Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
This office action is in response to an amendment received on 6/11/26.
Claims 1,9, 16 are amended.
Claims 1-20 are pending.
RESPONSE TO ARGUMENTS
Applicant argues#1
The Office Action at page 8 responds to Applicant's argument that the multi-variable approach constitutes a technical improvement by stating that "applicant is using a machine learning model implementing a multivariable approach, which is a commonly used machine learning technique when there are complex relationships between the data." The Office Action further states at page 9 that "the SHAP model or another suitable model recited in paras 35 and 51, is a machine learning explanatory model, that is recited at a high level of generality, operating in its ordinary capacity and as such is being used as a tool to implement the steps of the identified abstract idea."
Applicant respectfully traverses this characterization. The Office Action conflates the general concept of multi-variable machine learning with the specific technical mechanism recited in the pending claims. The pending claims, as amended, do not merely recite the use of a multi-variable approach in the abstract. Rather, the claims specifically recite that the one or more machine learning models generate one or more decision trees to determine the predicted severity output, that an explanatory model applied to the one or more decision trees generates an explainer value for each of the one or more claim variables, that each explainer value represents the relative contribution of the respective claim variable to the predicted severity output, and that the sum of the explainer values for the one or more claim variables is equivalent to the predicted severity output. This is not a generic recitation of multi-variable machine learning. It is a specific technical description of how the machine learning models operate to produce their output, including the internal mechanism by which the contribution of each variable to the output is computed and verified. Furthermore, the Office Action's assertion that the SHAP model or another suitable model is recited at a high level of generality is not accurate with respect to the pending claims as amended. The amended claims recite the specific functional role of the explanatory model, namely generating explainer values representing the relative contribution of each claim variable to the predicted severity output by identifying decisions made at each juncture of the decision trees, with the sum of those explainer values being equivalent to the predicted output. This is precisely the kind of specific technical mechanism that the Federal Circuit found patent eligible in McRO, where the court held that claims directed to a specific rule-based process that operated on defined inputs to produce a defined output were not abstract, distinguishing them from claims directed to the general concept of the process. 837 F.3d at 1314-1316.
Examiner Response
Examiner respectfully disagrees.
The decision tree that is recited in claim 1, is reciting a mental process (a human being with pen and paper can map out steps to decide on a choice).
The explainer values that are recited in claim 1, is reciting a Mathematical Concepts (mathematical calculations), see spec para 36.
Therefore the claimed invention is reciting both a Mental process and mathematical calculations.
The rejection is maintained.
Applicant argues#2
The Office Action at page 11-12 responds to Applicant's argument regarding Ex parte Desjardins by stating that "applicant's claims do not improve technology; the underlying technology remains unaffected by the claims" and that "applicant is addressing a business problem (steps for determining damage severity based on claim data for multiple time periods) with a business solution" and that "applicant is merely using existing technology (for its intended purpose) to implement the business solution."
Applicant respectfully traverses this characterization on multiple grounds. First, the characterization of the claimed invention as a business solution to a business problem is an impermissibly high-level characterization of the claims that ignores the specific technical limitations recited therein. The 2019 Guidance expressly instructs that claims must not be evaluated at such a high level of generality that specific technical features are subsumed within the characterization of the abstract idea. See MPEP § 2106.04(d). When the claims are considered at the appropriate level of specificity, the focus is not on the general concept of determining insurance damage severity but on a specific technical implementation comprising decision-tree-based machine learning models with an applied explanatory model generating per- variable explainer values, line-identifier-level aggregation of those explainer values into a single combined input processed simultaneously, iterative training calibrated to a tolerance threshold, and a two-step mathematical correction of percent impact values that accounts for the difference between model-predicted and actual severity in accordance with the equation I2 equals Il multiplied by P2 divided by P1, as described at paragraphs [0056] and [0057] of the specification.
Examiner Response
Examiner respectfully disagrees.
The limitations (determine percent impact values for respective variables of the first set of variables and the second set of variables, wherein determining a respective percent impact value comprises determining a difference between a first explainer value associated with a claim variable from the first period and a second explainer value associated with the corresponding claim variable from the second period, and dividing the difference by the predicted item severity corresponding to that claim variable within the second period; correct each percent impact value by multiplying the respective percent impact value by a ratio of a second percent change to a first percent change, wherein the first percent change is between the first plurality of average explainer values and the second plurality of average explainer values, and the second percent change is between the first plurality of average explainer values and a plurality of actual severity values corresponding to the second period) is reciting a Mathematical concept (mathematical calculations).
The rejection is maintained.
Applicant argues#3
Second, the assertion that the underlying technology remains unaffected by the claims is factually incorrect. The specification at paragraph [0016] identifies a specific deficiency in conventional machine learning approaches to severity assessment, namely that conventional systems use a single-variable approach in which the impact of each variable is determined separately, resulting in large amounts of data for each individual variable. The specification at paragraph [0017] identifies the claimed multi-variable approach as an improvement over those conventional systems. The specification at paragraph [0035] describes the specific mechanism by which the machine learning models operate differently from conventional models, namely through the generation of decision trees and the application of an explanatory model that generates per-variable explainer values whose sum is equivalent to the model output. The specification at paragraphs [0051] and [0052] describe the specific aggregation mechanism by which line-level explainer values are summed to produce a single claim-level combined input. The specification at paragraphs [0056] and [0057] describe the specific correction mechanism by which the percent impact values produced by the model are corrected to account for known model inaccuracies. These are not descriptions of a business process implemented on a generic computer. They are descriptions of a specific technical mechanism by which the machine learning models themselves operate differently from conventional approaches.
Examiner Response
Examiner respectfully disagrees.
This argument has been addressed above with respect to Applicant argues#2 above.
The rejection is maintained.
Applicant argues#4
Applicant respectfully traverses this distinction. As demonstrated above, the specification of the instant application expressly identifies improvements to how the machine learning models themselves operate at paragraphs [0016], [0017], [0035], [0051], [0052], [0056], and [0057]. The amended claims now directly reflect those improvements by reciting the specific decision tree and explanatory model mechanism, the line-identifier aggregation process, and the mathematical correction formula. The factual premise of the Office Action's distinction of Ex parte Desjardins is therefore no longer accurate with respect to the pending amended claims. Applicant maintains Ex parte Desjardins as supporting authority for the proposition that claims rooted in a specific improvement to model operation are not directed to a judicial exception, and notes that the Appeals Review Panel's caution against evaluating claims at an impermissibly high level of generality is directly applicable to the Office Action's characterization of the instant claims.
Examiner Response
Examiner respectfully disagrees.
Examiner reproduces the spec paras that applicant refers to below:
[0016] In conventional claims processing systems, item damage severity is determined retroactively-that is, when all factors that impact the item damage severity are fully known. Severity is also conventionally analyzed using a single-variable approach, where the impact of each variable is determined separately from other variables. Conventional severity investigations therefore result in large amounts of data for each individual variable, and, in some instances, may be inaccurate due to the limited single variable scope.
[0017] Accordingly, the systems, methods, and computer-executable media described herein provide an improved computing system for determining severity based on a multi-variable approach. The improved computing systems advantageously predict severity based on claims data such that severity for claims from a first time period can be predicted, rather than determined retroactively. Additionally, the systems, methods, and computer-executable media described herein provide an improved user interface that advantageously provides severity data. The improved user interface may reduce the amount of data transmissions necessary for a user to understand a determined severity, for example, by reducing the number of graphics (e.g., graphs, tables, text, etc.) needed to visually represent the determined severity. Further, the improved user interface advantageously filters and sorts the severity data such that relatively more relevant severity data is presented before and/or instead of relatively less relevant severity data. For example, relatively less relevant (e.g., lower magnitude) severity values may be automatically grouped into an "other" category and displayed as a single graphical feature. Thus the improved user interface provides at least one specific improvement over prior systems, for example, by reducing the number of graphical elements needed to understandably convey severity data. Additionally, the systems, methods, and computer-executable media described herein embody a self-correcting predictive system that is periodically re-trained using current data such that the accuracy of predictions for item damage severity is improved over time. The item damage severity modeling circuit 126 is structured to store computer- executable instructions embodying one or more machine learning models. The one or more machine learning models are configured to generate one or more statistical models of damage severity. The item damage severity modeling circuit 126 may be structured to train the one or more machine learning models based on the claims information and the severity information such that the one or more machine learning models outputs and/or determines a predicted severity. As used herein "predicted severity" is severity that is estimated or predicted, using one or more statistical methods, machine learning algorithms, and the like, by estimating the factors that are not fully known when damage is reported. For example, the one or more machine learning models may be trained using training data that includes claims data (e.g., stored at the claims database 134 or at the claims database 172) and actual severity data stored at the severity database 132. In some embodiments, the actual severity data may be provided by the item damage severity aggregation circuit 124. The one or more machine learning models are trained to generate predicted severity based on the training data. In some embodiments, the one or more machine learning models generate decision trees to output and/or determine predicted severity based on input claim data. For example, the item damage severity modeling circuit 126 may receive claim data and identify (e.g., parse) one or more claim variables from the received claim data. The item damage severity modeling circuit 126 may utilize the one or more trained machine learning models to determine and/or output the predicted severity. As briefly described above, the one or more machine learning models may include a machine learning explanatory model (e.g., SHAP or another suitable model). Accordingly, the item damage severity modeling circuit 126 may utilize the machine learning explanatory model with the one or more machine learning models to output and/or determine a base rate of expected severity and/or explainer values. For example, the machine learning explanatory model (e.g., SHAP or another suitable model) may identify the decisions made at the one or more decision trees and generate explainer values representing calculations performed at each decision juncture. The explainer values correspond to a claim variable of the claim data input into the one or more machine learning models. Specifically, the machine learning explanatory model generates explanatory values for each claim variable in the one or more decision trees. A sum of the explanatory values is equivalent to the output (e.g., the predicted severity). The base rate of severity is output and/or determined by the machine learning explanatory model by calculating an average actual severity of the training dataset.
[0051] Referring to the method 300 in more detail, at step 302, the provider computing system 110 generates an explainer value for each input variable of each claim received in a predetermined time period. The explainer values may be generated based on a relative impact each claim variable has on the total item damage severity. The explainer values may be generated using a machine learning explanatory model, such as SHAP. For example, the one or more machine learning models may generate one or more decision trees to arrive at an output. The provider computing system 110 and/or one or more components thereof may utilize the machine learning explanatory model with the one or more machine learning models. The machine learning explanatory model may identify the decisions made at the one or more decision trees and generate explanatory values representing calculations performed at each decision juncture. The explanatory values correspond to the claim variables of the claim data input into the one or more machine learning models. In the embodiments described herein, the one or more machine learning models generate decision trees to output and/or determine a predicted severity based on one or more claim variable inputs. The machine learning explanatory model generates explanatory values for each claim variable in the one or more decision trees, and a sum of the explanatory values is equivalent to the output (e.g., the predicted severity). At step 304, the provider computing system 110 aggregates the explainer values for each claim. The explainer values may be aggregated for a set of claims. For example, the explainer value for one type of claim variable is added up resulting in a total item damage severity for a single claim variable across the set of claims. In some embodiments, the provider computing system 110 sums explainer values for one or more line IDs to the claim level. For example, a single claim may have one or more line IDs and/or the claim may include more than one insured item. Accordingly, one or more of the line IDs may be related to a first insured item, a second insured item, and so on. The provider computing system 110 may sum the explainer values for each of the one or more line IDs of a claim. Accordingly, when the claim is a multi- coverage claim (e.g., a claim having more than one line ID and/or a claim related to more than one insured item) the line IDs are aggregated such that each claim is associated with a single, aggregated value. This process may be repeated for some or all of the claim variables. In various embodiments, the set of claims can be aggregated according to the FNOL date, loss date, insured item type, make and/or model, geographical location of loss (e.g., GPS coordinates, zip code, etc.), or any suitable combination thereof. In an example embodiment, the set of claims is aggregated according to a claim identifier (e.g., a claim number, a claim ID, etc.).
[0052] At step 306, the provider computing system 110 averages the aggregated explainer values. The average is calculated by multiplying a relative frequency (e.g., percent occurrence of each claim variable in the claims within the predetermined time period) of a claim variable by the corresponding aggregated explainer value (e.g., for the same claim variable). That is, an aggregated explainer value for a first claim variable (X₁) is multiplied by the percent occurrence of that claim variable (Y₁) within the predetermined time period (e.g., within a week, a month, a quarter, a year, etc.). For example, a first claim variable may be a type of vehicle where X₁ is an aggregated explainer value for the vehicle type and where Y₁ is a percentage of claims that include the vehicle type. The average explainer value is calculated as the product of X₁ and Y₁. In some embodiments, the average explainer value may be calculated for each claim variable of a plurality of claims within the predetermined time period. In some embodiments, the average explainer value may be calculated for at least one claim variable for the plurality of claims within the predetermined time period.
[0056] The provider computing system 110 may correct the percent impact because, in some embodiments, there may be a difference between the predicted severity in a time period (e.g., the first time period) and the actual severity in the same time period. The difference may be due to inaccuracies of the machine learning model(s) (including machine learning explanatory model(s)) used to generate the explainer values, predicted severity, etc. To correct the results, the item damage severity modeling circuit 126 calculates a percent change between the predicted severity of the first time period (S1) and the predicted severity of the second time period (S2) resulting in a predicted severity percent change (P1). The item damage severity modeling circuit 126 then calculates a percent change between the actual severity of the second time period (A2) and the predicted severity of the first time period (S1) resulting in an actual severity percent change (P2). The item damage severity modeling circuit 126 calculates the percent impact due to each claim variable based on the actual severity of the second time period (12) by multiplying the percent impact due to each claim variable based on the predicted severity (I1) by the actual severity percent change (P2) and dividing by the predicted severity percent change (P1). The result is a percent impact due to each claim variable based on the actual severity of the second time period (I2), and may be output as the output severity data packet. An example equation (2) is shown below.
[0057] I2 = 11(P²/p₁)
As can be seen from the specification the specific decision tree (is reciting a mental process) and the mathematical correction formula (is reciting a mathematical calculation), which are both abstract ideas.
The explanatory model mechanism is recited at a high level of generalty and is being used as a tool to implement the steps of the identified abstract idea.
The “the line-identifier aggregation process” as disclosed in the specification in para 51 is part of the identified abstract idea (para 51 states that the line identifiers ID is a line of a claim such as an item”.
Examiner maintains that the claims are unlike the claims in the Desjardins decision. The machine learning model recited in claim 1, does not improve the machine learning model.
The machine learning model is recited at a high level of generality, operating in its ordinary capacity and is being used as a tool to implement the steps of the identified abstract idea.
The rejection is maintained.
Applicant agues#5
The Office Action characterizes the claims as directed to "steps for determining damage severity based on claim data for multiple time periods." This characterization is made at an impermissibly high level of generality that strips the claims of all specific technical content. Under the 2019 Guidance and McRO, the abstract idea characterization must be made at a level of specificity that accurately reflects what the claims are actually directed to, not at a level that reduces all claims in a technical field to a single general concept. See MPEP § 2106.04(d); McRO, 837 F.3d at 1313. The Office Action's characterization would, if accepted, render patent ineligible any claim directed to machine learning-based damage assessment regardless of its specific technical content, which is precisely the outcome the Appeals Review Panel in Ex parte Desjardins cautioned against.
Examiner Response
Examiner respectfully disagrees.
All the claim elements have been evaluated, see the section 101 rejection below and are reciting the identified abstract idea.
The rejection is maintained.
Applicant argues#6
The Office Action's Step 2A Prong 2 analysis at pages 18-19 evaluates the additional claim elements individually and concludes that each is recited at a high level of generality. The Office Action does not conduct a meaningful analysis of whether the ordered combination of additional elements, considered as a whole, integrates the alleged abstract idea into a practical application. This is a legal error. Under the 2019 Guidance, the Prong 2 analysis must consider whether the claim as a whole integrates the judicial exception into a practical application, and the ordered combination of elements may demonstrate integration even if individual elements do not. See MPEP § 2106.04(d)(II).
Examiner Response
Examiner respectfully disagrees.
The section 101 rejection did consider the combination of elements when determining under step 2a prong 2 that there was no integration into a practical application (see page 19 of the Non-Final rejection mailed on 3/11/26; “Accordingly, these additional elements, when considered separately and as an ordered combination, do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea”.
Therefore, the claims were evaluated as a whole (the way that the additional elements interact with the identified abstract idea) under step 2a prong 2.
The rejection is maintained.
Applicant argues#7
C.2.b Legal Error 3: Unsupported Factual Assertion That Additional Elements Are
Routine and Conventional
The Office Action asserts at pages 18-19 that the machine learning models, processor, memory, computing device, user interface, iterative training, and multi-variable aggregation are all recited at a high level of generality and operate in their ordinary capacity. To the extent this assertion is intended to support a finding that these elements are well-understood, routine, and conventional, it is unsupported by any evidence. Under Berkheimer v. HP Inc., 881 F.3d1360 (Fed. Cir. 2018), and the USPTO's subsequent Berkheimer memorandum, a finding that additional elements are well-understood, routine, and conventional must be supported by evidence, such as a citation to the specification, the prosecution history, prior art, or a reasoned statement taking official notice. The Office Action provides none of these. The bare assertion that elements operate in their ordinary capacity does not satisfy the evidentiary requirement established in Berkheimer.
Examiner Response
Examiner respectfully disagrees.
Applicant misapprehends when a Berkheimer analysis is required under current examination policy. Simply put, Examiner is not required under current Examination policy to evaluate under Step 2B, whether additional elements constitute “well-understood, routine, and conventional activities,” [“WURC activities”] unless an additional element(s) were found to be insignificant extra-solution activity in Step 2A, Prong 2. MPEP § 2106.05(d)(I). Here, the condition precedent was not met and the Non-Final Office Action determined the additional elements were no more than mere instructions to apply the abstract idea exception using a computer. MPEP § 2106.05(f). Thus, Examiner was not required to determine a Berkheimer analysis. MPEP § 2106.05(d)(I). (See Section 101 rejection below).
The rejection is maintained.
Applicant argues#8
The Office Action at pages 16-18 characterizes the claimed steps as performable mentally as "a concept performed mentally by a human with pen and paper." This characterization is legally unsupportable with respect to the pending amended claims. The amended claims recite, among other things, generating one or more decision trees (can be done by hand by a human) through iterative training of machine learning models to a predetermined tolerance threshold, applying an explanatory model to those decision trees to generate per-variable explainer values, summing line-identifier-level explainer values to produce a single claim-level aggregated value, and correcting percent impact values by multiplying by the ratio of an actual severity percent change to a predicted severity percent change in accordance with the equation at paragraph [0057]. These operations are not performable mentally or with pen and paper. The mental process grouping of abstract ideas is limited to concepts that can practically be performed in the human mind. See MPEP § 2106.04(a)(2). The complexity and computational nature of the recited operations place them well outside that category.
Examiner Response
Examiner respectfully disagrees.
This argument has been addressed above with respect to Applicant argues#2,4 above.
The rejection is maintained.
Applicant argues#9
When the claims are considered at the appropriate level of specificity, the focus of the pending claims is not the general concept of determining insurance damage severity. Rather, the claims are directed to a specific computing system and corresponding method and media that implement the following defined technical mechanism:
First, the claims recite a machine learning architecture in which one or more machine learning models generate one or more decision trees to determine a predicted severity output, and in which an explanatory model applied to those decision trees generates an explainer value for each claim variable, wherein each explainer value represents the relative contribution of that variable to the predicted severity output and the sum of all explainer values is equivalent to the predicted severity output. This is not a generic recitation of machine learning. It is a specific description of a defined model architecture and output generation mechanism, as described at paragraphs [0035] and [0051] of the specification.
Second, the claims recite a specific aggregation mechanism in which explainer values for one or more line identifiers associated with each claim are summed to produce a single aggregated claim-level value regardless of the number of coverage lines or insured items associated with the claim. This specific data structure operation, described at paragraphs [0051] and [0052] of the specification, is what enables the simultaneous processing of the combined input through the machine learning models, thereby eliminating the separate per-variable data storage and processing required by conventional single-variable approaches as described at paragraph [0016] of the specification.
Third, the claims recite an iterative training process in which the machine learning models are trained until they can predict item damage severities to within a predetermined tolerance threshold of an actual item damage severity, creating a self-calibrating model architecture as described at paragraphs [0017] and [0035] of the specification.
Fourth, the claims recite a specific two-step mathematical process for determining and correcting percent impact values. The determination step comprises computing the difference between a first period and second period explainer value for each corresponding claim variable and dividing by the predicted item severity for that variable within the second period. The correction step comprises multiplying each percent impact value by the ratio of the actual severity percent change to the predicted severity percent change, in accordance with the equation I2 equals Il multiplied by P2 divided by P1 described at paragraphs [0056] and [0057] of the specification, thereby producing corrected percent impact values that account for known inaccuracies in the machine learning model's predictions.
Fifth, the claims recite a specific user interface that displays the corrected percent impact values as selectable features, enabling a user to interact with the severity data in a manner that reflects the corrected, model-adjusted impact of each variable on the predicted damage severity.
The pending claims are therefore not directed to an abstract idea under Step 2A, Prong 1 of the 2019 Guidance.
The first improvement is the elimination of per-variable data storage and processing through the specific aggregation and simultaneous processing mechanism. As described at paragraph [0016] of the specification, conventional single-variable approaches require separate data storage for each variable, separate processing operations for each variable, and separate network transmissions of individual results. The pending claims recite a specific mechanism that overcomes this deficiency by aggregating claim variables, specifically by summing line- identifier-level explainer values into a single claim-level combined input, and processing that combined input simultaneously through the machine learning models to generate a single predicted severity output. The result is a direct, measurable reduction in data storage and network transmission requirements, as explicitly recited in the claims. This is precisely the kind of improvement to computer functionality that the Federal Circuit recognized as patent eligible in Enfish, where a specific modification to a data structure improved memory efficiency, and in Visual Memory LLC V. NVIDIA Corp., 867 F.3d 1253 (Fed. Cir. 2017), where a specific cache configuration improved computer performance.
The second improvement is the self-calibrating iterative training mechanism. The claims recite iterative training of the machine learning models until the models can predict item damage severities to within a predetermined tolerance threshold of an actual item damage severity. As described at paragraph [0017] of the specification, this creates a self-correcting predictive system that is periodically retrained using current data such that the accuracy of predictions improves over time. This is a specific improvement to how the machine learning models themselves operate, not merely a description of a desired result. The parallel to the continual-learning logic recognized in Ex parte Desjardins as constituting an improvement to machine learning model operation is direct.
The third improvement is the mathematical correction mechanism that directly compensates for known machine learning model inaccuracies. As described at paragraph [0056] of the specification, there may be a difference between the predicted severity in a time period and the actual severity in the same time period due to inaccuracies of the machine learning models. The claims recite a specific correction mechanism that addresses this deficiency by multiplying each percent impact value by the ratio of the actual severity percent change to the predicted severity percent change, in accordance with the equation I2 equals Il multiplied by P2 divided by P1 at paragraph [0057]. This is a technically specific solution to a technically identified problem with the machine learning model's output and directly improves the accuracy and reliability of the model's results.
Examiner Response
Examiner respectfully disagrees.
With respect to the first improvement argument, “the summing of variables” (the sum of all explainer values) is reciting simple math.
Applicant argues that “the result is a direct measurable reduction in data storage and network transmission requirements, as explicitly recited in the claims”.
This limitation is not shown in the claims or in specification, on how this is achieved, in other words the claims do not delineate steps through which the technology achieves the asserted improvement.
With respect to the second improvement argument (the second improvement is the self-calibrating iterative training mechanism, the claims recite iterative training of a machine learning model until the models can predict item damage severities to within a predetermined tolerance threshold of an actual item damage severity).
Examiner does not agree. Improving a model until it reaches a certain level of accuracy is also abstract.
With respect to the third improvement (the mathematical correction mechanism, multiplying each percent value by the ratio of the actual severity percent change to the predicted severity percent change) is mathematical calculation (formula or equation) and is abstract.
With respect to the specific user interface, para 17 of the specification discloses that the graphical user interface is recited at high level of generality and operating in its ordinary capacity (for displaying data such as graphs, tables), and is being used as a tool to implement the steps of the identified abstract idea.
The rejection is maintained.
Applicant argues#11
The pending claims apply any alleged abstract idea with a particular machine, namely a computing system comprising at least one processor and memory storing the specific instructions recited in the claims, operating in conjunction with machine learning models implementing the specific decision tree and explanatory model architecture described above. The claimed computing system is not a generic computer performing generic functions. It is a specifically configured machine implementing a defined model architecture, a defined aggregation mechanism, a defined iterative training process, and a defined mathematical correction formula, all operating together to produce corrected percent impact values that are displayed in a specific interactive user interface. The specificity of the machine configuration distinguishes the pending claims from the claims found abstract in Alice Corp., where the computer implementation added nothing of substance to the underlying concept. Alice Corp., 573 U.S. at 225.
Examiner Response
Examiner respectfully disagrees.
The computing system comprising at least one processor and memory storing the specific instructions recited in the claims, operating in conjunction with machine learning models implementing the specific decision tree and explanatory model architecture is not a particular machine.
As was explained above, the use of the specific decision tree is a mental process.
The additional elements of the processor, memory, and machine learning model are recited at a high level of generality and are being used as a tool to implement the steps of the identified abstract idea.
The rejection is maintained.
Applicant argues#12
The ordered combination of elements in the pending claims also supports a finding of practical application under the framework of BASCOM. In BASCOM, the Federal Circuit held that although each individual element of the claims was known, the specific, non-conventional arrangement of those elements in an ordered combination constituted a practical application. BASCOM, 827 F.3d at 1350. Similarly here, even if individual elements such as decision trees, explanatory models, and iterative training were each individually known, the specific ordered combination of: a decision-tree-based machine learning architecture with an explanatory model generating per-variable explainer values whose sum equals the model output; a line-identifier- level aggregation mechanism producing a single claim-level combined input; iterative training to a predetermined tolerance threshold; a two-step percent impact determination and correction formula based on the ratio of actual to predicted severity percent changes; and a selectable user interface displaying the corrected percent impact values, constitutes a non-conventional, non- generic arrangement that integrates any alleged abstract idea into a practical application.
The claims are therefore directed to patent-eligible subject matter under Step 2A, Prong 2 of the 2019 Guidance, and the analysis need not proceed to Step 2B.
Applicant submits in the alternative that the claims recite significantly more than the alleged abstract idea under Step 2B of the 2019 Guidance.
As discussed in BASCOM, the question under Step 2B is whether the additional elements, considered individually and as an ordered combination, amount to significantly more than the judicial exception. BASCOM, 827 F.3d at 1349-1350. The BASCOM court held that a specific, unconventional arrangement of known elements satisfied this requirement. The same analysis applies here. The specific ordered combination of the decision tree and explanatory model architecture, the line-identifier aggregation mechanism, the tolerance-threshold iterative training, and the mathematical correction formula represents an unconventional technical arrangement that goes well beyond merely implementing an abstract idea on a generic computer.
Furthermore, as noted in Section C.2.c above, the Office Action has not provided the evidentiary support required under Berkheimer to establish that any of these elements, individually or in combination, are well-understood, routine, and conventional. In the absence of such evidence, the Office Action cannot support a finding under Step 2B that the additional elements fail to provide significantly more. Berkheimer, 881 F.3d at 1368-1369.
The claims therefore satisfy Step 2B of the 2019 Guidance and are patent eligible under 35 U.S.C. § 101.
Examiner Response
Examiner respectfully disagrees.
The claims are unlike the claims in BASCOM.
BASCOM was found patent eligible because the ordered combination was directed toward solving a problem arising in the realm of computer networks, providing a solution rooted in computer technology. BASCOM provided a filtering of content that was not independent of the internet. The ordered combination of elements as set forth by BASCOM, describe a customized filtering tool with customizable features specific to each end user.
It was the non-conventional and non-generic arrangement of known conventional elements which made BASCOM patent eligible.
Whereas as discloses in the instant claims and specification disclose the
processor, memory, user interface, and machine learning model are recited at a high level of generality and are being used as a tool to implement the steps of the identified abstract idea.
Also see the Response to Applicant argues# 4, 7 ,9 above.
Therefore there are no additional elements that amount to significantly more than the identified abstract idea.
The rejection is maintained.
Claim Rejections- 35 U.S.C § 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.
1. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
The claims are either directed to a method, system and computer readable medium which are one of the statutory categories of invention. (Step 1: YES).
Representative claim 1 recites the limitations of:
A computing system comprising:
at least one processor; and memory storing instructions that, when executed by the at least one processor, cause the computing system to:
iteratively train one or more machine learning models to predict respective item damage severities associated with one or more claims based on one or more claim variables of the one or more claims until the one or more machine learning models can predict the respective item damage severities to within a predetermined tolerance threshold of an actual item damage severity, wherein the one or more machine learning models implement a multi- variable approach by aggregating the one or more claim variables for each claim into a combined input and processing the combined input simultaneously through the one or more machine learning models to generate a predicted severity output, wherein the one or more machine learning models generate one or more decision trees to determine the predicted severity output, and wherein an explanatory model applied to the one or more decision trees generates an explainer value for each of the one or more claim variables, each explainer value representing a relative contribution of the respective claim variable to the predicted severity output, and wherein a sum of the explainer values is equivalent to the predicted severity output, thereby eliminating one or more data storage and network transmission requirements that would otherwise be required for individual variable analysis;
subsequently receive, for each first claim of a plurality of first claims submitted during a first period and each second claim of a plurality of second claims submitted during a second period that is before the first period, a first set of variables associated respectively with each first claim, and a second set of variables associated respectively with each second claim, wherein each second claim is associated with a respective actual damage severity and wherein a level of information specified by the first set of variables is insufficient on its own to determine respective actual damage severities associated with the first plurality of claims;
after aggregating the first set of variables into an aggregated first set of variables and aggregating the second set of variables into an aggregated second set of variables, wherein aggregating the first set of variables and the second set of variables respectively comprises summing explainer values for one or more line identifiers associated with each claim such that each claim is associated with a single aggregated explainer value regardless of the number of coverage lines or insured items associated with the claim, input, to the one or more machine learning models, the aggregated first set of variables and the aggregated second set of variables to receive, from the one or more machine learning models, a first predicted damage severity associated with the first plurality of claims and a second predicted damage severity associated with the second plurality of claims;
parse, using the one or more machine learning models and the explanatory model, a first plurality of explainer values from the first set of variables and a second plurality of explainer values from the second set of variables;
determine a first plurality of average explainer values for each of the first plurality of explainer values and a second plurality of average explainer values for each of the second plurality of explainer values;
determine percent impact values for respective variables of the first set of variables and the second set of variables, wherein determining a respective percent impact value comprises determining a difference between a first explainer value associated with a claim variable from the first period and a second explainer value associated with the corresponding claim variable from the second period, and dividing the difference by the predicted item severity corresponding to that claim variable within the second period;
correct each percent impact value by multiplying the respective percent impact value by a ratio of a second percent change to a first percent change, wherein the first percent change is between the first plurality of average explainer values and the second plurality of average explainer values, and the second percent change is between the first plurality of average explainer values and a plurality of actual severity values corresponding to the second period;
determine respective actual damage severities associated with each of the first plurality of claims based on the first predicted damage severity associated with the first plurality of claims, the second predicted damage severity associated with the second plurality of claims, and the respective actual damage severities of the second plurality of claims; and
cause a display of a computing device to generate a damage severity user interface comprising one or more selectable features representing one or more respective corrected percent impact values associated with respective variables of the first set of variables and the second set of variables, the respective corrected percent impact values being indicative of an amount of impact a particular variable has on the first predicted damage severity associated with the first plurality of claims and the second predicted damage severity associated with the second plurality of claims.
The bolded limitations above (to predict respective item damage severities associated with one or more claims based on one or more claim variables of the one or more claims can predict the respective item damage severities to within a predetermined tolerance threshold of an actual item damage severity, implement a multi- variable approach by aggregating the one or more claim variables for each claim into a combined input and processing the combined input simultaneously through the one or more machine learning models to generate a predicted severity output to determine the predicted severity output, generates an explainer value for each of the one or more claim variables, each explainer value representing a relative contribution of the respective claim variable to the predicted severity output, and wherein a sum of the explainer values is equivalent to the predicted severity output, subsequently receive, for each first claim of a plurality of first claims submitted during a first period and each second claim of a plurality of second claims submitted during a second period that is before the first period, a first set of variables associated respectively with each first claim, and a second set of variables associated respectively with each second claim, wherein each second claim is associated with a respective actual damage severity and wherein a level of information specified by the first set of variables is insufficient on its own to determine respective actual damage severities associated with the first plurality of claims; after aggregating the first set of variables into an aggregated first set of variables and aggregating the second set of variables into an aggregated second set of variables, wherein aggregating the first set of variables and the second set of variables respectively comprises summing explainer values for one or more line identifiers associated with each claim such that each claim is associated with a single aggregated explainer value regardless of the number of coverage lines or insured items associated with the claim, a first predicted damage severity associated with the first plurality of claims and a second predicted damage severity associated with the second plurality of claims; parse, a first plurality of explainer values from the first set of variables and a second plurality of explainer values from the second set of variables; determine a first plurality of average explainer values for each of the first plurality of explainer values and a second plurality of average explainer values for each of the second plurality of explainer values; determine percent impact values for respective variables of the first set of variables and the second set of variables, wherein determining a respective percent impact value comprises determining a difference between a first explainer value associated with a claim variable from the first period and a second explainer value associated with the corresponding claim variable from the second period, and dividing the difference by the predicted item severity corresponding to that claim variable within the second period; correct each percent impact value by multiplying the respective percent impact value by a ratio of a second percent change to a first percent change, wherein the first percent change is between the first plurality of average explainer values and the second plurality of average explainer values, and the second percent change is between the first plurality of average explainer values and a plurality of actual severity values corresponding to the second period; determine respective actual damage severities associated with each of the first plurality of claims based on the first predicted damage severity associated with the first plurality of claims, the second predicted damage severity associated with the second plurality of claims, and the respective actual damage severities of the second plurality of claims; and representing one or more respective corrected percent impact values associated with respective variables of the first set of variables and the second set of variables, the respective corrected percent impact values being indicative of an amount of impact a particular variable has on the first predicted damage severity associated with the first plurality of claims and the second predicted damage severity associated with the second plurality of claims), which under its broadest reasonable covers performance of the limitation(s) as a fundamental economic practice (insurance, mitigating risk), (steps for determining damage severity based on claim data for multiple time periods).
If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation as a fundamental economic practice, then it falls within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas. Accordingly claim 1 recites an abstract idea.
Claims 1,9, 16 involve multiple abstract ideas, however are primarily focused on one (A fundamental practice of insurance), see RecogniCorp, LLC v. Nintendo Co., 855 F.3d 1322, 1327, 122 USPQ2d 1377 (Fed. Cir. 2017) ("Adding one abstract idea (math) to another abstract idea (encoding and decoding) does not render the claim non-abstract").
The claims also recite Mathematical concepts (mathematical relationships/calculations).
The claims also recite a Mental Process (a concept performed mentally by a human with pen and paper).
Claims 9,16 recite substantially the same subject matter as claim 1 and are abstract for similar reasons.
(Step 2A-Prong 1: YES. The claims are abstract)
This judicial exception is not integrated into a practical application. Limitations that are not indicative of integration into a practical application include: (1) Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05.f), (2) Adding insignificant extra solution activity to the judicial exception (MPEP 2106.05.g), (8) Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05.h).
Claims 1, 9, 16 includes the following additional elements:
- One or more machine learning models
-A processor
-A memory
-A display of a computing device
-A user interface
-Iteratively training of the one or more machine learning models
The one or more machine learning models, processor, memory, display of a computing device, user interface and iteratively training of the one or more machine learning models are recited at a high level of generality and are being used in their ordinary capacity and are being used as a tool for implementing the steps of the identified abstract idea, see MPEP 2106.05(f), where applying a computer or using a computer as a tool to perform the abstract idea is not indicative of a practical application.
Therefore, there are no additional elements in the claim that amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use.
Accordingly, these additional elements, when considered separately and as an ordered combination, do not integrate the abstract idea into a practical application because
they do not impose any meaningful limits on practicing the abstract idea.
Therefore claims 1, 9, 16 are directed to an abstract idea without a practical application. (Step 2A-Prong 2: NO). The additional claimed elements are not integrated into a practical application)
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered separately and as an ordered combination, they do not add significantly more (also known as an “inventive concept”) to the exception. As discussed above with respect to integration of the abstract idea into a practical application, there are no additional elements recited in the claim beyond the judicial exception.
Mere instructions to implement an abstract idea, on or with the use of generic computer components, or even without any computer components, cannot provide an inventive concept - rendering the claim patent ineligible.
Thus claims 1,9, 16 are not patent eligible. (Step 2B: NO.) The claims do not provide significantly more)
Dependent claims 2-8, 10-15, 17-20 which further define the abstract idea that is present in their respective independent claims 1, 9, 16 and thus correspond to a fundamental economic practice and a Mental process and hence are abstract for the reasons presented above.
Claims 3, 10, 17 recites, recites the additional element of “one or more sensors of a telematics device”. The one or more sensors of the telematics device are recited a high level of generality and is operating in its ordinary capacity and is being used as a tool to implement the steps of the identified abstract idea.
Therefore, the dependent claims do not include any additional elements that integrate the abstract idea into a practical application or are sufficient to amount to significantly more than the judicial exception when considered both individually and as an ordered combination.
Therefore, the dependent claims 2-8, 10-15, 17-20 are directed to an abstract idea. Thus, claims 1-20 are not patent-eligible.
CONCLUSION
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMMAD Z SHAIKH whose telephone number is (571)270-3444. The examiner can normally be reached M-T, 9-600; Fri, 8-11, 3-5.
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, BENNETT SIGMOND can be reached at 303-297-4411. 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.
/MOHAMMAD Z SHAIKH/Primary Examiner, Art Unit 3694 8/25/2026
