, 15).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad to incorporate the teachings of Tran provide updating clinical insights based on a threshold and generating a notification in response. Tran is directed to a smart device that includes a decision support system for a patient and treatment. Incorporating the dynamically adapting clinical insights and clinical protocols of Tran with the machine learning and precision treatment platform and techniques of Mohammad would thereby increase the applicability, utility, and efficacy of a system for personalized clinical assessment and care management.
Regarding claims 10 and 17, claims 10 and 17 recite substantially the same limitations as included in independent claim 1. Thus, the claims are rejected under the same grounds of rejection and for the same reasoning applied to claim 1, above.
Regarding claims 11-13 and 18-20, claims 11-13 and 18-20 recite substantially the same limitations as included in claims 2-4, respectively. Thus, the claims are rejected under the same grounds of rejection and for the same reasoning applied to claims 2-4, above.
Regarding claims 14 and 15, claims 14 and 15 recite substantially the same limitations as included in claims 5 and 6, respectively. Thus, the claims are rejected under the same grounds of rejection and for the same reasoning applied to claims 5 and 6, above.
Claims 7 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Mohammad, in view of Tran and further in view of U.S. 2024/0404687 A1 to Bell et al., hereinafter “Bell.”
Regarding claim 7, Mohammad as modified by Tran disclose the limitations of claim 1. The references may not specifically describe but Bell teaches augmenting, by the processor, the trained LLM using a retrieval-augmented generation (RAG) technique, wherein the RAG technique retrieves relevant data from a healthcare knowledge database comprising psychiatric research, historical patient data, treatment outcomes, and medical literature (See Bell at least at Paras. [0003], [0060] (medical research, EHR), [0066] (research databases), [0120] (“agent module types include […] data collection agent modules (e.g., RAG modules).”), [0152] (“[D]ata from the knowledge database (e.g., a question-and-answer dataset) is used to finetune the model component (e.g., the model API (e.g., for a model 228), a task-specific machine-learning model.”), [0161] (“[I]mplementing a retrieval-augmented generative (RAG) approach to identify relevant portions of EHR text.”), [0179], [0315], [0317]-[0321], [0335]; Figs. 2, 4, 6-9, 13, 14, 17, 19).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad and Tran to incorporate the teachings of Bell and provide a RAG technique for LLM. Bell is directed to AI routing for tasks and devices. Incorporating the AI routing and retrieval augmented generation as in Bell with the dynamically adapting clinical insights and clinical protocols of Tran and the machine learning and precision treatment platform and techniques of Mohammad would thereby improve an AI based clinical recommendation system.
Regarding claim 16, claims 16 recites substantially the same limitations as included in claim 7. Thus, claim 16 is rejected under the same grounds of rejection and for the same reasoning applied to claim 7, above.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over U.S. Mohammad, in view of Tran and further in view of U.S. 2024/0395369 A1 to Mariappan et al., hereinafter “Mariappan.”
Regarding claim 8, Mohammad as modified by Tran discloses the limitations of claim 1. The references may not specifically describe but Mariappan teaches encrypting, by the processor, the health data using an end-to-end encryption protocol, wherein the encrypting comprises: encrypting patient responses received from a patient interface prior to processing by a trained large language model (LLM) and processing, by the processor, the health data within a security boundary (See Mariappan at least at Paras. [0038]-[0041] (LLM and encryption for secure communication by patient), [0065] (patient responses), [0071]-[0078] (“[I]t will be understood that robust security measures and compliance with global data protection regulations (e.g., GDPR, HIPAA) to protect patient data are used and enforced. This includes the use of encrypted databases and secure connections for data transmission. Thus, the communications described over both Internet and intranets are secured, and the database where the records are stored are suitably encrypted […] the system employs advanced machine learning algorithms to predict potential urgent care needs […] the system doesn't just follow a rigid script but adapts the questions based on the patient's initial inputs and responses.”); Figs. 1-5, 7).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad and Tran to incorporate the teachings of Mariappan and provide encrypted communications and databases. Mariappan is directed to automating medical history documentation using AI. Incorporating the encryption and medical documentation as in Mariappan with the dynamically adapting clinical insights and clinical protocols of Tran and the machine learning and precision treatment platform and techniques of Mohammad would thereby increase the applicability, utility, and efficacy of an AI system for personalized clinical assessment and care management.
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over U.S. Mohammad, in view of Tran and further in view of U.S. 2017/0188976 A1 to Kalra et al., hereinafter “Kalra.”
Regarding claim 21, Mohammad as modified by Tran discloses the limitations of claim 1. The references may not specifically describe but Kalra teaches wherein the decision-tree algorithm defines branching conditions based on the analysis of the previous patient responses (See Kalra at least at Abstract; Paras. [0031] (“[T]he system can serve to the patient a dynamic psychological test in which each subsequent question served to the patient is selected from a question tree based on a response provided by the patient in response to a previous question. The system can thus navigate through a sequence of increasingly pointed questions to identify the presence and/or severity of a variety of psychological conditions that may be affecting the patient.”), [0040] (decision tree), [0095] (machine learning), [0099] (reordering nodes in decision tree), [0102]-[0106] (“[T]he system models a meditation program in the form of a decision tree including a set of nodes specifying a sequence of meditation tracks and related symptom-, meditation score-, and/or patient feedback-based triggers for navigating through these nodes.” Markov chains where the probability of each event depends on the state attained in the previous event.); Figs. 4-6).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the disclosure of Mohammad and Tran to incorporate the teachings of Kalra and provide a decision tree that is updated based on patient data and responses to queries. Kalra is directed to addressing psychological conditions of a patient and custom decision trees. Incorporating the decision trees based on patient responses of Kalra with the dynamically adapting clinical insights and clinical protocols of Tran and the machine learning and precision treatment platform and techniques of Mohammad would thereby increase the applicability, utility, and efficacy of an AI system for clinical recommendations.
Response to Arguments
Applicant’s remarks filed July 24, 2026 have been fully considered, but they are not entirely persuasive. The following explains why:
Applicant’s arguments pertaining to subject matter eligibility are not persuasive. The claims have been addressed with regard to the updated 35 U.S.C. §101 rejection discussed above, and considered under relevant sections of the MPEP. The arguments at pages 14-27 of Applicant’s Remarks are not persuasive. At pages 14-18 the Examiner disagrees that there is not an abstract idea or that there is any practical application thereof, or a technological improvement to the underlying computer technology in the claims. The claimed decision tree LLM is merely leveraged as a tool to employ the abstract idea and are merely instructions. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Applicant argues at Pages 15-16 the specific computational pipeline and not a method of organizing human activity. However, the current claims are included as a mental process(es) for the judicial exception as well as organizing human activity as following rules or instructions. That it may be tedious or laborious to perform analyses in the mind or manually is not of consequence in the eligibility analysis. The Examiner disagrees at Page 17 “the "decision-tree algorithm" producing the "enriched patient clinical profile" is a specific improvement to computer functionality” as argued by Applicant. This is not found in the claim limitations. There is no improvement to underlying machine learning models with this clinical support system. There is no clear technological improvement that is described in the claims for the underlying computer technology.
The Examiner disagrees the limitations are like Example 47 from the PEG at Pages 18-19. The LLM and functions are recited at a high level, and amount to applying the exception using a generic computer component. In Example 47, for claim 2, the “detecting” and “analyzing” were mental processes, and “using the trained ANN” (similar to software executing on a computer for analyzing and comparing healthcare decision trees and detecting variations and same as the claimed LLM and decision tree, and amounted to generic computer implementation). Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Thus, the claims recite an abstract idea. The Examiner disagrees at Pages 19-22, the automated “closed loop pipeline” system for generating and re-generating patient clinical profiles is part of the abstract idea. It is not analogous to Example 47, as providing clinical recommendations is not the same as dropping identified packets and blocking future traffic from a source address. The Examiner disagrees at Pages 20-22 regarding Example 49 - there is not a specific medical treatment being implemented at Claim 9 as there is no specific treatment being called for and actually implemented in the claims. The Examiner disagrees at Pages 22-24 with the clinical profile being made by a computational pipeline as a practical application or significantly more than the abstract idea. It is leveraging computers as a tool to employ the abstract idea of obtaining patient data and providing recommendations after analyzing and comparing it to other data, in any number of iterations. The Examiner disagrees at Pages 23-24 that the limitations are akin to Example 42 – there is no conversion of standardized and non-standardized data between disparate systems that otherwise could not communicate found in the current limitations, and the mention of LLM decision trees is not sufficient as it is merely a tool to employ the abstract idea. The Examiner disagrees at Pages 25-28 there is an inventive concept or that the additional elements have not been shown to be conventional computer devices leveraged as tools to employ the abstract idea. The LLM and decision tree logic is how a LLM functions and there is no claim to improving the underlying LLM or decision tree logic. These are commonly used today and here in the limitations merely as tools to apply the judicial exception. The Examiner disagrees at Page 27 that “(evaluating the enriched patient clinical profile against a pre-configured clinical threshold and automatically triggering a clinical protocol comprising re-prioritization, routing, and profile re- generation) is absent from every reference of record, confirming that this specific element of the combination was not well-understood, routine, or conventional as of the effective filing date.” These limitations are part of the abstract idea itself and not additional elements. The additional elements are the computer components and generic LLM and logic tree and display interface – they are leveraged to employ the abstract idea of re-generating enriched profiles and the like. The Examiner further disagrees at Pages 26-28 that the iterative refinement here is improving the machine learning model merely by outputting a useful clinical insight or that any technical improvement to LLMs using iterative refinement is in the claims or specification. For these reasons and those stated above, the claims are not patent eligible.
Applicant’s arguments pertaining to prior art rejections are not persuasive. The amended claims have been addressed with regard to the 35 U.S.C. §103 rejection discussed above. The arguments pertaining to prior art references of the Applicant’s Remarks at Pages 28-29 are not persuasive and moot in light of at least new citations to references Tran, Mariappan and Kalra. Thus, the cited prior art reads on the claim limitations and are in the same field of endeavor, namely, clinical support and decision recommendation and related health determinations for a patient. Therefore, the claims are rejected.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure: U.S. 2025/0095642 A1 to Ghose et al. at Paras. [0018]-[0020] (NLP model for answering queries posed as questions), [0035], [0045]-[0046], [0057]-[0059] (decision tree), [0102] (“[V]erifying the automated response outputted by the LLM further comprises comparing a recommended treatment of the automated response with a previous recommended treatment of a previous patient stored in a patient decision workflow of the previous patient.”) [0103]-[0104] (patient data, training prompts, fine-tuning LLM based on previous patient data and responses/prompts); Claim 7, 12 (“[I]nputting a set of generated natural language training prompts into the fine-tuned LLM a plurality of times to generate a plurality of alternative outputs.” Decision tree); Figs. 2, 5-9).
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to WILLIAM T. MONTICELLO whose telephone number is (313)446-4871. The examiner can normally be reached M-Th; 08:30-18:30 EST.
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.
/WILLIAM T. MONTICELLO/Examiner, Art Unit 3682
/FONYA M LONG/Supervisory Patent Examiner, Art Unit 3682
