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 the Claims
The status of the claims as of the response filed 02/04/2026, is as follows:
· Claims 1-5, 8-10, 12-16, 20, 25, 26, and 45 are pending.
· Claims 6, 7, 11, 17-19, 21-24, and 27-44 are canceled.
· The applicant has amended Claims 1, 26, and 45 are amended and have been considered below.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 02/04/2026 has been entered.
Response to Arguments
35 USC 101 Rejection
Applicant’s arguments, presumed to begin on page 14-25, filed 02/04/2026, with respect to amended Claims 1, 26, and 45 have been fully considered and are not persuasive.
Applicant argues that amended Claim 1 is now directed to a server-client architecture with multiple client terminals, GUIs, dynamically adjusted global list weights, server-side inference, and different scores or rankings returned to different terminals. Applicant asserts that these amendments clarify that the claims are directed to specific technical structure rather than an abstract idea.
The Examiner respectfully disagrees because, under proper BRI, the amended server-client limitations specify the computer environment used to implement the prioritization calculation, but do not change the claim’s operative substance. Claim 1 still weights numerical priority-list scores by global list weights and ranks subjects for treatment or evaluation. The specification confirms that the asserted improvement is customization of prioritization... based on assignment of global list weights, not an improvement to server architecture, GUI technology, network operation, processor operation, memory structure, or machine-learning model operation. Therefore, the amendment does not integrate the judicial exception into a practical application and does not add significantly more. The § 101 rejection is maintained.
Applicant argues that Claim 1 does not recite a mental process because training a respective specific machine learning model, training a combined prioritization machine learning model, presenting ... within a graphical user interface (GUI), transmitting ... to the server, and receiving from the server cannot practically be performed in the human mind or with pen and paper, especially in a server-client network environment.
The Examiner respectfully disagrees that this argument overcomes the § 101 rejection. The rule applied is MPEP § 2106.04(a)(2), because Step 2A, Prong One asks whether the claim recites any judicial exception, including mathematical concepts, mental processes, or certain methods of organizing human activity. MPEP § 2106.04(a)(2) explains that a mental process includes concepts that can be performed in the human mind, or by a human using a pen and paper, such as observations, evaluations, judgments, and opinions. The rule does not provide that AI, GUI, server, or network language removes all abstract subject matter from the claim.
Under proper BRI, the Examiner does not treat the computer-specific acts of training machine-learning models, presenting a GUI, transmitting data, or receiving data as mental steps. Those limitations are additional elements evaluated under Step 2A, Prong Two and Step 2B. However, Claim 1 still recites abstract prioritization logic by generating a respective numerical score for each respective subject, creating lists by ranking subjects according to respective numerical scores, and computing a respective weighted score by multiplying each respective numerical score... by the respective global list weight. The specification confirms the same calculation, stating at paragraph [0154] that the aggregated score is computed by multiplying... each respective numerical score... by the respective global list weight and summing the weighted scores.
Thus, Applicant’s position is not persuasive. The strongest Prong One basis is the mathematical-concept grouping because Claim 1 expressly recites scores, weights, multiplication, weighted scores, and ranking. The claim also includes mental-process subject matter to the extent it evaluates patient information, applies prioritization judgment, compares scores, and ranks subjects for treatment or evaluation. The rejection does not assert that a human mind literally trains the recited machine-learning models or literally transmits data through a network; it identifies the abstract substance as mathematical weighting, scoring, ranking, and patient-prioritization decision-making, then evaluates the AI, GUI, server, network, and processor limitations as additional elements. Therefore, the § 101 rejection is maintained.
Applicant argues that the amended claims do not recite a certain method of organizing human activity because they do not recite a fundamental economic practice, contractual relationship, commercial transaction, or rule governing interpersonal behavior, and because the claims recite technical steps in a network-based client-server environment.
The Examiner respectfully does not find this argument persuasive. The rule applied is MPEP § 2106.04(a)(2), because the issue is whether the claim recites an abstract idea within the enumerated groupings. MPEP § 2106.04(a)(2) identifies certain methods of organizing human activity as including managing personal behavior or relationships or interactions between people, and explains that the determination is based on whether the activity itself falls within a listed subgrouping, not merely on the number of people or the presence of a computer. Here, the claim activity is ordering subjects for healthcare evaluation or treatment using priority lists, weighted scores, and rankings. Claim 1 recites priority lists of subjects scheduled in a prioritized sequence for treatment and evaluation for a target clinical outcome and by a resource-limited medical intervention. The specification confirms that the invention addresses human healthcare prioritization, stating that patients are prioritized for treatment and/or evaluation... by a best manual guess, and/or treated reactively, with emergencies and/or sudden deteriorations being prioritized (par.0003). Thus, the claims organize which patients should be evaluated or treated first in view of clinical outcomes and resource-limited interventions. The use of a server-client implementation does not change the nature of the claimed prioritization activity.
Applicant argues that PowerBlock supports eligibility because the Federal Circuit required consideration of the whole claim, cautioned against oversimplification, and found claims eligible where they recited specific structure. Applicant states that, similarly, the present claims recite a specific GUI and specific hardware structure where a server communicates with multiple client terminals.
The Examiner respectfully does not find this argument persuasive. The rule applied is MPEP § 2106, because claims must be considered as a whole, but concrete components do not automatically make an abstract idea eligible. MPEP § 2106 states that mere recitation of concrete, tangible components is insufficient to confer patent eligibility to an otherwise abstract idea and that an abstract idea does not become nonabstract by limiting the invention to a particular field of use or technological environment, such as the Internet or a computer. The present claims do not recite a comparable mechanical or computer-structural improvement than recited in PowerBlock. They recite generic processors, a server, client terminals, a network, and a GUI used to receive numerical weights, perform inference using those weights, and display numerical scores or rankings. Considering the whole claim does not require treating generic computer implementation of mathematical prioritization as a technological improvement.
Applicant argues that the claims integrate any alleged abstract idea into a practical application because they improve the technical field of machine-learning models and/or computer devices by allowing different users to dynamically adjust preselected global list weights for the same trained machine-learning models, thereby reducing the need for different training datasets or different models and reducing storage, system complexity, and training burden.
The Examiner respectfully does not find this argument persuasive. The rule applied is MPEP § 2106.05(a), because Applicant asserts an improvement to computer or machine-learning technology. MPEP § 2106.05(a) requires more than a bare assertion of improvement; it states that if the specification asserts an improvement, the disclosure must provide sufficient technical details such that a person of ordinary skill would recognize the claimed invention as providing the improvement, and the claim itself must reflect the disclosed technological improvement. The specification passage relied upon by Applicant states that the alleged improvement is providing an approach for customization of prioritization of specific priority lists obtained as outcomes of machine learning models and that the solution is based on assignment of global list weights to the specific priority lists. That disclosure describes improvement to prioritization customization, not a technical improvement to the internal operation of machine-learning models or computers. The claim confirms the same point: it recites known model categories, neural network, logistic regression, decision tree, and boosting, but does not recite a new model architecture, training algorithm, loss function, optimization procedure, feature representation, memory structure, inference protocol, GUI control mechanism, or network protocol. The asserted benefit therefore flows from changing numerical weights used in the prioritization calculation, not from improving machine-learning or computer technology.
Applicant argues that Claim 1 is analogous to Desjardins because the claimed GUI dynamically adjusts preselected global list weights, allegedly like Desjardins’ adjustment of machine-learning parameters to optimize performance on another machine-learning task. Applicant further argues that using dynamically adjustable weights with the same trained machine-learning models reduces the need for different training datasets or additional models, thereby reducing storage, system complexity, and training burden.
The Examiner respectfully does not find this argument persuasive. Because the claim must reflect a technological improvement to the computer, the machine-learning model, or another technology; merely using AI or improving an analytical result is not enough.
Under proper BRI, Claim 1 does not adjust internal machine-learning model parameters to improve learning, transfer learning, model convergence, model accuracy, storage structure, or inference efficiency. Claim 1 adjusts preselected global list weights that indicate the magnitude of prioritization of priority-list outputs, and the server then uses those different weights to return different numerical scores or rankings to different client terminals.
The specification confirms this reading. Paragraph [0075] states that the alleged improvement is customization of prioritization of specific priority lists obtained as outcomes of machine learning models and that the solution is based on assignment of global list weights to the specific priority lists. Paragraph [0154] further explains that the aggregated score is computed by multiplying each numerical score by the respective global list weight and summing the weighted scores.
Thus, the Desjardins analogy is not commensurate with the actual claim language. Because Claim 1 recites user-adjustable weighting values applied to priority-list scores for patient evaluation or treatment ranking. That is an improvement to prioritization customization implemented with generic machine-learning and computer tools, not a claimed technological improvement to machine-learning training or computer functionality. Applicant’s argument is not persuasive, and the § 101 rejection is maintained.
Applicant argues that the claims satisfy Step 2B because the additional elements are meaningful limitations, including GUI elements functionally updated through a machine-learning model, dynamic adjustment of weights by different users, and different values generated for different users, and because those elements go beyond generic technology implementing an abstract idea.
The Examiner respectfully disagrees because under proper BRI, Claim 1 uses the GUI, server, client terminals, network, and machine-learning inference to receive adjusted weighting values, apply those values to priority-list scores, and display resulting scores or rankings. The claim itself recites that the server performs inference using “different values of the preselected global list weights” and returns scores or rankings “computed according to the respective dynamically adjusted preselected global list weights.” The record shows generic computer implementation: paragraph [0098] describes program instructions downloaded over the Internet, a local area network, a wide area network and/or a wireless network; paragraph [0099] describes execution partly on the user’s computer... or entirely on the remote computer or server; and paragraph [0101] describes instructions provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus. Paragraphs [0153]-[0155] show that the GUI changes prioritization values, not computer functionality: global list weights may be adjustable to different values by different users... using a user interface such as a GUI, and the aggregated score is computed by multiplying... each respective numerical score... by the respective global list weight. Thus, Applicant’s position is not persuasive because Step 2B requires additional elements that amount to significantly more than the judicial exception, while these elements perform generic input, transmission, inference, receipt, and display functions for the abstract weighting-and-ranking calculation. Therefore, the § 101 rejection is maintained.
Applicant argues that Classen supports eligibility because the claims allegedly recite meaningful limitations beyond generic technology and are not manifestly abstract.
The Examiner respectfully disagrees because under proper BRI, Claim 1 recites training/providing models, computing weighted scores, ranking subjects, and displaying scores or rankings, while Claims 16 and 20 merely use the priority result for treatment/evaluation order or instructions. The rule applied is MPEP § 2106.05(e), because Applicant relies on “other meaningful limitations”; the additional elements must meaningfully limit the judicial exception rather than merely link the exception to a field of use or add insignificant post-solution activity.
The record shows that the operative claimed result is prioritization, not a particular treatment. Claim 1 recites “ranking subjects according to respective numerical scores” and computing “a respective weighted score by multiplying each respective numerical score... by the respective global list weight.” The specification states that the improvement is customization of prioritization of specific priority lists obtained as outcomes of machine learning models and is based on assignment of global list weights to the specific priority lists at paragraph [0075]. Paragraph [0154] confirms that the score is computed by multiplying... each respective numerical score... by the respective global list weight and summing the weighted scores. Claim 16 recites sequentially treating or evaluating subjects according to the prioritized order, and Claim 20 recites generating instructions according to the target combined priority list, but neither recites a specific treatment, dosage, medical-device operation, biological transformation, or prophylactic intervention.
Thus, Applicant’s Classen analogy is not persuasive because the claims use the abstract prioritization result to guide who should be treated or evaluated first, but do not claim a particular treatment application that integrates the mathematical prioritization exception into a practical application. Therefore, the § 101 rejection is maintained.
Applicant argues that independent Claims 26 and 45 should be allowed for the same reasons asserted for amended Claim 1, because the Office rejected those claims using the same rationale.
The Examiner respectfully disagrees because Applicant presents no separate eligibility argument for Claims 26 and 45 beyond the arguments made for Claim 1. Under proper BRI, Claims 1, 26, and 45 recite the same abstract core: mathematical weighting of priority-list scores, ranking subjects, and using the ranked result for treatment or evaluation priority. The server, client terminal, GUI, network, and machine-learning elements are considered, but they are recited functionally as tools for implementing the scoring and prioritization logic. Because the Claim 1 arguments are not persuasive for the reasons stated above, the rejection of Claims 1-5, 8-10, 12-16, 20, 25, 26, and 45 under 35 U.S.C. § 101 is maintained.
Applicant argues that the dependent claims are eligible because they add weight-adjustment details, aggregation, time correlation, resource scheduling, risk or diagnosis data, use of target EMRs, treatment/evaluation steps, and instruction generation.
The Examiner respectfully disagrees because, under proper BRI, the dependent claims narrow the same abstract prioritization workflow without adding a technological improvement or an inventive concept. Claims 2-5 and 25 further define the global list weights as numerical, location-based, rule-based, range-limited, or user-adjustable values. These limitations narrow the weighting input. Claims 8 and 9 recite aggregation, multiplication, weighted scores, and ranking, which further define the mathematical calculation. Claims 10, 12, and 13 recite time-correlated priority lists and resource-availability scheduling, which organize timing and allocation of medical resources. Claim 14 adds risk or diagnosis data used in the prioritization analysis. Claim 15 recites feeding target EMRs into models and obtaining priority lists, which supplies and processes data for the prioritization workflow. Claims 16 and 20 recite treating/evaluating subjects or generating instructions according to the priority result, but do not recite a particular treatment, dosage, medical-device operation, biological transformation, or technical implementation.
Thus, Applicant’s dependent-claim arguments are not persuasive. The dependent claims refine the abstract scoring, weighting, ranking, scheduling, and treatment-prioritization process, but do not improve a computer, GUI, network, processor, memory, or machine-learning model. The rejection of Claims 2-5, 8-10, 12-16, 20, and 25 under 35 U.S.C. § 101 is maintained.
Applicant argues that Claims 1, 26, and 45 are eligible because the server, client terminals, GUI, network, and machine-learning models provide a practical application and significantly more than the abstract idea.
The Examiner respectfully disagrees because, under MPEP §§ 2111 and 2106, Claims 1, 26, and 45 are properly construed as reciting mathematical weighting and healthcare resource prioritization. The claims weight numerical scores by global list weights, rank subjects according to weighted scores, and return different scores or rankings to different client terminals for treatment or evaluation priority. The server, client terminals, GUI, network, and machine-learning models are considered, but they are recited functionally as tools for implementing that scoring and ranking logic.
The specification confirms that the asserted improvement is prioritization customization, not a technical improvement. Paragraph [0075] describes the improvement as customization of prioritization of specific priority lists obtained as outcomes of machine learning models based on assignment of global list weights to the specific priority lists. Paragraph [0154] explains that the aggregated score is computed by multiplying... each respective numerical score... by the respective global list weight and summing the weighted scores. Therefore, the claims do not integrate the judicial exception into a practical application under Step 2A, Prong Two and do not add significantly more under Step 2B. The rejection of Claims 1-5, 8-10, 12-16, 20, 25, 26, and 45 under 35 U.S.C. § 101 is maintained.
35 USC 103 Rejection
Applicant’s arguments, presumed to begin on page 25-34, filed 02/04/2026, with respect to amended Claims 1, 26, and 45 have been fully considered and are not persuasive.
Applicant argues that Claim 1, “a plurality of client terminals ... presenting ... a graphical user interface (GUI) configured for dynamically adjusting the respective preselected global list weights ... wherein different GUIs dynamically adjust the preselected global list weights to different values ... transmitting the respective dynamically adjusted preselected global list weights ... to the server ... and receiving ... numerical score[s] and/or rankings ... according to the respective dynamically adjusted preselected global list weights”, is not taught or suggested by Nida or Micaelian, alone or in any feasible combination.
The Examiner respectfully disagrees because Under 35 U.S.C. § 103, a POSITA. Applying Micaelian’s known client-adjustable weighting technique to Nida’s expressly customizable multi-list prioritization would have predictably allowed different client terminals to submit different list weights and receive corresponding ranked outputs. Refer obvious rational below. Therefore, the rejection is maintained.
Applicant argues that Claim 1, “wherein the respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list are adjustable to different values by different users via respective user interfaces”, and the related requirement that “each respective preselected global list weight comprises an absolute value of one or greater”, are not taught or suggested by Nida in view of Micaelian because Micaelian teaches only relative feature importance using slider bars whose positions are translated into “a normalized value between Zero and one” (Micaelian, col. 8, lines 19-39), which Applicant contends is different from the claimed absolute weighting.
The Examiner respectfully disagrees because under proper BRI, Claim 1, “preselected global list weight”, requires a numerical value indicating list priority for a client terminal, and “absolute value of one or greater” specifies only the chosen numeric scale, not a different weighting principle. The Specification states “respective preselected global list weights indicating the magnitude of prioritization ... comprise a numerical value” and “are adjustable to different values by different users” (Spec. [0006], [0010]). Nida teaches the list-combination framework and provider-specific prioritization logic, including that different users or facilities may prefer different rubrics when queues are generated (Nida [0046]-[0047]). Micaelian teaches the missing user-interface adjustment of numerical weights and server use of those weights in ranking, even though one disclosed example uses “a normalized value between Zero and one” (Micaelian, col. 8, lines 19-39; col. 10, lines 30-45). Applicant is correct that Micaelian’s illustrated example is relative and normalized, but that does not persuasively distinguish the claim, because once numerical client-adjustable weights are taught, choosing a different positive scale, including values of 1 or greater, would have been a predictable matter of routine parameterization rather than a different mechanism. Therefore, the rejection is maintained.
Applicant argues that Claim 1, “wherein the respective preselected global list weights ... are adjustable to different values by different users via respective user interfaces”, is not suggested by a combination of Nida and Micaelian because Nida addresses combining machine-learning-generated clinical priority lists using model performance and business logic, while Micaelian addresses subjective user preferences in consumer-product searching, and thus a POSITA would not have combined the references.
The Examiner respectfully disagrees because under proper BRI, Claim 1, “preselected global list weights”, requires user-adjustable numerical priorities applied to ranking inputs, not that the weighting originate from the same application domain as the underlying list generator. Nida teaches combining and prioritizing multiple model outputs into a clinical queue and expressly states that “any arbitrary business logic may be use[d] to prioritize and combine lists” and that a business user may prioritize one model over another (Nida [0046]); Nida further teaches that different providers may prefer queues scored according to different rubrics (Nida [0047]). Micaelian teaches a known user-interface technique for inputting adjustable numerical weights and using those weights to rank outputs. Applicant is not persuasive that Nida’s disclosure of a trained model or, alternatively, instructions, teaches away from user-adjustable weighting; those are disclosed implementation options for queue management, not a criticism or disavowal of weighting inputs. Applicant is also not persuasive that Micaelian is irrelevant merely because it ranks search results rather than pre-existing priority lists, because the relied-upon teaching is the known technique of client-side adjustable weighting to alter ranked output, which is reasonably applicable to Nida’s expressly customizable prioritization framework. On this record, using Micaelian’s adjustable weighting interface to implement Nida’s user- or rubric-dependent prioritization would have been a predictable use of a known ranking technique for its established purpose. Therefore, the rejection is maintained.
Applicant argues that Claims 1, 26, and 45, “the respective preselected global list weights ... adjustable to different values by different users via respective user interfaces”, together with server-side inference and different client-specific scores/rankings, would not have been obvious over Nida and Micaelian because the proposed combination is technically inoperable. Applicant contends Nida’s list combination uses a trained machine-learning model or, alternatively, instructions, whereas Micaelian uses transparent weighted multiplication, and that adding user-adjustable weights would require retraining Nida’s model for each weight configuration or abandoning Nida’s trained-model approach.
The Examiner respectfully disagrees because under proper BRI, Claims 1, 26, and 45, “preselected global list weights”, require user-adjustable numerical weighting inputs used in generating ranked outputs, not retraining the combined model for every new client weight selection. Nida teaches that the queue management module may include a model trained to combine and sort lists and “alternatively” may use instructions, and further teaches that the module combines lists using model scores, model performance, and “any arbitrary business logic” including user-selected priorities (Nida [0046]). That disclosure does not limit Nida to a single non-adjustable black-box architecture, and it does not teach that client-specific prioritization inputs would render the system inoperable. Micaelian teaches the known technique of client-adjustable numerical weights used to alter ranked output, and the Office Action relies on Micaelian for that adjustable-weighting feature, not for replacing all of Nida’s queue-generation logic. Applicant’s asserted need to retrain for each weight configuration is unsupported attorney argument; the present record does not show that Nida’s disclosed queue management framework could not accept adjusted weighting inputs at inference or through its alternative instruction-based implementation. On this record, applying Micaelian’s adjustable weighting technique within Nida’s expressly flexible list-prioritization framework would have been a predictable and operable combination, and the same reasoning applies to Claims 26 and 45 because Applicant states the argument is the same. Therefore, the rejection is maintained.
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 1-5, 8-10, 12-16, 20, 25, 26, and 45 are rejected under 35 U.S.C. § 101 because the claimed
subject matter is directed to a judicial exception (an abstract idea) without reciting elements that
integrate the exception into a practical application or provide an inventive concept amounting to
significantly more than the exception itself.
Step 1: Statutory Categories Analysis
The claims are directed to statutory subject matter, encompassing the following statutory category:
Machine (Claims 1-5, 8-10, 12-16, 20, 25, 26, and 45): The language reciting "A system... comprising: at
least one processor executing a code for..." (Claims 1, 26, and 45) describes a concrete thing consisting
of parts, aligning with the definition of a machine in MPEP § 2106.03.
Having confirmed the claims are directed to statutory subject matter, the analysis proceeds to Step 2A.
Step 2A, Prong One: Judicial Exception Analysis
Prong One asks whether the claim recites a judicial exception.
The following representative claim 1 recite abstract idea non-bold and additional element in bold:
Independent Claim 1 Analysis
A system for training a plurality of machine learning models and for training a combined prioritization machine learning model for dynamic prioritization of target subjects for at least one of evaluation, and treatment, comprising:
at least one processor of a server in communication with a plurality of client terminals over a network, executing a code for:
accessing electronic medical records (EMR) of a set of a plurality of subjects;for at least one of each of a plurality of specific medical interventions that are resource limited, and for each respective target clinical outcome:
accessing a respective specific priority list of a respective sub- set of the plurality of subjects scheduled in a prioritized sequence for at least one of treatment and evaluation for at least one of the respective target clinical outcome and by the respective specific medical intervention;
creating a respective specific training dataset that includes data extracted from the EMRs of at least the respective sub-set of the plurality of subjects labelled with the specific priority list;
training a respective specific machine learning model on the respective specific training dataset for generating an outcome of a respective target specific priority list of a sub-set of target subjects for prioritized at least one of evaluation and treatment for the at least one of respective target clinical outcome and by the respective specific medical intervention, in response to an input of data extracted from EMR of at least the sub-set of target subjects;
accessing code of an intermediate component for computing a plurality of weighted specific priority lists by assigning a respective preselected global list weight to each respective target specific priority list outcome of respective specific machine learning models, creating a main training dataset that includes the plurality of weighted specific priority lists obtained as outcomes of the plurality of specific machine learning models assigned respective preselected global list weights, labelled with a respective combined priority list of the set of the plurality of subjects for at least one of treatment and evaluation;training a combined prioritization machine learning model on the main training dataset for generating an outcome of a target combined priority list of target subjects for prioritized at least one of evaluation and treatment in response to an input of the plurality of weighted specific priority lists; andproviding the plurality of specific machine learning models, the code of the intermediate component, and the combined prioritization machine learning model,wherein the plurality of specific machine learning models and the combined prioritization machine learning model are implemented as one or more or combination of the following architectures: neural network, logistic regression, decision tree, and boosting,wherein the plurality of specific machine learning models and the combined prioritization machine learning model generate a respective numerical score for each respective subject, and the specific priority lists and the combined priority list are created by ranking subjects according to respective numerical scores,wherein the combined prioritization machine learning model computes,for each respective subject of the plurality of target subjects, a respective weighted score by multiplying each respective numerical score assigned to the respective subject on each respective priority list by the respective global list weight assigned to the respective list;
anda plurality of client terminals in communication with the server over the network, each client terminal including at least one processor executing a code for:
presenting on each respective display of each respective client terminal of the of the plurality of client terminals, graphical user interface (GUI) configured for dynamically adjusting the respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list to respective values,wherein different GUIs dynamically adjust the preselected global list weights to different values,wherein each respective preselected global list weight indicates a magnitude of prioritization of the respective target specific priority list for the respective client terminal,wherein each respective preselected global list weight comprises an absolute value of one or greater,wherein the same respective preselected global list weight is applied to all scores of all subjects of the same respective target specificity priority list for the respective client terminal,wherein each respective global weight of each respective priority list is adjustable to a different value by a different client terminal;transmitting the respective dynamically adjusted preselected global list weights of the respective client terminal to the server,wherein the server performs inference using the combined prioritization machine learning model using the different values of the preselected global list weights received from the plurality of client terminals; andreceiving from the server and presenting within the GUI on each respective display, for each respective subject, at least one of:
the respective numerical score computed by the combined prioritization machine learning model, and a ranking of the respective subject on the combined priority list, computed according to the respective dynamically adjusted preselected global list weights,wherein the server sends different numerical scores and/or rankings to corresponding client terminals according to the dynamically adjusted preselected global weights.
Claim Abstract Idea Classification Rational
Under the broadest reasonable interpretation consistent with the specification, Claims 1, 26, and 45 are directed to a server-client computerized prioritization system that uses EMR-derived subject data, specific priority lists, machine learning model outputs, preselected global list weights, weighted numerical scores, and rankings to generate different client-specific priority results for treatment or evaluation.
The claim language requires training or using specific machine learning models and a combined prioritization machine learning model, but the claimed functional result remains the generation of prioritized medical evaluation or treatment lists by numerical weighting and ranking. Claim 1 expressly recites numerical score, ranking subjects according to respective numerical scores, and multiplying each respective numerical score assigned to the respective subject on each respective priority list by the respective global list weight. The specification par. 0003 confirms that the invention addresses patient prioritization for treatment or evaluation, where traditional prioritization was by manual guess or reactive emergency treatment.
Claims 1, 26, and 45 recite mathematical concepts. The claims require numerical scores, global list weights, multiplication of each subject score by a list weight, and ranking according to weighted scores. These are mathematical relationships and calculations. The mathematical operation is not incidental because it is the claimed mechanism used to compute the combined priority result. Claims 1, 26, and 45 also include mental-process subject matter to the extent the claims evaluate subject information, apply prioritization judgments, compare scores, and rank subjects for treatment or evaluation. Under BRI, the abstract decision logic could be conceptually performed by a human medical committee using patient records, separate priority lists, assigned weights, multiplication, comparison, and ranking.
Claims 1, 26, and 45 also recite certain methods of organizing human activity, specifically managing personal behavior or relationships or interactions between people, because the claims determine the order in which target subjects are prioritized for medical evaluation or treatment by healthcare resources. The claim language expressly recites priority lists of subjects scheduled in a prioritized sequence for treatment and evaluation, for target clinical outcome, and by specific medical intervention where the intervention may be resource limited. The ranked list therefore does not merely calculate a number; it organizes clinical decision-making and coordinates which patients should be evaluated or treated first in view of limited medical resources. Thus, under BRI, the claim recites a healthcare resource-prioritization and treatment/evaluation-ordering process that falls within the certain-methods-of-organizing-human-activity grouping, sub-category managing personal behavior or relationships or interactions between people.
The machine learning and server-client language does not remove the recited abstract idea at Prong One. Those components are consider additional elements for prong two evaluation, because they specify the environment and tools used to perform the weighting, scoring, ranking, transmission, and display of the prioritization result.
Manual Replication Scenario (Human Equivalence)
The following human equivalence show why and how abstract, a human medical committee could perform the same abstract prioritization logic by reviewing patient records, maintaining separate clinical priority lists for different outcomes or resource-limited interventions, assigning an importance weight to each priority list, multiplying each patient’s list score by the assigned list weight, comparing the resulting weighted scores, and producing a final ranked schedule identifying which patients should be evaluated or treated first. The human analogy shows that the abstract core is not the hardware or software implementation, but the mathematical and organizational process of weighting and ranking patients for clinical prioritization.
Dependent Claims Analysis
The dependent claims are also directed to an abstract idea.
Claims 2-5, 25: These claims recite details of the "global list weights." Under BRI (MPEP 2111), these
limitations merely narrow the abstract concept of judgmental evaluation and analysis (Mental Process)
regarding how the prioritization magnitude is determined or constrained.
Claims 8, 9: These claims recite "aggregation code" (Cl. 8) and the steps of computing "weighted score[s]
by multiplying" and "ranking" (Cl. 9). Under BRI, this explicitly recites the analytical and comparison
steps used in the abstract idea (Mental Process).
Claims 10, 12, 13: These claims recite "time correlated priority lists" (Cl. 10, 12) or basis on a "schedule
of availability of resources" (Cl. 13). Under BRI, this involves organizing schedules and managing the
timing of interventions, which is a Certain Method of Organizing Human Activity (sub-category:
Managing Personal Behavior or Relationships / Resource allocation).
Claim 14: This claim recites that the training data includes "indication of risk" or "diagnosis." Under BRI,
this involves analyzing specific data points to determine priority, which is a Mental Process.
Claims 15, 16, 20: These claims recite the implementation of the system, such as sequentially
treating/evaluating subjects (Cl. 16). Under BRI, these claims describe the implementation of the
organized workflow, which is a Certain Method of Organizing Human Activity (sub-category Managing
Personal Behavior or Relationships).
Step 2A, Prong Two: Integration into a Practical Application
Prong Two evaluates only the additional elements beyond the abstract idea identified in Prong One. The abstract idea includes EMR-based medical prioritization using specific priority lists, global list weights, numerical scores, multiplication of scores by weights, weighted scores, and ranking subjects for treatment or evaluation. The additional elements evaluated here are:
Additional elements:server processor, client-terminal processors, network communication, client terminals, GUI on each client-terminal display, code executed by the server and client terminals, transmitting adjusted values to the server, server-side inference using the adjusted values, receiving results from the server, and presenting scores or rankings within the GUI.
These additional elements do not integrate the abstract idea into a practical application because they implement the weighted medical-prioritization calculation using generic computer components, generic GUI input/output, generic server-client communication, and generic machine-learning execution, without improving the operation of the computer system, GUI, network, processor, memory, or machine-learning technology.
Server processor, client-terminal processors, network communication, and client terminals
The server processor, client-terminal processors, network communication, and client terminals do not improve computer functionality. The claims use these components to execute code, communicate adjusted values, perform inference, return results, and present scores or rankings. The claims do not recite a new server architecture, network protocol, processor operation, memory structure, transmission protocol, or distributed-computing mechanism.
The specification supports this determination because it describes the computing environment generically. Paragraph [0098] describes program instructions downloaded over the Internet, a local area network, a wide area network and/or a wireless network. Paragraph [0099] states that instructions may execute on the user computer, partly on a remote computer, or entirely on a remote computer or server. Paragraph [0101] describes instructions provided to a general purpose computer, special purpose computer, or other programmable data processing apparatus.
Under MPEP § 2106.05(a), these elements do not show an improvement to computer functionality or another technology because the claims do not reflect a technical improvement in how the server, processors, network, or client terminals operate. MPEP § 2106.05(a) requires the claim to reflect the asserted technological improvement, and the specification must provide enough technical detail for a person of ordinary skill to recognize that improvement. Here, the specification supports flexible generic implementation, not a specific improvement in server, processor, network, or client-terminal operation. These elements therefore amount to using computer components as tools under MPEP § 2106.05(f) and generally linking the abstract prioritization logic to a computer-network environment under MPEP § 2106.05(h).
GUI on each client-terminal display, receiving adjusted values, transmitting adjusted values, receiving results, and presenting scores or rankings
The GUI and its related input/output functions do not improve GUI technology. The claims use the GUI to dynamically adjust values, transmit adjusted values to the server, receive numerical scores or rankings from the server, and present those results. The claims do not recite a specific GUI layout, control structure, rendering technique, synchronization technique, accessibility improvement, display-memory technique, or reduction in computing-resource usage.
The specification confirms that the GUI is used as an interface for prioritization values. Paragraph [0153] states that global list weights may be adjustable to different values by different users, for example, using a user interface such as a GUI. Paragraph [0155] states that each respective weight may be dynamically adjustable by a user, for example, via a GUI presented on a display. These disclosures support a GUI as an input/display interface for values used in the prioritization calculation, not as a claimed improvement in GUI operation.
Under MPEP §§ 2106.05(a), 2106.05(f), and 2106.05(g), these GUI-related elements do not integrate the abstract idea into a practical application. They receive input values, transmit values, receive computed output, and display the result of the abstract weighted-prioritization calculation. They do not improve the GUI or display technology itself.
Server-side inference and execution of machine-learning models
The server-side inference and machine-learning execution do not improve machine-learning technology. The claims recite use of machine-learning models implemented as neural network, logistic regression, decision tree, and boosting, but do not recite a specific training algorithm improvement, model architecture improvement, loss-function improvement, feature-engineering mechanism, memory-saving mechanism, catastrophic-forgetting solution, or inference-efficiency technique.
The specification describes the asserted improvement as customization of medical prioritization by assigning global list weights to outputs of machine-learning models, not as an improvement to the internal operation of the models. Paragraph [0075] states that the improvement is based on assignment of global list weights to the specific priority lists and that the weights are applied to numerical values to compute weighted values used to create the combined priority list. Paragraph [0077] explains that different global-list-weight magnitudes generate different combined priority lists even when relative prioritization is maintained. This is an improvement to prioritization customization, not a claimed improvement to machine-learning architecture, training operation, or inference technology.
Under MPEP § 2106.05(a), an asserted improvement must be reflected in the claim as a technological improvement. Under MPEP § 2106.05(f), using generic machine-learning execution to perform the abstract analysis is a mere instruction to apply the exception using computer technology.
Client-specific computer implementation of the weighting logic
The client-specific computer implementation of the weighting logic does not supply technological integration. The abstract idea includes the global-list-weight prioritization logic itself. The additional elements are the GUI/client-server implementation steps that allow different client terminals to adjust values, transmit those values to the server, and receive corresponding different scores or rankings.
The specification confirms that the changed values affect the priority calculation, not the operation of the computer components. Paragraphs [0148]-[0153] describe global list weights as numerical values, geographical selections, rule-defined values, range-limited values, constrained values, and values adjustable by different users using a GUI. Paragraph [0154] states that the aggregated score is computed by multiplying each numerical score by the respective global list weight and summing the weighted scores. Thus, different client-specific values produce different prioritization outputs because the mathematical input changes, not because the GUI, server, network, processor, or model operates in a technically improved way.
Under MPEP §§ 2106.05(a), 2106.05(f), and 2106.05(h), this client-specific implementation does not integrate the abstract idea into a practical application because it applies mathematical weighting within a healthcare-computer environment rather than improving the technology itself.
When viewed individually and as an ordered combination, the additional elements use generic server-client computing, generic GUI input/output, generic network communication, and generic machine-learning execution to implement the abstract weighted medical-prioritization calculation. MPEP § 2106.04(d) and the 2024 AI guidance require evaluation of the additional elements individually and in combination to determine whether the claim integrates the exception into a practical application. The claims do not recite the type of technological improvement found eligible in the USPTO AI examples, where the claim reflected a disclosed technical solution in the technology itself.
Dependent Claims Analysis
The dependent claims add only minor limitations that fail to provide the necessary integration.
Claims 2-5, 8, 9, 14, 25: These claims merely narrow the abstract ideas (Mental Processes) by detailing
the nature of the weights or the analytical aggregation process. They do not add additional elements
that integrate the exception.
Claims 10, 12, 13: These claims add the abstract idea of time correlation and scheduling (Method of
Organizing Human Activity). This addition of another abstract idea does not integrate the exceptions
into a practical application.
Claim 15: This claim adds the steps of using the system (e.g., "feeding the set of target EMRs"). This is
insignificant pre-solution activity (g) (data gathering).
Claims 16, 20: These claims add "treating and evaluating" subjects or "generating instructions" for the
same. This is a mere field-of-use limitation (h) (medical treatment) and insignificant post-solution
activity (g) (acting upon the results of the abstract prioritization).
Because the claims are directed to an abstract idea without integrating it into a practical application, the
analysis proceeds to Step 2B.
Step 2B: Inventive Concept Analysis
Step 2B evaluates the same additional elements considered in Prong Two, but with a different focus: whether those elements, individually or in ordered combination, amount to significantly more than the abstract idea. The abstract idea remains EMR-based medical prioritization using specific priority lists, global list weights, numerical scores, multiplication of scores by weights, weighted scores, and ranking subjects for treatment or evaluation. The additional elements evaluated here are the same additional elements evaluated in Prong Two:
Additional elements:server processor, client-terminal processors, network communication, client terminals, GUI on each client-terminal display, code executed by the server and client terminals, transmitting adjusted values to the server, server-side inference using the adjusted values, receiving results from the server, and presenting scores or rankings within the GUI.
These additional elements do not amount to significantly more than the abstract idea because they perform ordinary computer, communication, interface, inference, and display functions at a high level of generality.
Server processor, client-terminal processors, network communication, and client terminals
The server processor, client-terminal processors, network communication, and client terminals do not add an inventive concept. The claims use these components for ordinary computer functions: executing code, communicating values, performing inference, returning results, and presenting scores or rankings. The claim language does not require any non-generic computer arrangement or technical improvement to these components.
The specification supplies factual support for this determination. Paragraph [0098] describes ordinary network delivery over Internet, LAN, WAN, or wireless networks. Paragraph [0099] describes execution using ordinary programming languages and execution locally, remotely, or on a server. Paragraph [0101] describes use of a general-purpose computer, special-purpose computer, or programmable data-processing apparatus. Paragraphs [0107]-[0109] further describe broad implementation as client terminal, server, cloud, EMR server, virtual server, mobile device, desktop computer, thin client, smartphone, tablet, laptop, wearable computer, CPU, GPU, FPGA, DSP, ASIC, RAM, ROM, magnetic media, semiconductor memory, hard drive, removable storage, and optical media.
Under MPEP § 2106.05(d), generic computer functions specified at a high level of generality do not supply significantly more. Under MPEP § 2106.05(f), mere instructions to implement an abstract idea on a computer do not add an inventive concept. Under MPEP § 2106.05(g), data gathering and output activity do not add significantly more.
GUI on each client-terminal display, receiving adjusted values, transmitting adjusted values, receiving results, and presenting scores or rankings
The GUI and its related input/output functions do not add an inventive concept. The claims use the GUI and related communication functions to receive adjusted values, transmit those values to the server, receive computed scores or rankings, and display the results. The claim language does not require a non-generic GUI mechanism, display mechanism, input control, rendering process, synchronization process, or interface architecture.
The specification supports this conclusion. Paragraph [0153] states that global list weights may be adjusted by different users using a GUI. Paragraph [0155] states that each weight may be dynamically adjustable by a user via a GUI presented on a display. Paragraph [0148] describes the weights as numerical values; paragraph [0150] describes rule-defined weights; paragraph [0151] describes range-limited weights; and paragraph [0152] describes criteria restricting the weights. These disclosures show mathematical and administrative control of prioritization values, not an unconventional GUI mechanism.
Under MPEP §§ 2106.05(d), 2106.05(f), and 2106.05(g), the GUI adjustment, transmission, receipt, and presentation limitations do not add significantly more because they receive, transmit, and display abstract prioritization values using generic interface functions at a high level of generality.
Server-side inference and execution of machine-learning models
The server-side inference and machine-learning execution do not add an inventive concept. The claims recite broad machine-learning architecture categories and use them to generate numerical scores, priority lists, weighted scores, and rankings. The claims do not recite a non-conventional ML structure, training technique, inference technique, data structure, or model-operation improvement.
The specification describes the asserted improvement as assigning global list weights to specific priority lists to create customized combined priority lists. Paragraph [0075] states that the improvement is based on assignment of global list weights to the specific priority lists. Paragraph [0077] explains that different global-list-weight magnitudes generate different combined priority lists. Paragraph [0154] confirms that the score computation is multiplication of numerical scores by list weights followed by summing weighted scores. These disclosures support customized scoring and prioritization, not a non-conventional ML model structure or training process.
Under MPEP § 2106.05(d), additional elements recited at a high level of generality must add more than generic implementation. Under MPEP § 2106.05(f), use of a trained ANN or ML model merely to perform the abstract idea, without limiting how the model operates, does not provide an inventive concept. The USPTO AI examples apply the same reasoning where a generic DNN or trained ANN merely performs the abstract analysis without claim-recited technical details of how the model operation is improved.
Client-specific computer implementation of the weighting logic
The client-specific computer implementation of the weighting logic does not add an inventive concept. The global-list-weight calculation itself is part of the abstract idea. The additional elements are the GUI/client-server implementation steps that allow different client terminals to adjust values, transmit those values to the server, and receive corresponding different scores or rankings. These implementation steps change the mathematical prioritization input and resulting output for each client terminal, but do not recite an unconventional GUI mechanism, server architecture, network protocol, processor arrangement, memory structure, or ML training improvement.
The specification supports this conclusion. Paragraphs [0148]-[0153] describe global list weights as numerical values, geographical selections, rule-defined values, range-limited values, constrained values, and values adjustable by different users using a GUI. Paragraph [0154] confirms that the score computation is multiplication of numerical scores by list weights followed by summing weighted scores. These disclosures show that the client-specific implementation changes the values used in the prioritization calculation, not the operation of the computer components.
Under MPEP §§ 2106.05(d), 2106.05(f), and 2106.05(g), these elements do not add significantly more because they implement the abstract calculation with ordinary input, transmission, inference, receipt, and display functions.
The ordered combination does not add significantly more. As an ordered combination, the claims use server processors, client-terminal processors, network communication, GUIs, server-side inference, receipt of results, and display of scores or rankings to implement the abstract weighted medical-prioritization calculation. The ordered combination automates and distributes the prioritization calculation across a server-client environment, but does not recite a non-generic computer arrangement, a specific GUI improvement, a network-function improvement, a processor or memory improvement, a particular ML training improvement, or a particular treatment step.
The specification describes the same combination as flexible computer implementation and weighted prioritization. Paragraphs [0098]-[0103] describe generic computer-program and network implementation. Paragraphs [0075]-[0077] and [0153]-[0155] describe assignment and adjustment of global list weights to generate customized combined priority lists.
Under MPEP § 2106.05, Step 2B carries forward the Prong Two conclusions and evaluates whether the same additional elements, individually and in combination, amount to significantly more. MPEP § 2106.05 identifies mere computer implementation, high-level generic computer activity, insignificant extra-solution activity, and general technological-environment limitations as insufficient.
Dependent Claims Analysis
Group 1 (Claims 2-5, 25): These claims add details of the weights (numerical, geographical, rule-based, ranged, user-adjustable). Applying weights based on judgment to prioritize items is a well-understood, routine, and conventional activity (d) in decision support. The specification confirms these are standard ways to define weights: "The global list weights may be defined according to a set of rules." (Spec., para. [0155]).
Group 2 (Claims 8, 9): These claims recite "aggregation code" (Cl. 8) and the steps of weighting and ranking (Cl. 9). Implementing data aggregation and ranking via generic code is a mere instruction to perform a mathematical operation (f) and is conventional (d). The specification describes this generically: "The combined prioritization component comprises aggregation code that when executed... aggregates... the respective numerical scores..." (Spec., para. [0159] and Claim 8).
Group 3 (Claims 10, 12, 13): These claims add time correlation and scheduling based on resource availability. Scheduling based on priority and time constraints is a well-understood, routine, and conventional activity (d) in resource management and project management.
Group 4 (Claim 14): This claim specifies the use of risk or diagnosis data. Analyzing specific types of medical data is a conventional application of data analysis techniques (d) and constitutes a mere field-of-use limitation (h) that does not add an inventive concept.
Group 5 (Claims 15, 16, 20): These claims involve using the models (Cl. 15) and subsequent treatment steps (Cl. 16, 20). These are insignificant post-solution activities (g) that are conventional (d) in the medical field. The specification confirms that "accessing and processing EMRs is conventional" (Spec., para. [0120]), and applying the results for treatment is the conventional end goal of any medical analysis (Spec., para. [0218]).
The combination of the dependent claims with the independent claims does not add an inventive concept, as they only add other abstract ideas, insignificant activities, or conventional elements to the underlying abstract prioritization methodology.
The claims are directed to abstract ideas (Certain Methods of Organizing Human Activity and mathematical) and lack an inventive concept. Therefore, Claims 1-5, 8-10, 12-16, 20, 25, 26, and 45 are rejected under 35 U.S.C. § 101.
Claim Rejections - 35 USC § 103
1. 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.
2. 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.
Claim(s) 1-5, 8-10, 12-16, 25-26, and 45 is/are rejected under 35 U.S.C. 103 as being unpatentable over US20200279641A1- Dean Nida, and further in view of US6714929B1-Fadi Victor Micaelian.
Claim 1.
Nida teaching, A system for training a plurality of machine learning models and for training a combined prioritization machine learning model for dynamic prioritization of target subjects for at least one of evaluation, and treatment, comprising: (Nida, See at least, abstract … actively improving clinical queues are provided . …generating , with a plurality of predictive models , a plurality of risk scores for each individual a plurality of individuals based on medical claims of each individual ,transmitting a clinical queue to a client device associated with a healthcare provider , the clinical queue comprising a list of individuals prioritized for healthcare intervention based on the plurality of risk scores , receiving feedback from the client device regarding the clinical queue , and updating at least one predictive model of the plurality of predictive models based on the feedback . In this way , members may be identified for intervention in a timely manner , multiple assessment methods may be combined to generate clinical queues … par.0049 …model training / evaluation module 240 [is] configured to… improve one or more models of the modeling module 201 based on the queue feedback… model training / evaluation module 240 [is] configured to receive the queue feedback… and improve one or more models of the modeling module 201 based on the queue feedback…., par. 0046 … machine learning model specifically trained to combine and sort the lists output by the modeling module 201 into the clinical queue…, par. 0060… retrains the model based on the evaluated list output…, par. 0057… updates one or more models of the plurality of models based on the indication of the opened case…, par. 0045… combine the lists output by the plurality of models…, fig.2, par. 0042-0046, 0060)
Nida teaches the claimed system because modeling module 201 includes plural ML-capable models 202–204 ([0043]-[0044]); queue management module 210 combines their outputs into prioritized clinical queues and may include an ML model “specifically trained” to do so ([0045]-[0046]); and training/evaluation module 240 uses feedback to improve/update queue-generation models ([0049]).
Fig. 2 shows one operative system capable of the claimed function: models 202–204 generate outputs, module 210 combines/sorts them, module 230 receives feedback, and module 240 updates/retrains models using feedback, opened-case indications, and “good”/“bad” labels ([0060]). Thus, a POSITA would understand Nida as disclosing the preamble: 201 provides the plurality of ML models, 210 provides the combined prioritization ML model, and 240 provides the training/evaluation loop.
The system combines multiple assessments and updates (trains) the models based on feedback.
at least one processor of a server in communication with a plurality of client terminals over a network, executing a code for: accessing electronic medical records (EMR (Nida, See at least, par. 0020, 0024, 0026, 0042-0043, 0049, 0059-0060, 0093), The queue management system accesses "healthcare databases" storing "medical claims." Medical claims are a core component of EMRs. Accessing these databases constitutes accessing EMRs.
for at least one of each of a plurality of specific medical interventions that are resource limited, and for each respective target clinical outcome: (Nida, See at least, abstract, par. 0043, abstract, 0019, 0003, par. 0090, par. 0004, par. 0082, 0089, 0056, 0082, claim 3),
Nida discloses prioritized clinical queues for healthcare interventions, with examples like follow-ups and appointments. These queues address resource limitations and target clinical outcomes, such as risk and cost, supporting intervention prioritization.
accessing a respective specific priority list of a respective sub-set of the plurality of subjects scheduled in a prioritized sequence for at least one of treatment and evaluation for at least one of the respective target clinical outcome and by the respective specific medical intervention; (Nida, See at least, par. 0019, 0046-0047, 0043, 0050-0051, 0055-0056, 0060, 0084, 0005, Claim 1-3), Nida reads on this under a broad but reasonable interpretation. Its “specialized lists” and “second queue” are respective specific priority lists; the “members identified as high risk” are a respective sub-set; and the lists are expressly “sorted” and “prioritized.” Nida also ties the lists to treatment/evaluation because the queue is used for case opening, review, contact, follow-up visits, and scheduling visits. On the outcome side, Nida discloses distinct predicted outcomes or targets such as future healthcare cost, in-patient length of stay, and risk for healthcare episodes. On the intervention side, Nida discloses specific intervention actions such as contact, follow-up visit, and scheduling a visit. Read together, those passages support that the prioritized subset is used for intervention/evaluation with respect to specific predicted clinical targets.
creating a respective specific training dataset that includes data extracted from the EMRs of at least the respective sub-set of the plurality of subjects labelled with the specific priority list; (Nida, See at least, abstract, par. 0019, 0026, 0043, 0045, 0049, 0052, 0059, 0060, 0081-0082, fig.2, fig.5), Nida’s accessible “healthcare databases” storing member medical claims, and its patient data including claims, diagnoses, and prescriptions, are electronic medical patient records/sources under ordinary meaning. Nida also discloses creating training inputs for the relevant subset by constructing feature vectors “for each reviewed member,” and expressly discloses a “training dataset.” On labeling, Nida ties the labels to the specific generated list by identifying which feedback and opened cases “correspond to the list generated by the model,” and labeling those instances as “good”/“bad” or positive/negative also use feedback that include score for a subject in a list. A POSITA would understand that as a labeled training dataset for the specific prioritized list/subset.
and training a respective specific machine learning model on the respective specific training dataset for generating an outcome of a respective target specific priority list of a sub-set of target subjects for prioritized at least one of evaluation and treatment for the at least one of respective target clinical outcome and by the respective specific medical intervention, in response to an input of data extracted from EMR of at least the sub-set of target subjects; (Nida, See at least, abstract, par. 0005, 0043-0045, 0049, 0051, 0059-0060, 0089-0090), Nida reads on this limitation under that BRI. It expressly discloses machine-learning models, a training dataset, and training based on feedback/opened cases. It also discloses electronic patient medical data inputs from healthcare databases, including claims, diagnoses, and prescriptions, which under the ordinary-meaning BRI read on EMR-derived patient data. Nida further teaches that the trained model outputs lists and specialized lists for subsets such as high-risk members, and those lists are prioritized for provider review, case opening, contact, follow-up, or scheduling a visit. It also ties the generated list to target clinical outcomes such as future healthcare cost, length of stay, and risk for healthcare episodes, and to specific intervention actions by learning which at-risk members may benefit from an available intervention.
accessing an intermediate component for computing a plurality of weighted specific priority lists by assigning a respective preselected global list weight to each respective target specific priority list outcome of respective specific machine learning models, (Nida, par. 0019, 0043 – 0046, 0050-0051, 0082, 0089, fig.2, fig.5)
Nida reads on this limitation under that BRI. The “queue management module 210” is the intermediate component. It operates on a plurality of model-specific list outcomes, combines them, sorts them, and derives one or more queues and specialized lists. Nida also discloses list-level weighting in substance: prioritization can be based not only on the member scores within each list, but also on the “performance of individual models,” where one model is prioritized over another, and on preselected business logic, where a business user decides one model type should be prioritized over another. A POSITA would understand those model-level prioritizations as assigning relative global weights or preferences to the corresponding list outputs before computing the resulting prioritized queues.
creating a main training dataset that includes the plurality of weighted specific priority lists obtained as outcomes of the plurality of specific machine learning models assigned respective preselected global list weights, labelled with a respective combined priority list of the set of the plurality of subjects for at least one of treatment and evaluation; (Nida, Abstract; paras. 0005, 0019, 0043-0047, 0049-0050, 0058-0061), Nida's system uses multiple machine learning models to create a ranked patient list and then retrains those models using real-world feedback on that list, such as which patients were actually selected for treatment. This retraining process is functionally the same as creating a new, weighted, and labeled training dataset based on the models' own outcomes.
training a combined prioritization machine learning model on the main training dataset for generating an outcome of a target combined priority list of target subjects for prioritized at least one of evaluation and treatment in response to an input of the plurality of weighted specific priority lists; (Nida, See at least, par. 0045, Abstract; paras. 0043, 0046, 0049, 0058, 0059, 0061)
Nida teaches using a "queue management module" to process a "plurality of risk scores" from multiple models into a single "clinical queue," directly showing the refinement of multiple lists into one. Machine learning further improves this process, as the system is configured to "update one or more of the predictive models... based on the feedback" to generate better subsequent queues.
And providing the plurality of specific machine learning models, the code of the intermediate component, and the combined prioritization machine learning model, (Nida, See at least, par. 0043, 0045, 0048, fig. 2, fig. 12, 0096), Nida explicitly discloses providing the claimed components as part of an integrated "queue management system 200," as shown in FIG. 2. The system is described as including "a plurality of predictive models 202" and "a queue management module 204," which directly correspond to the claimed specific models and intermediate component. By providing the system itself, for example as a "server configured to communicate with a plurality of client devices," Nida provides all the functional software components contained within it.).
wherein the plurality of specific machine learning models and the combined prioritization machine learning model are implemented as one or more or combination of the following architectures: neural network, logistic regression, decision tree, and boosting, (Nida, par. 0069)
wherein the plurality of specific machine learning models and the combined prioritization machine learning model generate a respective numerical score for each respective subject, and the specific priority lists and the combined priority list are created by ranking subjects according to respective numerical scores, (Nida, See at least, 0043, 0045-0046, 0019, 0051), The queue management module sorts individuals "according to the scores," which is equivalent to ranking them based on their numerical scores. This applies to both the individual model lists ("specific priority lists") and the final combined list.
wherein the combined prioritization machine learning model computes, for each respective subject of the plurality of target subjects, a respective weighted score (Nida, See at least, par. 0046),
a plurality of client terminals in communication with the server over the network, each client terminal including at least one processor executing a code for: (Nida, par. 0031, 0038-0041: “the server 101 provides a network service that is accessible to a plurality of users through a plurality of client systems”; “any number of user devices may be communicatively coupled to the server 101 via the network 115”; “client device 121 may include executable instructions 131… executed by the logic subsystem 123”; “client device 121 may be configured… to receive one or more clinical queues generated and transmitted by the server 101.”))
presenting on each respective display of each respective client terminal of the of the plurality of client terminals, a graphical user interface (GUI)
Nidia “the server 101 provides a network service that is accessible to a plurality of users through a plurality of client systems such as the client device 121” ([0031]); “the client device 121 may be configured ... to receive one or more clinical queues generated and transmitted by the server 101, display the one or more clinical queues via a graphical user interface on the display subsystem 125” ([0038]).
configured for dynamically adjusting the respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list to respective values wherein different GUIs dynamically adjust the preselected global list weights to different values,
Nidia, “the client device 121 may be configured ... to receive one or more clinical queues generated and transmitted by the server 101, display the one or more clinical queues via a graphical user interface on the display subsystem 125 ... and receive feedback regarding the one or more clinical queues” ([0038]); “a user of client device 121 may input feedback regarding a clinical queue ... [and] the server 101 may update one or more models for generating clinical queues based on the user input” ([0040]); “a model with a higher rate of positive cases may be prioritized over a model with a lower rate of positive cases ... any arbitrary business logic may be use[d] to prioritize and combine lists” ([0046]) “…transformed to visually represent changes in the underlying data… Such display devices may be combined with logic subsystem 103 ”([0029], “…any number of user devices may be communicatively coupled to the server 101 via the network 115…”([0031-0032]), “…outputs the dashboard to the client device 121 for display via the display subsystem 125.. In some examples, the modular dashboard module 730 adjusts the order of display modules according to predictions of the actionability of display modules determined by the actionability prediction model ”([0068]), “…dashboard displaying the display modules in the particular order…”([0070-0073])).
transmitting (Nida, 0031, 0046-0049, 0054-0060, 0070, 0087)
Nida teaches a plurality of client systems, server receipt of client-originated information, and a server-side machine learning model that combines and sorts outputs from multiple predictive models into prioritized clinical queues. Nida also teaches that prioritization may vary by user or facility preference, including different scoring rubrics and user-directed priority choices. Further, Nida teaches client-side adjustments and feedback that are transmitted back to the server and used to update a machine learning model. Under a broad reading, these disclosures come close to client-specific prioritization inputs used by the server.andreceiving from the server and presenting within the GUI on each respective display, for each respective subject, at least one of:
the respective numerical score computed by the combined prioritization machine learning model,
and
a ranking of the respective subject on the combined priority list, computed according to the respective dynamically adjusted preselected global list weights, wherein the server sends different numerical scores and/or rankings to corresponding client terminals according to the dynamically adjusted preselected global weights
Nida teaches server-to-client transmission, GUI presentation, model-generated scores, and a combined machine-learning-based queue prioritization process. Nida also teaches sorted/prioritized queues, which reasonably reads on ranking. It further teaches that different client devices may receive different queue outputs.
35 USC 103 Rational:
Nida weighted score is presented, but the method of calculation is not specified neither adjusting the specific numerical magnitude.
Method of Calculation
Micaelian teaches, the weighted summation method, where each alternative’s score is multiplied by a criteria weight and accumulated column 9, line 30 – line 41. A person skilled in the art would seek to improve the prioritization accuracy of clinical queues by incorporating a weighting system that accounts for the importance of different risk factors. Incorporating weighted summation from Micaelian into the clinical queue prioritization from Nida results in a more refined and adjustable ranking method.
Specific Numerical Magnitude
Nida teaches a system for combining multiple clinical priority lists into a master queue, explicitly stating that prioritization (weighting) occurs based on user-defined logic: "any arbitrary business logic may be use[d] to prioritize and combine lists. For example, a business user may decide..." (Nida [0046]). This establishes the motivation and framework for customizable weighting by different users. To the extent that Nida does not explicitly detail a user interface for adjusting the specific numerical magnitude of these weights, this feature is explicitly taught by Micaelian. Micaelian discloses a weight adjustment interface (e.g., slider bars) allowing users to assign numerical weights representing the magnitude of importance, normalized between zero and one (Micaelian, Col. 7, lines 20-40; Col. 8, lines 18-39). It would have been obvious to one of ordinary skill in the art to implement Nida's customizable prioritization logic using the known technique of user-adjustable magnitude weights taught by Micaelian to achieve a predictable result: a refined and customizable method for combining clinical priority lists.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the teachings of Nida with Micaelian because both references address the shared challenge of prioritizing and ranking data based on multiple, customizable criteria. Nida establishes the need for user-defined prioritization in clinical queues (Nida [0046]), and Micaelian provides a known methodology for implementing user-adjustable weighted prioritization (Micaelian, Col. 2, lines 34-39). A POSITA would recognize Micaelian's weighting interface as a concrete method to implement Nida's generalized “arbitrary business logic” (Nida [0046]).
The above limitation requires a preselected per-list, per-client-terminal weight that expresses how strongly a particular target specific priority list is to be prioritized for that particular client terminal.
Nida teaches combines multiple model-specific lists into client-delivered queues and prioritizes those lists when forming the queue. The output of each model may therefore comprise a list of members with associated scores in a given category. Nida para. 0043. Nida further states the queue management module 210 may be configured to sort the plurality of individuals in the plurality of lists according to the scores as well as according to the performance of individual models and a business user may decide that a high cost model should be prioritized over an inpatient model. Nida para. 0046. Under the BRI, this shows different target specific lists can be comparatively prioritized when generating a queue for a client device. However it does not expressly disclose a respective preselected weight value tied to each list for a particular client terminal that numerically indicates degree of priority.
Micaelian teaches that missing weighting feature. Micaelian discloses that a weighted preference generator generates weighted preference information including at least a plurality of weights corresponding to a plurality of search criteria. Micaelian Abstract. Micaelian further states that the method includes determining weighted preference information including a plurality of search criteria and a corresponding plurality of weights signifying the relative importance of the search criteria, and querying a data source and ranking the results based upon the weighted preference information. Micaelian Abstract. Micaelian explains that the weighted preference generator can include a user interface which allows a human user to input preferences into the generator, including the adjustment of weights with respect to the search criteria. Micaelian col. 2., ll. 53-67 Micaelian also teaches that the position of the indicator 110 along the slider bar 112 is translated into a numeric output, typically a normalized value between Zero and one, which is the weight for the criterion. Micaelian col. 8.ll.30-45 These passages teach express user or client specific numerical weights that indicate the degree of priority assigned to particular ranking inputs.
A POSITA would have combined Nida with Micaelian because Nida already requires combining and prioritizing multiple model-output lists, while Micaelian teaches a known and structured way to express relative priority through numerical weights. The specific modification would have been to apply Micaelian’s client-adjustable weighting technique to Nida’s queue management module so that each client terminal or associated user could assign a respective weight to each target-specific list, and the server could use those weights when combining and sorting the lists into a clinical queue. That result would have been predictable because it merely uses Micaelian’s known weighting approach to make Nida’s already-disclosed list prioritization explicit, tunable, and client-specific.
The above limitation requires only that each list weight be expressed as a numerical value at above 1. Under the applicant’s own context, this does not change how the weighting concept functions; it merely selects one numerical scale for expressing priority magnitude.
Nida teaches states the output of each model may therefore comprise a list of members with associated scores in a given category and that the queue management module 210 may combine all of the lists into a master queue and generate clinical queues comprising sorted and combined lists of members. Nida paras. 0043-0045. Nida further teaches that a model with a higher rate of positive cases may be prioritized over a model with a lower rate of positive cases and that a business user may decide that a high cost model should be prioritized over an inpatient model. Nida para. 0046. Thus, Nida shows the operative environment in which different list outputs are assigned different relative priority in forming a queue for a given client device or provider.
Nida does not expressly teach an absolute value of one or greater for each respective preselected global list weight.
Micaelian teaches the missing concept of express numerical weights used to signify relative priority. By give one example using a normalized range (fig.10, Col. 3, ll. 1-15, Col. 8, ll. 35-39), that disclosure confirms that the weight magnitude is a selectable numeric parameter.
A POSITA would have combined Nida with Micaelian and would have found it obvious to implement Nida’s list-prioritization scheme with numerical weights and to choose the particular absolute scale of those weights, including values of 1 or greater, as a matter of routine design choice. Nida already teaches that different lists may be prioritized according to model performance or business logic, while Micaelian teaches expressing such relative importance through adjustable numeric weights. Once a numerical weighting approach is adopted, selecting whether the weights are 0 to 1, 1 and above, 10 to 100, or another positive scale would have been a predictable matter of parameterization because the claimed threshold does not change the underlying ranking function and merely reflects one of many obvious ways to encode relative importance.
The limitation is therefore obvious under 35 U.S.C. 103. The claimed requirement that the weight have an absolute value of one or greater does not appear to add patentable significance beyond choosing a particular numeric expression for a known weighting parameter.
The limitation requires that, for a given client terminal and a given target-specific priority list, the same list-level weight is applied uniformly to the scores of all subjects appearing on that list.
Nida teaches the underlying list-and-score framework as explained above.
Micaelian teaches that missing weighting technique. Micaelian discloses weighted preference information including weights corresponding to search criteria and a corresponding plurality of weights signifying the relative importance of the search criteria, with results ranked based upon the weighted preference information. Micaelian further teaches that for each criterion, the normalized distance data is multiplied by its corresponding weight and accumulated to obtain a score for the alternative. It also explains that operation 180 accumulates the product of each normalized distance by its criteria weight (Col.10, ll. 30-45) and that wj is the weight for the jth criterion. Those passages teach uniform application of the same criterion weight across all alternatives scored under that criterion.
A POSITA would have combined Nida with Micaelian because Nida already prioritizes and combines multiple model-output lists, while Micaelian teaches a known numerical mechanism for expressing and applying relative importance during ranking. The specific modification would have been to assign one preselected weight to each Nida target-specific list for a respective client terminal and apply that same list weight to every subject score on that list before combining the weighted lists into the clinical queue. The result would have been predictable because uniform application of a single weight to all scores produced by the same ranking input preserves that input’s intended relative importance during combination. This is the use of a known weighting technique to improve a known prioritization method in the same way, yielding predictable results.
wherein each respective global weight of each respective priority list is adjustable to a different value by a different client terminal. The above limitation requires that different client terminals can assign different weight values to the same priority list, such that the same list may have one degree of priority for one client terminal and a different degree of priority for another.
Nida teaches the surrounding framework, but not the full limitation. Nida states that the output of each model may therefore comprise a list of members with associated scores in a given category and that the queue management module 210 may combine all of the lists into a master queue and generate clinical queues comprising sorted and combined lists of members. Nida para. 0043-0045. Nida also teaches that a model with a higher rate of positive cases may be prioritized over a model with a lower rate of positive cases and that a business user may decide that a high cost model should be prioritized over an inpatient model. Nida para. 0046. Nida further teaches client-terminal-specific queueing because the queue output module 220 may identify a specific client device 121 for transmitting a particular clinical queue and may distinguish queues according to a type of healthcare facility or provider. Nida para. 0047. Nida further states that healthcare providers at a facility specializing in mental health care may prefer clinical queues scored according to different rubrics than healthcare providers at a facility specializing rheumatology. Nida para. 0047. Thus, Nida teaches different client devices, different queue rubrics, and prioritization among different model-output lists.
However, Nida does not teach that each respective global weight of each respective priority list is adjustable to a different value by a different client terminal. Nida describes different rubrics and different queues for different providers, but it does not expressly disclose that each client terminal can adjust a numerical global weight for each priority list, nor that different client terminals can assign different values to the same list weight.
Micaelian teaches that missing feature. Micaelian states that the weighted preference generator can include a user interface which allows a human user to input preferences into the generator and that those preferences include the adjustment of weights with respect to the search criteria. Micaelian col. 3, lines 31-35. Micaelian further teaches that the weighted preference generator is preferably a client to the weighted preference data search engine and that the client can provide weighted preference information to the search engine. Micaelian col. 3, lines 28-30 and col. 4, lines 44-46. Micaelian also explains that the position of the indicator 110 along the slider bar 112 is translated into a numeric output, typically a normalized value between Zero and one, which is the weight for the criterion. Micaelian col. 9, lines 1-5. These passages teach client-side adjustment of numerical weights and thus permit different clients to assign different values to the same ranking input.
A POSITA would have combined Nida with Micaelian because Nida already teaches that different providers or facilities may prefer differently scored queues, but leaves unspecified the mechanism for implementing those different rubrics. Micaelian supplies that missing mechanism by teaching client-adjustable numerical weights used to express relative importance during ranking. The specific modification would have been to configure Nida’s client terminals so each terminal could set the weight value assigned to each target-specific priority list, and so different terminals could assign different values to the same list before the server combines the lists into a clinical queue. The result would have been predictable because varying client-selected weight values is a straightforward way to produce the different queue priorities that Nida already says different providers may prefer.
The cited teachings and rationale are sufficient to support obviousness of this limitation under 35 U.S.C. 103.
The above limitation requires that different client terminals can assign different weight values to the same priority list, such that the same list may have one degree of priority for one client terminal and a different degree of priority for another.
Nida teaches different client devices, different queue rubrics, and prioritization among different model-output lists.
However, Nida does not teach that each respective global weight of each respective priority list is adjustable to a different value by a different client terminal.
Micaelian teaches that missing feature. Micaelian states that the weighted preference generator can include a user interface which allows a human user to input preferences into the generator and that those preferences include the adjustment of weights with respect to the search criteria. Micaelian col. 3, lines 51-67. Micaelian further teaches that the weighted preference generator is preferably a client to the weighted preference data search engine and that the client can provide weighted preference information to the search engine (Col.5, ll. 45-67). Micaelian also explains that the position of the indicator 110 along the slider bar 112 is translated into a numeric output, typically a normalized value between Zero and one, which is the weight for the criterion (Col.8, ll. 19-39). These passages teach client-side adjustment of numerical weights and thus permit different clients to assign different values to the same ranking input.
A POSITA would have combined Nida with Micaelian because Nida already teaches that different providers or facilities may prefer differently scored queues, but leaves unspecified the mechanism for implementing those different rubrics. Micaelian supplies that missing mechanism by teaching client-adjustable numerical weights used to express relative importance during ranking. The specific modification would have been to configure Nida’s client terminals so each terminal could set the weight value assigned to each target-specific priority list, and so different terminals could assign different values to the same list before the server combines the lists into a clinical queue. The result would have been predictable because varying client-selected weight values is a straightforward way to produce the different queue priorities that Nida already says different providers may prefer.
transmitting the respective dynamically adjusted preselected global list weights of the respective client terminal to the server, wherein the server performs inference using the combined prioritization machine learning model using the different values of the preselected global list weights received from the plurality of client terminals. The above limitation requires that the client terminal send its adjusted list-weight values for the respective priority lists to the server, and that the server use those received client-specific weight values when performing inference with the combined prioritization machine learning model to generate the resulting queue.
Nidia teaches multiple model-output lists, server-side queue combination, client-specific queue delivery, and client-to-server transmission of queue-related user input.
However, Nida does not complete teach transmitting
Micaelian teaches that missing feature. Micaelian states that the weighted preference generator is preferably a client to the weighted preference data search engine (Col.2, ll. 45-67) and that the client can provide weighted preference information to the search engine Micaelian col. 5, lines 47-67; Micaelian further teaches that the user may perform the adjustment of weights with respect to the search criteria, and that the position of the indicator 110 along the slider bar 112 is translated into a numeric output ... which is the weight for the criterion Micaelian col. 8, lines 10-40. Micaelian also teaches that the server-side engine uses those provided weights to rank output because the weighted preference data search engine uses the weight of the preference data ... to provide an ordered result list based upon the weighted preference information. These disclosures teach client-adjusted weight values being sent to the ranking engine and then used there to produce ranked results.
A POSITA would have combined Nida with Micaelian because Nida already teaches the need for different client-specific scoring rubrics and already provides the server-side framework for combining multiple model-output lists into client-delivered queues, but Nida does not specify how those rubric differences are concretely represented and used at inference time. Micaelian supplies that exact missing mechanism by teaching client-adjusted numerical weights transmitted to the server-side engine and used by that engine in ranking. The specific modification would have been to configure each Nida client terminal to send its adjusted weight values for the respective target-specific lists to the server, and to configure Nida’s queue management module to use those received values when combining the model-output lists into the client’s queue. The result would have been predictable because it would implement Nida’s disclosed rubric variation using Micaelian’s known client-to-server weighting technique.
The cited teachings and rationale are sufficient to support obviousness of this limitation under 35 U.S.C. 103.
Claim 2. Nida in combination with Micaelian teaches, The system of claim 1, wherein respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority lists comprise a numerical value. (Nida See at least, 0043, 0046), Nida teaches the weights as numerical (score).
Claim 3.
Nida in combination with Micaelian teaches The system of claim 1, wherein respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list are selected according to a geographical location. (Nida See at least, 0047), Nida states that queues are generated based on geographical position.
Claim 4.
Nida in combination with Micaelian teaches The system of claim 1, wherein respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list are defined according to a set of rules. (Nida See at least, par.0046), Nida describes prioritizing models based on their performance ("higher rate of positive cases") and allows for "arbitrary business logic" to be used. Both of these are examples of "a set of rules" defining prioritization.
Claim 5.
Nida in combination with Micaelian teaches, The system of claim 1, wherein respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list are fixed within a defined range of values. (Micaelian See at least, column 8, line18 – line39), Micaelian, discloses a weight adjustment interface (column 7, lines 20-40) where properties, analogous to priority lists, are assigned weights. These weights, controlled by slider bars, indicate the magnitude of importance ("not important" to "very important") and are fixed within a defined range (normalized values between zero and one).
Claim 8.
Nida in combination with Micaelian teaches, The system of claim 1, wherein the combined prioritization component comprises aggregation code that when executed by a processor aggregates, for each subject of the plurality of subjects, the respective numerical scores of the plurality of specific priority lists into an aggregated score; (Nida See at least, par.0046, 0043, 0045), Nidia teaches, the queue management module 210 includes a set of instructions for sorting and combining multiple lists, which functions as the aggregation code. Each subject is assigned a plurality of numerical scores, where models generate scores for different categories such as risk, cost, and predicted length of stay. The system aggregates these scores and ranks individuals accordingly in a specialized combined list.
and generates the target combined priority list by ranking the plurality of subjects according to respective aggregated scores. (Nida See at least, 0045-0046), Nida describes a queue management module that generates a combined priority list (master queue) from multiple predictive model outputs and ranks individuals based on scores computed by those models.
Claim 9.
Nida in combination with Micaelian teaches, The system of claim 8, wherein each of the plurality of priority lists is assigned a respective weight, and wherein aggregating comprises, computing, for each respective subject a respective weighted score by multiplying each respective numerical score assigned to the respective subject on each respective priority list by the respective weight assigned to the respective list, and ranking the plurality of subjects according to respective weighted scores, the combined priority list comprising the ranked weighted scores. (Nida See at least, par.0045-0046) 0092, 0094, Nidia, describes combining multiple lists into a master queue and sorting individuals based on computed scores, which aligns with the claim of ranking individuals based on weighted scores to generate a combined priority list. Although explicit mention of multiplication is not provided, the sorting of individuals based on risk scores and prioritization based on business logic implies a computed ranking, which aligns with the computation of a weighted score. Additionally, the prior art describes assigning different priority levels to different models based on their performance or business logic considerations, implying respective weights to each priority list.
Claim 10.
Nida in combination with Micaelian teaches, The system of claim 1, wherein at least one of a plurality of the target specific priority lists and the target combined priority list are time correlated priority lists, denoting, for each respective rank within each respective time correlated priority list, a maximal recommended time interval for performing a corresponding medical intervention on the respective subject at the respective rank for reducing or preventing a target clinical outcome. (Nida See at least, par. 0054-0055, 0093-0094, 0043, 0045, 0003, 0047), Nidia, describes a clinical queue system that prioritizes subjects for healthcare intervention based on their risk scores, ensuring timely medical intervention. Since the system orders individuals according to urgency, risk scores, and required medical intervention, this meets the requirement of time-correlated priority lists with recommended intervention times to reduce medical risks
Claim 12.
Nida in combination with Micaelian teaches, The system of claim 1, wherein at least one of a plurality of the target specific priority lists and the target combined priority list are time correlated priority lists, wherein each respective rank within each respective time correlated priority list denotes a recommended maximal time interval for performing a corresponding medical intervention on the respective subject for an effective allocation of resources for performing the medical intervention based on an input of a schedule of availability of resources for performing each respective corresponding medical intervention for each of a plurality of time intervals. (Nida See at least, par. 0003, 0019, 0075,0082, 0043, 0045, 0047, 0056), Nida addresses time-correlated priority lists and the efficient allocation of resources. In addition, Nida, display the patient data in timelines and use schedule that are analogous to recommended maximal time interval.
Claim 13.
Nida in combination with Micaelian teaches, The system of claim 12, wherein at least one of each respective specific training dataset and a main training dataset further includes a schedule of availability of resources for performing each respective corresponding medical intervention for each of a plurality of time intervals, and the respective machine learning model is trained for generating a respective time correlated priority list in response to an input of a schedule of time correlated availability of the respective resource for performing the respective medical intervention. (Nida See at least, par. 0054-0055,0093-0094), Nida, describes a system where medical interventions are scheduled based on priority queues. The dashboard feature displays patient priority for intervention, considering availability and scheduling, which meets the requirement of including a resource schedule for interventions. It also features a machine learning model that prioritizes individuals by risk scores, creating clinical queues that dictate timing. Since these queues are updated in real-time, they correlate to available medical resources, anticipating the feature of time-correlated priority lists based on resource availability.
Claim 14.
Nida in combination with Micaelian teachers, The system of claim 1, wherein the respective specific training dataset further includes, for each respective sub-set of the plurality of subjects, an indication of risk of at least one of a respective clinical outcome and/or a diagnosis of a respective clinical diagnosis, wherein each respective specific machine learning model is for at least one of a respective clinical outcome and for the respective clinical diagnosis wherein each respective specific machine learning model is trained to generate the outcome of the respective target priority list of the sub-set of target subjects at risk for the at least one of clinical outcome and diagnosed with the respective clinical diagnosis, for prioritized at least one of evaluation and treatment for the at least one of respective clinical outcome, respective clinical diagnosis, and target clinical outcome by the respective specific medical intervention. (Nida See at least, par. 0043, 0083, 0045, 0093-0094), Nidia, describes models that generate scores related to "risk, cost, predicted length of stay", and that these models are used to generate "clinical queues". The system uses multiple models for different prediction "categories”.
Claim 15.
Nida in combination with Micaelian teachers, The system of claim 1, further comprising code for:
receiving a set of target EMRs of a plurality of target subjects; (Nida See at least, par. 0026 0043), Nida states the system works with "members", implying multiple individuals. The system must receive data on these members to function.
feeding the set of target EMRs into the plurality of specific machine learning models; (Nida See at least, par. 0026 0043), Nida states the models "process" data, which implies the data is fed into the models.
obtaining a plurality of specific priority lists as outcomes of the plurality of specific machine learning models; (Nida See at least, par. 0026 0043), Nida states that the output of each model is a "list of members".
feeding the plurality of specific priority lists into the combined prioritization machine learning model; (Nida See at least, par. 0045), Nida, states the queue management system combines "all of the lists.
and obtaining a combined priority list as an outcome of the combined prioritization machine learning model. (Nida See at least, 0045), Nida, states the system generates "clinical queues" which are "sorted and combined lists".
Claim 16. Nida in combination with Micaelian teachers, The system of claim 15, further comprising code for:
sequentially at least one of treating and evaluating the plurality of target subjects for a plurality of respective target clinical outcomes by the plurality of specific medical interventions according to a prioritized order defined by the plurality of specific priority lists; (Nida See at least, par. 0019, 0089, 0043, 0045), Nida describes a queue management system that evaluates subjects using predictive models, assigns risk scores for different healthcare outcomes, and prioritizes medical interventions accordingly. The system operates sequentially by processing individuals in order of their priority ranking and generates multiple lists based on predictive modeling outputs. Additionally, the queue management system combines these lists into a master queue for further prioritization. This process directly correlates to the claimed system’s function of sequentially treating and evaluating subjects using priority lists for different interventions.
and sequentially at least one of treating and evaluating the plurality of target subjects according to the prioritized order defined by the combined priority list. (Nida See at least, par. 0045), Nida, discloses that the queue management system consolidates multiple priority lists into a master queue, which then determines the order of medical interventions and evaluations. The system ensures that patients are treated and evaluated in a sequential manner based on the combined priority ranking.
Claim 25.
Nida in combination with Micaelian teachers, The system of claim 1, wherein each respective weight of each respective priority list is dynamically adjustable by a user via a user interface. (Nida See at least,, par. 0089, 0086, 0088)
Claim 26. (Note: for the amended limitations refer to claim 1 analysis)
Nida teachers, A system for prioritization of target subjects for at least one of treatment and evaluation, comprising
at least one processor of a server in communication with a plurality of client terminals over a network executing a code for:
accessing EMRs of a set of a plurality of subjects; (Nida See at least, par. 0093, 0026), Nida, describes a queue management system that accesses healthcare databases, which contain electronic medical records (EMRs) and medical claims data.
feeding the EMRs into each of a plurality of specific machine learning models; (Nida See at least, par. 0044, 0049), Nida, describes machine learning models that process EMR data, including medical claims, diagnoses, and prescriptions to generate predictions and assessments.
obtaining a plurality of specific priority lists as outcomes of the plurality of specific machine learning models, wherein each respective specific priority list includes a respective sub-set of the plurality of subjects scheduled in a prioritized sequence for at least one of treatment and evaluation for at least one of a respective target clinical outcome and by a respective specific medical intervention; (Nida See at least, par. 0043, 0045,0089) Nida, describes a system generating patient-specific priority lists, where each list contains a subset of patients ranked for medical interventions based on their risk assessment scores.
feeding the plurality of specific priority lists into an intermediate component that assigns a respective preselected global list weight to each respective target specific priority list, (Nida See at least, par. 0005, 0026, 0019, 0045 – 0046, 0089), Nida, teaches generating "a plurality of risk scores for each individual a plurality of individuals based on medical claims of each individual”. This is analogous to the "plurality of specific priority lists" of the Patent Under Examination. The prior art then transmits a "clinical queue comprising a list of individuals prioritized for healthcare intervention based on the plurality of risk scores" to a client device. While the term "intermediate component" is not explicitly used, the transmission of the clinical queue implies a processing step where the risk scores (priority lists) are used to generate the queue. This transmission can be seen as the "feeding" of the risk scores into a component that generates the clinical queue. The prior art describes a queue management module that combines and sorts lists from individual models, prioritizing them based on model performance. This is functionally equivalent to assigning a preselected weight to each list and computing weighted priority lists.
obtaining a plurality of weighted specific priority lists from the intermediate component; (Nida See at least, par. 0019), Nida discloses a plurality of weight for example risk, cost, potential length of hospitalization and so on.
feeding the plurality of weighted specific priority lists into combined prioritization machine learning model; (Nida See at least, par. 0019, 0043, 0045)
wherein the combined prioritization machine learning model is trained on a main training dataset that includes the plurality of weighted specific priority lists obtained as outcomes of the plurality of specific machine learning models assigned respective preselected global list weightss, labelled with a respective combined priority list of the set of the plurality of subjects for at least one of treatment and evaluation, (Nida See at least, par. 0059, 0093, 0043-0045,0049- 0050)
Nida describes the claimed invention. The system evaluates a model's performance, generates a list, and then retrains the model using feedback and indications of opened cases, which serves as a labeled training dataset derived from the model's outcomes.
wherein the plurality of specific machine learning models and the combined prioritization machine learning model are implemented as one or more or combination of the following architectures: neural network, logistic regression, decision tree, and boosting. (Nida See at least, par. 0043-0045, 0049, 0058-0059, 0069)
Nida describes the predictive models and prioritization models may include architectures such as artificial neural networks, logistic regression, decision trees, and boosting models, among others.
obtaining a combined priority list as an outcome of the combined prioritization machine learning model, wherein the combined priority list is of the plurality of subjects indicating priority for at least one of evaluation and treatment, wherein the plurality of specific machine learning models and the combined prioritization component generate a respective numerical score for each respective subject, and the specific priority lists and the combined priority list are created by ranking subjects according to respective numerical scores, wherein the combined prioritization machine learning model computes, for each respective subject of the plurality of target subjects, a respective weighted score by multiplying each respective numerical score assigned to the respective subject on each respective priority list by the respective global list weight assigned to the respective list. (Nida See at least, par. 0019, 0026, 0043, 0045-0046)
Nida's queue management system uses machine learning to generate prioritized clinical queues by combining and sorting patient lists from a master queue based on predictive models and weighted scores.
a plurality of client terminals in communication with the server over the network, each client terminal including at least one processor executing a code for:presenting on each respective display of each respective client terminal of the of the plurality of client terminals, a graphical user interface (GUI) configured for dynamically adiusting the respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list to respective values,wherein different GUIs dynamically adjust the preselected global list weights to different values,wherein each respective preselected global list weight indicates a magnitude of prioritization of the respective target specific priority list for the respective client terminal,wherein each respective preselected global list weight comprises an absolute value of one or greater,wherein the same respective preselected global list weight is applied to all scores of all subjects of the same respective target specificity priority list for the respective client terminal,wherein each respective global weight of each respective priority list is adjustable to a different value by a different client terminal;transmitting the respective dynamically adjusted preselected global list weights of the respective client terminal to the server,wherein the server performs inference using the combined prioritization machine learning model using the different values of the preselected global list weights received from the plurality of client terminals; andreceiving from the server and presenting within the GUI on each respective display, for each respective subject, at least one of:
for each respective subject, at least one of: the respective numerical score computed by the combined prioritization machine learning model, and a ranking of the respective subject on the combined priority list.( Nida See at least, par. 0005, 0019, 0065, 0068, fig. 12, 0096)
Nida discloses a GUI that presents a combined, prioritized queue showing each subject’s position in the ordered list produced by the combined predictive models.
Nida, teaches a combine output of priority list using artificial intelligence models based on weights.
computed according to the respective dynamically adjusted preselected global list weights,wherein the server sends different numerical scores and/or rankings to corresponding client terminals according to the dynamically adjusted preselected global weights.
35 USC 103 Rational:
Nida disclosed all limitation, however weighted score is presented, but the method of calculation is not specified neither adjusting the specific numerical magnitude.
Method of Calculation
Micaelian teaches, the weighted summation method, where each alternative’s score is multiplied by a criteria weight and accumulated column 9, line 30 – line 41. A person skilled in the art would seek to improve the prioritization accuracy of clinical queues by incorporating a weighting system that accounts for the importance of different risk factors. Incorporating weighted summation from Micaelian into the clinical queue prioritization from Nida results in a more refined and adjustable ranking method.
Specific Numerical Magnitude
Nida teaches a system for combining multiple clinical priority lists into a master queue, explicitly stating that prioritization (weighting) occurs based on user-defined logic: "any arbitrary business logic may be use[d] to prioritize and combine lists. For example, a business user may decide..." (Nida [0046]). This establishes the motivation and framework for customizable weighting by different users. To the extent that Nida does not explicitly detail a user interface for adjusting the specific numerical magnitude of these weights, this feature is explicitly taught by Micaelian. Micaelian discloses a weight adjustment interface (e.g., slider bars) allowing users to assign numerical weights representing the magnitude of importance, normalized between zero and one (Micaelian, Col. 7, lines 20-40; Col. 8, lines 18-39). It would have been obvious to one of ordinary skill in the art to implement Nida's customizable prioritization logic using the known technique of user-adjustable magnitude weights taught by Micaelian to achieve a predictable result: a refined and customizable method for combining clinical priority lists.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the teachings of Nida with Micaelian because both references address the shared challenge of prioritizing and ranking data based on multiple, customizable criteria. Nida establishes the need for user-defined prioritization in clinical queues (Nida [0046]), and Micaelian provides a known methodology for implementing user-adjustable weighted prioritization (Micaelian, Col. 2, lines 34-39). A POSITA would recognize Micaelian's weighting interface as a concrete method to implement Nida's generalized “arbitrary business logic” (Nida [0046]).
Claim 45.
Nida teaches, A system for dynamic prioritization of target subjects for at least one of evaluation and treatment, comprising:
at least one processor of a server in communication with a plurality of client terminals over a network executing a code for: (Nida See at least, abstract, 0023), Nida discloses a system that uses predictive models, has a logic subsystem including one or more physical devices configured to execute instructions that are part of one or more applications, services, programs, routines, libraries, objects, components, data structures, or other logical constructs 1 to generate risk scores, transmit a clinical queue to a healthcare provider, the clinical queue including a list of individuals prioritized for healthcare intervention.
accessing a plurality of specific priority lists of a respective sub-set of a plurality of subjects scheduled in a prioritized sequence for at least one of treatment and evaluation for at least one of a respective target clinical outcome and by a respective specific medical intervention,
wherein the plurality of specific priority lists are generated by the plurality of specific machine learning models trained according to Claim 1 (Refer to Nida in combination with Micaelian claim 1 ), wherein each subject in each of the plurality of specific priority lists is associated with a respective numerical score indicative of risk of a respective clinical outcome; (Nida See at least, par. 0089, 0005, 0019, 0093)
computing, for each respective subject of the plurality of subjects, a resfpective weighted score by multiplying each respective numerical score assigned to the respective subject on each respective priority list by the respective global list weight assigned to the respective list; (Nida See at least, par. 0045-0046), The prior art explicitly states that models can be prioritized over others, meaning their outputs contribute more to the final result. This directly corresponds to the weight indicating the magnitude of prioritization.
And generating a combined priority list by ranking the plurality of subjects according to respective weighted scores, wherein the combined priority list comprising the ranked weighted scores. (Nida See at least, 0043, 0045,0096, 0019), Nida describes a method where subjects (individuals) are evaluated and ranked based on different weighted scores derived from predictive models. The combined priority list (clinical queue) is formed by integrating these ranked scores, ensuring that individuals are prioritized for healthcare intervention based on their overall risk and other relevant factors. This aligns with the document's method of generating clinical queues by sorting and combining lists according to various scores, thus fulfilling the limitation of ranking subjects by weighted scores to create a combined priority list.
wherein the combined priority list is generated by the combined prioritization machine learning model according to Claim 1. (Refer to Claim 1 analysis)
In summary, Nidia teaches all the elements with the exceptions of Multiplying operation, that it is teaches by Micaelian, in the abstract and figure 10. It would have been obvious to one of ordinary skill in the art, to combine Nida regarding multiple priority list to addressing the inability for user or automated clients of a database search engine to specify preferences or “weights”. Refer to Micaelian Column 2, Lines 20-28
a plurality of client terminals in communication with the server over the network, each client terminal including at least one processor executing a code for: presenting on each respective display of each respective client terminal of the of the plurality of client terminals, within-a graphical user interface (GUI configured for dynamically adjusting the respective preselected global list weights indicating the magnitude of prioritization of the respective target specific priority list to respective values. wherein different GUIs dynamically adjust the preselected global list weights to different values. wherein each respective preselected global list weight indicates a magnitude of prioritization of the respective target specific priority list for the respective client terminal, wherein each respective preselected global list weight comprises an absolute value of one or greater. wherein the same respective preselected global list weight is applied to all scores of all subjects of the same respective target specificity priority list for the respective client terminal, wherein each respective global weight of each respective priority list is adjustable to a different value by a different client terminal: transmitting the respective dynamically adjusted preselected global list weights of the respective client terminal to the server, wherein the server performs inference using the combined prioritization machine learning model using the different values of the preselected global list weights received from the plurality of client terminals; and receiving from the server and presenting within the GUI on each respective display, for each respective subject, at least one of: the respective numerical score computed by the combined prioritization machine learning model, and a ranking of the respective subject on the combined priority list, computed according to the respective dynamically adjusted preselected global list weights. wherein the server sends different numerical scores and/or rankings to corresponding client terminals according to the dynamically adjusted preselected global weights.(Refer to Claim 1 analysis to this part of claim 45, as is substantially same than in claim 1)
Claim(s) 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over US20200279641A1- Dean Nida in combination with US6714929B1-Fadi Victor Micaelian, and further in view of US20190034591A1-Alexander Mossin
Claim 20.
Nida in combination with Micaelian teachers, The system of claim 1, further comprising code for generating instructions for at least one of treating and evaluation target subjects by a main medical intervention according to the target combined priority list. (Nida See at least, par. 0042-0043, 0048, 0019, 0045, 0089), Nida, discloses an AI-generated clinical queues that rank patients based on risk, and merging into a single risk-based prioritization queue. However, does not explicitly teaches, by a main medical intervention, that it is teaches by Mossin in par. 0129, since the system recommends primary interventions based on predictive analysis. It would have been obvious to combine Missing with Nidia in combination with MIcaelian driven by the need to integrate AI-driven patient prioritization with real-time predictive clinical decision support for optimized treatment planning. See par. 0006 Mossin
Conclusion
Relevant Prior Art:
US 20150324531 A1
[0012] To better address one or more of these concerns, in a first aspect of the invention a system for scheduling a sequence of examinations for a patient in a hospital is presented that includes a first data base comprising data relating to a specific type of examination, the data including time consumption, and the data relating usage of resources associated with the specific type of examination, a second data base comprising data relating a patient to an associated list of examinations that the patient is to participate in, a scheduling unit operatively coupled to the first data base and the second data base, wherein the scheduling unit is arranged for scheduling the sequence of examinations for a group of patients based on their associated list of examinations and data from the first data base under consideration of a maximum total time used on a specific patient, the sequence including information relating to time and type of examination. See also figure 6 and figure 7
US 20210295984 A1
[0036] The workflow schedule optimizer can be embodied as an add-on package (e.g., OptTek-OptQuest™, available at https://www.opttek.com) to the simulator, and operates to adjust aspects of the simulated workflow schedule in accordance with a set of business constraints/restrictions/priorities in order to generate schedule adjustments. For example, if a laboratory worker calls in sick, the simulator may estimate that this will lead to afternoon patients being delayed by delay times that accumulate over the course of the day. The workflow schedule optimizer then may simulate hypothetical workflow schedules for various candidate adjustments or combinations of adjustments, such as shifting times of adjustable patient appointments (e.g. in-patients), cancelling one or more patients, adding a temporary worker, contacting remote personnel to help in maintaining a workflow schedule, providing overtime to laboratory personnel in order to extend the work day, and/or so forth. Each such hypothetical simulation can be scored using one or more Key Performance Indicators (KPIs). The system may automatically choose one or more adjustments scoring highest in terms of KPIs, or may propose the highest scoring adjustment(s) to laboratory personnel via the user interface for user selection. see also fi. 2
CA 2763209 A1
Abstract
In a method and system of controlling patient care logistics, a computer determines for each of a number of user devices a unique priority sorted list of queue tasks on the basis of global criterion. Each user device is dispatched the unique priority sorted list of queue tasks determined for the user device. In response to receiving a change in at least one global criterion, the computer determines for each user device either an amendment to the unique priority sorted list of queue tasks for the user of the user device or a new unique priority sorted list of queue tasks for the user of the user device, and then dispatches the unique priority sorted list of queue to the user device. See also claim 1
US 20090164236 A1
Abstract
The claimed subject matter provides a system and/or a method that facilitates scheduling an incoming patient appointment for a medical facility. A medical facility can provide healthcare to a patient, wherein the medical facility can utilize a schedule with an available time slot to assign an appointment to a patient. A match component can evaluate a portion of transportation data to select a patient to which an appointment on the schedule is allotted. A dynamic schedule component can automatically adjust the schedule based upon the evaluation.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOSHUA RUIZ whose telephone number is (571)272-0409. The examiner can normally be reached M-F 8:00-4:00 ET.
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, Shahid Merchant can be reached at 571-270-1360. 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.
/JOSHUA DAMIAN RUIZ/Examiner, Art Unit 3684
/Shahid Merchant/Supervisory Patent Examiner, Art Unit 3684
Prosecution Insights