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 action is in response to the application filed 22 December 2025. Claims 21, 22, and 23 are cancelled. Claims 1, 10, and 19 are amended. Claims 1-20 and 24-28 are pending and have been examined.
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 22 December 2025 has been entered.
Response to Arguments
Applicant's arguments, see pages 8-12, filed 22 December 2025, with respect to the rejections of Claims 1-8, 10-17, and 21-28 under 35 U.S.C. 103 have been fully considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
APPLICANT'S ARGUMENT: Applicant argues (page 9, continued paragraph) that "the same input (e.g., the data of the claim) is provided to both the supervised machine learning model and the unsupervised machine learning model. ... ¶ ... [T]he Office Action argues that 'Claim 1 does not recite or require parallel processing of the inputs through the supervised and unsupervised models, only that the claim data be processed by [] both models.'"
EXAMINER'S RESPONSE: Examiner notes that applicant's arguments pertain to newly claimed limitations and are moot. Amended Claim 1 is now rejected as being unpatentable over Galia in view of Machnicki in view of Hayward in view of Liao. Liao is relied on to teach the feature of inputting data into a supervised and an unsupervised model in parallel.
APPLICANT'S ARGUMENT: Applicant argues (page 10, paragraph 1) that "Galia expressly teaches using the output of the text mining platform as an input for the decision tree and, therefore, does not teach inputting the data of the claim into a supervised machine learning model, as previously recited by the claim. As depicted in FIG. 3, reproduced above, this is extremely different from the claimed process, which provides unaltered claim data from new dataset 340 as inputs to each of the supervised model 323 and unsupervised model 333 and then combines 350 the outputs of each."
EXAMINER'S RESPONSE: Examiner notes that the features upon which applicant relies (i.e., "unaltered claim data") are not recited in the rejected claims. Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims.
In the rejection of amended Claim 1 below, Galia is relied on to teach inputting data of a claim into an unsupervised machine learning mode and inputting data of the claim into an unsupervised machine learning model, and combining the outputs of the models. Liao is further relied on to teach inputting data into a supervised model and an unsupervised model in parallel.
APPLICANT'S ARGUMENT: Applicant argues (page 10, paragraph 2) that "Applicant respectfully disagrees with the assertion that Galia teaches the determination of claim complexity at all. Galia expressly teaches the determination of 'volatility,' which it describes as '[t]he propensity for a value to differ from its predicted value.' Para. [0002]. Conversely, the Subject Application explains that 'complexity,' as recited by the claims, 'may refer to a value, or range of values, that correspond to an outcome of a claim.'"
EXAMINER'S RESPONSE: Examiner respectfully disagrees. Under BRI in light of the specification, the volatility of Galia's teaching reasonably represents "a value, or range of values, that correspond[s] to an outcome of a claim" (instant specification, [0026]). The teaching of Galia regarding volatility indicates that it may be understood as the result of a calculation that is used for scoring or ranking. For further example, see Galia, Fig. 17, Step 1708, "Determine volatility score using structured data, predictive model, and/or information about the Semantic event."
APPLICANT'S ARGUMENT: Applicant argues (page 10, paragraph 2) that "The Subject Application separately explains how the term 'escalation potential' corresponds to 'a probability that the predicted complexity range is inaccurate.' Para. [0040]. If anything, the 'volatility' of Galia is akin to the 'escalation potential' of the Subject Application, but Galia fails to teach a supervised model that generates a claim complexity, as recited by Claim 1."
Applicant argues (page 11, continued paragraph) that "since Galia explains how volatility may facilitate early escalation, any combination allegedly taught by FIG. 11 of Galia cannot possibly be based on escalation potential, as presently recited by Claim 1. See Para. [0105]. Escalation is already represented by Galia's determined volatility."
EXAMINER'S RESPONSE: Examiner respectfully disagrees.
The instant specification at [0040] provides non-limiting, example usage of "escalation potential" with regard to instant Fig. 7: "The term escalation potential, as used herein, may correspond to a probability that the predicted complexity range is inaccurate and/or is likely to be higher than predicted. Escalation determination module 225 determines, using the historical claim data, the probability of inaccuracy." The instant specification also refers to escalation potential as an "additional factor" ([0016]) used in a combination of outputs.
As indicated in the rejection of amended Claim 1 below, Galia teaches escalated intervention into claim administration facilitated by determination and by further determination of claim volatility, as calculated using results of supervised and unsupervised models. Under BRI in light of the specification, Galia's teaching of determination of claim escalation by means of the results of the models teaches or reasonably suggests the claimed determination of escalation potential for a claim.
Claim Rejections - 35 USC § 103
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-8, 10-17, and 21-28 are rejected under 35 U.S.C. 103 as being unpatentable over Galia, et al. (US 2018/0107734 A1, hereinafter "Galia") in view of Machnicki, et al. (US 2015/0178850 A1, hereinafter "Machnicki") in view of in view of Hayward, et al. (US 11,783,422 B1, hereinafter "Hayward") in view of Liao, et al. (US 2009/0099959 A1, hereinafter "Liao").
Regarding Claim 1, Galia teaches:
a method for combining output (Galia, Claim 17: "A computer-implemented method for predicting a future performance characteristic associated with an electronic record," where Galia's future characteristic corresponds to the instant combined output) of supervised (Galia, [0110]: "The predictive model(s), in various implementation, may include one or more of neural networks, Bayesian networks (such as Hidden Markov models), expert systems, decision trees, collections of decision trees, support vector machines, or other systems known in the art for addressing problems with large numbers of variables. Preferably, the predictive model(s) are trained on prior text data and outcomes known to the insurance company," where Galia's predictive model trained on known outcomes corresponds to the instant supervised model, and where a further example of training data includes: [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") and unsupervised machine learning models (Galia, [0079]: "[0079] In some embodiments, the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used to discover common patterns in data. ... In some cases, cluster analysis ... may be used to explore the business context of such themes") the method comprising:
receiving, from a client device, an indication of a claim and a request ... specific to the claim (Galia, [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104. The client terminal 107 also includes an input module where a new claim may be filed, and where information pertaining to the claim may be entered, such as a notice of loss, for example," where Galia's request to administer the associated claim corresponds to the instant request specific to the claim, as in [0091]: "FIG. 9 is flowchart 900 illustrating a method of claim administration based upon a claim's predicted likelihood of exhibiting cost volatility, according to one embodiment of the invention. The method begins at step 901, when an insurance company receives a notice of loss");
responsive to receiving the request (Galia, [0093]: "After a period of time in which additional claim characteristics are collected by the employee assigned to process the claim ( e.g., 30, 45, 60, or 90 days after the notice of loss) the back-end application computer server 103 may access text mined data and other data at step 907 to calculate a claim likelihood of volatility," where Galia's claim initiation plus arbitrary window of time corresponds to the instant responsive to receiving):
inputting data of the claim (Galia, [0049]: "Some examples of characteristics that might be considered in connection with a volatility prediction include medical invoice totals ... visit counts ... primary injury ... accident state, claimant age, ... and prior claimant injuries and/or medical conditions") into a supervised (Galia, [0050]: "the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") machine learning model (Galia, [0047]: "The predictive model 104 may be a ... neural network") and receiving as output from the supervised machine learning model a complexity of the claim (Galia, [0046]: "The predictive model 104 is used by the back-end application computer server 103 to estimate the likelihood that a claim will exhibit increased volatility in comparison to other claims" and [0029]: "the confidence an insurer can have in predicting the total cost of resolving a claim may be much lower. The propensity for a claim to exceed its predicted total resolution cost, including medical costs, is referred to herein as 'claim volatility,'" where Galia's predicted claim volatility corresponds to the instant complexity of the claim, where the instant "complexity" is interpreted under BRI to represent a value that corresponds to the outcome of a claim, as in the specification of the instant application, [0026]: "complexity, as used herein, may refer to a value, or range of values, that correspond to an outcome of a claim"), wherein the complexity of the claim is generated by inputting the data of the claim into a baseline ... model of the supervised machine learning model (Galia, [0068]: "The predictive model 104 may be updated from time to time as an insurance company receives additional claim data to use as a baseline for building the predictive model 104. The updating includes retraining the model based on the updated data using the previously selected parameters. Alternatively, or in addition, updating includes carrying out the model generation process again based on the new data");
... inputting ... data of the claim into an unsupervised machine learning model and receiving as output from the unsupervised machine learning model an identification of a cluster of candidate claims to which the claim belongs (Galia, [0079]: "the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used to discover common patterns in data. ... In some cases, cluster analysis ... may be used to explore the business context of such themes. For example, ... it might be noted that a particular automobile model is frequently experiencing a particular unintended problem" and [0076]: "According to some embodiments, the text mining associated with the data flow is a 'big data' activity that may use machine learning to sift through large amounts of unstructured data to find meaningful patterns to support business decisions") ... ;
combining the complexity and the identification of the cluster into a combined result by determining ... an escalation potential for the claim (Galia, [0110]: "text data may be used in conjunction with one or more predictive models to take into account a large number of underwriting and/or other parameters," where Galia's cluster analysis occurs by way of mining of text data, and [0105]: "embodiments may utilize text mining to help determine which characteristics of an insurance claim ... may indicate a large deviation from expectation. Moreover, text mining information ... may further impact the volatility of a claim. ... Note that ... there is a population of volatile claims whose outcomes vary significantly from their expectations. It is these claims that can drive the total severity, and the predictive model described herein may ... facilitate early intervention via claim prioritization, escalation, and/or re-assignment");
... a matrix corresponding to the combined result ... (Galia, Fig. 11, depicting the user interface for a "Claim Volatility Tool," which comprises a matrix relating indemnity reserved/paid values to predicted claim volatility by a degree of correlation, and [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times"); and
providing, for display at the client device ... a display of the matrix (Galia, [0100]: "Fig. 11 is a claim volatility tool machine learning cluster analysis example display 1100.... According to some embodiments, a user might select a graphically displayed element to see more information about that element" where [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104").
Galia may not explicitly teach receiving, from a client device ... a request for an analysis specific to the claim; identifying a cell in a matrix ... having a probability curve corresponding to the claim; providing, for display at the client device, an identification of the cell, the cell to be emphasized to a user within a display of the matrix , the cell emphasized because it corresponds to the claim , other cells of the matrix having respective probability curves that are not specific to the claim.
However, Machnicki teaches:
receiving, from a client device (Machnicki, [0138]: "According to some embodiments, the request to analyze the third insurance claim is received via a GUI of a mobile computing device," where Machnicki's mobile device corresponds to the instant client device), ... a request for an analysis specific to the claim (Machnicki, [0138]: "the method may further comprise generating, by the processing device and based on the third liability data for the third insurance claim, the determination of the third insurance claim. According to some embodiments, the determination of the third insurance claim comprises an indication of whether the third insurance claim should be allowed or denied," where Machnicki's determination of allowability of the third claim corresponds to the instant analysis specific to the claim);
identifying a cell in a matrix (Machnicki, [0088]: "The total premium calculated for a potential insurance policy offering ... may, to continue the example, be graded between 'B' and 'C' (e.g., at 810 of FIG. 8) or between 'Fair' and 'Average'. The resulting combination of risk score and premium rating may be plotted on the risk matrix 900, as represented by a data point 904 shown in FIG. 9," where Machnicki's combination plotted on the matrix corresponds to the instant identifying, and where Machnicki's quadrant containing figure element 1004 from Fig. 9 corresponds to the instant cell) ... having a probability curve corresponding to the claim (Machnicki, [0085]: "One example of how the risk matrix 900 may be output and/or implemented with respect to VED [Virtual Engineering Data] of an account and/or group of objects will now be described. ... Typical risk metrics ... may be utilized to produce expected loss frequency and loss severity distributions," where Machnicki's loss severity distribution corresponds to the instant probability curve, and where the distribution corresponds to a claim by way of Machnicki's associated engineering data).
providing ... an identification of the cell (Machnicki, [0084]: "The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher," where quadrant 902b is identified in Fig. 9 as "Good Money Despite Risk"), the cell to be emphasized to a user within a display of the matrix, the cell emphasized because it corresponds to the claim (Machnicki, [0088]: "The data point 904, based on the VED-influenced risk score and the corresponding VED-influenced premium calculation, is plotted in the second quadrant 902b," where Machnicki's point plotted in the second quadrant corresponds to the instant cell emphasized as corresponding to the claim), other cells of the matrix having respective probability curves (Machnicki, [0084]: "the risk matrix 900 may comprise four (4) quadrants 902a-d .... The first quadrant 902a represents the most desirable situations where risk scores are low and premiums are highly graded. The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher. ... The third quadrant 902c represents less desirable characteristics of having poorly graded premiums with low risk scores and the fourth quadrant 902d represents the least desirable characteristics of having poorly graded premiums as well as high risk scores") that are not specific to the claim (Machnicki, [0088]: "The data point 904 ... is plotted in the second quadrant 902b, in a position indicating ... the calculated premium is probably large enough to compensate for the level of risk").
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 Galia regarding receiving an indication of a claim and a request specific to the claim from a client device with those of Machnicki regarding receiving a request for an analysis specific to a claim from a client device and displaying at the client device a cell in a matrix having a probability curve corresponding to the claim, emphasized because it corresponds to the claim.
The motivation to do so would be to facilitate evaluating a level of risk and other processing related to an insurance claim based on related engineering data (Machnicki, [0062]: "such engineering data 602a-n may comprise data indicative of a level of risk ... at the time of casualty or loss (e.g., as defined by the one or more claims). Information on claims ... to update, improve, and/or enhance these procedures and/or associated software and/or devices. In some embodiments, engineering data 602a-n may be utilized to determine, inform, define, and/or facilitate a determination or allocation of responsibility related to a loss").
The Galia/Machnicki combination does not explicitly teach inputting the data .. into a ... deep learning model.
However, Hayward teaches:
inputting the data .. into a ... deep learning model (Hayward, Fig. 1, depicting historical claim data processed by an artificial intelligence platform, where block 150 depicts a neural network component, as in col. 9, lines 56-58: "ANN unit 150 may process claim data by training models with the data and/or by operating a trained ANN model with the data" and where the ANN may be a deep learning model comprising hidden layers, as depicted in Fig. 3, and described at col. 19, lines 32-43: "FIG. 3 depicts an exemplary artificial neural network (ANN) 300 which may be trained by ANN unit 150 of FIG. 2.... The exemplary ANN 300 may include layers of neurons, including input layer 302, one or more hidden layers 304-1 through 304-n, and an output layer 306").
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 the Galia/Machnicki combination regarding inputting data of the claim into a supervised machine learning model with those of Hayward regarding inputting the data into a deep learning model.
The motivation to do so would be to facilitate training and use of a model in such a manner that it produces results according to the relevance of inputs for a task (Hayward, col. 19, 44-58: "the number of elements used by ANN 300 may change during the 55 training process, and some neurons may be bypassed or ignored if, for example, during execution of the ANN, they are determined to be of less relevance" and col. 20, lines 4-13: "each layer may have a discrete, recognizable, function with respect to input data ... where all dimensions are analyzing a distinct and unrelated aspect of the input data. For example, the dimensions may correspond to aspects of a health insurance considered strongly determinative" and col. 22, lines 63: "portions of a claim that are identified as most dispositive of payment may be identified and processed first. If, in that example, they are dispositive of payment, then processing of further claim elements may be abated to save processing resources").
The Galia/Machnicki/Hayward combination does not explicitly teach inputting data ... into a supervised machine learning model ...; in parallel to inputting the data ... into the supervised machine learning model, also inputting the data ... into an unsupervised machine learning model; and combining ... into a combined result ... via an ... determination model.
However, Liao teaches:
inputting data ... into a supervised machine learning model ...; in parallel to inputting the data ... into the supervised machine learning model, also inputting the data ... into an unsupervised machine learning model (Liao, Fig. 2, depicting supervised model 132 and unsupervised model 140 receiving data in parallel, and [0041]: "The data preprocessing module 124 provides application data to one or more models .... [A]pplication data is provided to one or more loan models 132 that generate data indicative of fraud based on application and applicant data. ... The data preprocessing module 124 can also provide application data to one or more entity models 140 that are configured to identify fraud based on data associated with entities involved in the processing of the application," where Luo's loan models 132 are supervised, as in Fig. 3, Supervised Model 170, and entity models 140 are unsupervised, as in [0063]: "FIG. 4 is a functional block diagram illustrating examples of the entity models 140 in the fraud detection system 100. ... [I]n one embodiment, an unsupervised model, e.g., a clustering model such as k-means, is applied to risk indicators for historical transactions for each entity") and receiving as output from the unsupervised machine learning model an identification of a cluster ... (Liao, [0063]: "FIG. 4 is a functional block diagram illustrating examples of the entity models 140 in the fraud detection system 100. ... [I]n one embodiment, an unsupervised model, e.g., a clustering model such as k-means, is applied to risk indicators for historical transactions for each entity. A score for each risk indicator, for each entity, is calculated based on the relation of the particular entity to the clusters") wherein the identification of the cluster ... is performed independent of the output from the supervised machine learning model (Liao, Fig. 2, depicting independence of output of the supervised model 132 and the unsupervised model 140 based on common inputs);
combining ... into a combined result ... via an ... determination model (Liao, Fig. 2, Integrator 136, depicting a model receiving output of the supervised model 132 and the unsupervised model 140, and [0041]: "The data indicative of fraud generated by the loan models 132 can be provided to an integrator 136 that combines scores from one or more models into a final score" and [0045]: "the entity scoring model 150 combines each of the risk indicator scores for a particular entity using a weighted average or other suitable combining calculation to generate an overall entity score. In addition, the risk indicators having higher scores can also be identified and provided to the integrator 136" and [0056]: "The integrator 136 can be configured to apply weights and for processing rules to generate one or more scores and risk indicators based on the data indicative of fraud provided by one or more of the loan models 132, the entity models 140 and entity scoring modules 160").
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 Galia/Machnicki/Hayward combination regarding inputting data of the claim into a supervised machine learning model and an unsupervised machine learning model with those of Liao regarding inputting data into a supervised machine learning model and in parallel to inputting the data into the supervised machine learning model, also inputting the data into an unsupervised machine learning model, and combining into a combined result via a determination model.
The motivation to do so would be to facilitate performing fraud analysis of accounts under circumstances where ensuring accuracy and scaling with large volumes of data are improved by the combination of results of the types of models (Liao, [0028]: "Existing fraud detection systems can use transaction data in addition to data related to the transacting entities to identify fraud. ... However, the fraud detection capabilities of existing systems have not kept pace with either the types of fraudulent activity that have evolved or increasing processing and storage capabilities of computing systems" and [0029]: "it has been found that ... fraud detection can be improved by using stored past transaction data in place of, or in addition to, summarized forms of past transaction data. In addition, in one embodiment, fraud detection can be improved by using statistical information that is stored according to groups of individuals that form clusters").
Regarding Claim 2, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the supervised machine learning model was trained using a generic set of training data (Galia, [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above"), wherein the client device corresponds to an enterprise (Galia, [0032]: "the system 100 includes a text mining platform 150 that receives information from a semantic rules database 110, a claim system 120, one or more text input data sources 130 ( e.g., internal to an insurance enterprise)"), wherein the enterprise has access to historical claim data (Galia, [0043]: "The data warehouse 101 is the main electronic depository of an insurance company's current and historical data."), and wherein the training of the supervised machine learning model is supplemented by undergoing a training process using the historical claim data of the enterprise (Galia, [0068]: "The predictive model 104 may be updated from time to time as an insurance company receives additional claim data to use as a baseline for building the predictive model 104.").
Claim 11 recites the limitations of Claim 2 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 3, rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the data of the claim comprises structured and unstructured data (Galia, [0110]: "the parameters can be selected from any of the structured data parameters stored in the present system, whether the parameters were input into the system originally in a structured format or whether they were extracted from previously unstructured text, such as from big data."), wherein the supervised machine learning model is a multi-branch machine learning model with a first branch trained to process structured data and a second branch trained to process unstructured data (Galia, Fig. 7B/756 and [0080]: "Text flags 754 might be identified as described in connection with 7B. According to some embodiments, static claim variables like age, geography, profession, wage levels, etc. may be considered").
Claim 12 recites the limitations of Claim 3 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 4, the rejection of Claim 3 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the multi-branch model comprises shared layers trained to combine the unstructured data and the structured data in order to output the complexity (Galia, [0060]: "Note that the collected data might include any other type of data, including data identified by the text mining platform 150. At step 502, the volatility of each claim in this data set may be determined by computing the claim's target variable as described in relation to FIG. 4." and Fig. 5/503).
Claim 13 recites the limitations of Claim 4 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 5, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein combining the complexity and the identification of the cluster into a combined result comprises: determining an escalation potential of the claim (Galia, [0105]: "it is these claims that can drive the total severity, and the predictive model described herein may deliver an ability to predict volatility (or lack of volatility) and facilitate early intervention via claim prioritization, escalation, and/or re-assignment.") and weighting the complexity based on the determined escalation potential (Galia, [0042]: " The insurance applications might be associated with, for example … volatile claim detection…. Note that the transmitted indication might be used to trigger an insurance application … and/or update an insurance application (e.g., by updating a variable or weighing factor of a predictive model).").
Claim 14 recites the limitations of Claim 5 in computer-readable medium forms and is rejected under the same rationale.
Regarding Claim 6, the rejection of Claim 5 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein determining the escalation potential comprises: identifying historical cost predictions of historical claims and actual claim costs for those historical claims (Galia, [0053]: "Looking at the historical medical spend, the system can classify certain patterns as volatile. … For example, if the claim associated with the bottom right graph of FIG. 3A was expected to flatten early in the claim's life-but the insurance enterprise instead continued to increase payouts over time, then the claim may be classified as 'volatile.'"); determining a relative amount of the historical claims having an actual claim cost higher than a historical cost prediction (Galia, [0064]: "The model was generated using a sample data set including several hundred thousand historical claims. Approximately 1% of these claims were identified as being volatile"); and determining the escalation potential based on the relative amount (Galia, [0105]: "there is a population of volatile claims whose outcomes vary significantly from their expectations. It is these claims that can drive the total severity, and the predictive model described herein may deliver an ability to predict volatility (or lack of volatility) and facilitate early intervention via claim prioritization, escalation, and/or re-assignment").
Claims 15 recites the limitations of Claim 6 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 7, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the matrix is generated as having, in a first dimension, a first axis corresponding to an amount of clusters of candidate claims, and in a second dimension, a second axis corresponding to different ranges of complexity values (Galia, Fig. 11 and [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times").
Claims 16 recites the limitations of Claim 7 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 8, the rejection of Claim 7 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the matrix comprises, at each intersection of the first axis and the second axis, a cell, the cell indicating a relative complexity value with respect to surrounding cells (Galia, Fig. 11 and [0100]: "For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation.").
Claims 17 recites the limitations of Claim 8 in computer-readable medium form and is rejected under the same rationale.
Regarding Claim 10, Galia teaches:
a non-transitory computer-readable medium comprising memory with instructions encoded thereon for combining output (Galia, Claim 19, "a non-transitory computer-readable medium storing instructions adapted to be executed by a computer processor to perform a method to predict a future performance characteristic for an electronic record," where Galia's future characteristic corresponds to the instant combined output) of supervised (Galia, [0110]: "The predictive model(s), in various implementation, may include one or more of neural networks, Bayesian networks (such as Hidden Markov models), expert systems, decision trees, collections of decision trees, support vector machines, or other systems known in the art for addressing problems with large numbers of variables. Preferably, the predictive model(s) are trained on prior text data and outcomes known to the insurance company," where Galia's predictive model trained on known outcomes corresponds to the instant supervised model, and where a further example of training data includes: [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") and unsupervised machine learning models (Galia, [0079]: "In some embodiments, the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used to discover common patterns in data"), the instructions when executed causing one or more processors to perform operations (Galia, Claim 19, "instructions adapted to be executed by a computer processor"), the instructions comprising instructions to:
receive, from a client device, an indication of a claim and a request ... specific to the claim (Galia, [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104. The client terminal 107 also includes an input module where a new claim may be filed, and where information pertaining to the claim may be entered, such as a notice of loss, for example," where Galia's request to administer the associated claim corresponds to the instant request specific to the claim, as in [0091]: "FIG. 9 is flowchart 900 illustrating a method of claim administration based upon a claim's predicted likelihood of exhibiting cost volatility, according to one embodiment of the invention. The method begins at step 901, when an insurance company receives a notice of loss");
responsive to receiving the request (Galia, [0093]: "After a period of time in which additional claim characteristics are collected by the employee assigned to process the claim ( e.g., 30, 45, 60, or 90 days after the notice of loss) the back-end application computer server 103 may access text mined data and other data at step 907 to calculate a claim likelihood of volatility," where Galia's claim initiation plus arbitrary window of time corresponds to the instant responsive to receiving):
input data of the claim (Galia, [0046]: "The predictive model 104 is used by the back-end application computer server 103 to estimate the likelihood that a claim will exhibit increased volatility in comparison to other claims" and [0049]: "Some examples of characteristics that might be considered in connection with a volatility prediction include ... estimated incurred (reserved amount) at end of evaluation period, estimated total medical spend, ... estimated indemnity payment," where Galia's costs associated with a claim correspond to the instant data of the claim) into a supervised (Galia, [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") machine learning model (Galia, [0047]: "The predictive model 104 may be a ... neural network") and receiving as output from the supervised machine learning model a complexity of the claim (Galia, [0063]: "FIG. 6 is a flowchart of a method 600 of using the predictive model generated in FIG. 5 to obtain a future volatility prediction on a particular test claim," where Galia's predicted claim volatility corresponds to the instant a complexity output from the model) ... ;
input the data of the claim into an unsupervised machine learning model (Galia, [0079]: "the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used" and [0073] "The pulled data may be associated with, for example, various insurance applications and/or data types 720, such as claim handler notes, loss descriptions, injury descriptions, FNOL [first notice of loss] statements, call transcripts, and/or OCR documents") and receiving as output from the unsupervised machine learning model an identification of a cluster of candidate claims to which the claim belongs (Galia, [0100]: "FIG. 11 is a claim volatility tool machine learning cluster analysis example display.... In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times"), wherein the cluster of candidate claims is identified by inputting the data of the claim into a clustering algorithm of the unsupervised machine learning model (Galia, [0100]: "Clustering may, for example, find patterns that are resident in a volume of data ... In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times (e.g., at initiation, at 30 days, etc.). For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation," where Galia's clustering of claims according to timeframe and correlation corresponds to the instant clustering algorithm);
combine the complexity and the identification of the cluster into a combined result (Galia, [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times (e.g., at initiation, at 30 days, etc.). For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation," where Galia's depicted degree of correlation of volatility based according to reserved/paid status corresponds to the instant combined result);
identify ... a matrix corresponding to the combined result (Galia, Fig. 11, depicting the user interface for a "Claim Volatility Tool," which comprises a matrix relating indemnity reserved/paid values to predicted claim volatility by a degree of correlation, and [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times"); and
provide, for display at the client device ... a display of the matrix (Galia, [0100]: "Fig. 11 is a claim volatility tool machine learning cluster analysis example display 1100.... According to some embodiments, a user might select a graphically displayed element to see more information about that element" where [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104").
Galia may not explicitly teach receiving, from a client device ... a request for an analysis specific to the claim; identifying a cell in a matrix ... having a probability curve corresponding to the claim; providing, for display at the client device, an identification of the cell, the cell to be emphasized to a user within a display of the matrix , the cell emphasized because it corresponds to the claim , other cells of the matrix having respective probability curves that are not specific to the claim.
However, Machnicki teaches:
receive, from a client device (Machnicki, [0138]: "According to some embodiments, the request to analyze the third insurance claim is received via a GUI of a mobile computing device," where Machnicki's mobile device corresponds to the instant client device), ... a request for an analysis specific to the claim (Machnicki, [0138]: "the method may further comprise generating, by the processing device and based on the third liability data for the third insurance claim, the determination of the third insurance claim. According to some embodiments, the determination of the third insurance claim comprises an indication of whether the third insurance claim should be allowed or denied," where Machnicki's determination of allowability of the third claim corresponds to the instant analysis specific to the claim);
identifying a cell in a matrix (Machnicki, [0088]: "The total premium calculated for a potential insurance policy offering ... may, to continue the example, be graded between 'B' and 'C' (e.g., at 810 of FIG. 8) or between 'Fair' and 'Average'. The resulting combination of risk score and premium rating may be plotted on the risk matrix 900, as represented by a data point 904 shown in FIG. 9," where Machnicki's combination plotted on the matrix corresponds to the instant identifying, and where Machnicki's quadrant containing figure element 1004 from Fig. 9 corresponds to the instant cell) ... having a probability curve corresponding to the claim (Machnicki, [0085]: "One example of how the risk matrix 900 may be output and/or implemented with respect to VED [Virtual Engineering Data] of an account and/or group of objects will now be described. ... Typical risk metrics ... may be utilized to produce expected loss frequency and loss severity distributions," where Machnicki's loss severity distribution corresponds to the instant probability curve, and where the distribution corresponds to a claim by way of Machnicki's associated engineering data);
provide ... an identification of the cell (Machnicki, [0084]: "The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher," where quadrant 902b is identified in Fig. 9 as "Good Money Despite Risk"), the cell to be emphasized to a user within a display of the matrix, the cell emphasized because it corresponds to the claim (Machnicki, [0088]: "The data point 904, based on the VED-influenced risk score and the corresponding VED-influenced premium calculation, is plotted in the second quadrant 902b," where Machnicki's point plotted in the second quadrant corresponds to the instant cell emphasized as corresponding to the claim), other cells of the matrix having respective probability curves (Machnicki, [0084]: "the risk matrix 900 may comprise four (4) quadrants 902a-d .... The first quadrant 902a represents the most desirable situations where risk scores are low and premiums are highly graded. The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher. ... The third quadrant 902c represents less desirable characteristics of having poorly graded premiums with low risk scores and the fourth quadrant 902d represents the least desirable characteristics of having poorly graded premiums as well as high risk scores") that are not specific to the claim (Machnicki, [0088]: "The data point 904 ... is plotted in the second quadrant 902b, in a position indicating ... the calculated premium is probably large enough to compensate for the level of risk").
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 Galia regarding receiving an indication of a claim and a request specific to the claim from a client device with those of Machnicki regarding receiving a request for an analysis specific to a claim from a client device and displaying at the client device a cell in a matrix having a probability curve corresponding to the claim, emphasized because it corresponds to the claim.
The motivation to do so would be to facilitate evaluating a level of risk and other processing related to an insurance claim based on related engineering data (Machnicki, [0062]: "such engineering data 602a-n may comprise data indicative of a level of risk ... at the time of casualty or loss (e.g., as defined by the one or more claims). Information on claims ... to update, improve, and/or enhance these procedures and/or associated software and/or devices. In some embodiments, engineering data 602a-n may be utilized to determine, inform, define, and/or facilitate a determination or allocation of responsibility related to a loss").
The Galia/Machnicki combination has been shown to teach inputting data of the claim into a supervised machine learning model and receiving as output from the supervised machine learning model a complexity of the claim. The Galia/Machnicki combination does not explicitly teach the complexity of the claim is generated by reconciling data with different dimensionalities within the data of the claim.
However, Hayward teaches:
the complexity of the claim is generated by reconciling data with different dimensionalities within the data of the claim (Hayward, col. 20, lines 4-13: "each layer may have a discrete, recognizable, function with respect to input data. For example, if n=3, a first layer may analyze one dimension of inputs, a second layer a second dimension, and the final layer a third dimension of the inputs, where all dimensions are analyzing a distinct and unrelated aspect of the input data. For example, the dimensions may correspond to aspects of a health insurance considered strongly determinative, then those that are considered of intermediate importance, and finally those that are of less relevance" and col. 19, lines 54-58: "the number of elements used by ANN 300 may change during the 55 training process, and some neurons may be bypassed or ignored if, for example, during execution of the ANN, they are determined to be of less relevance," where Hayward's training and use per relevance corresponds to the instant reconciling).
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 the Galia/Machnicki combination regarding inputting data of the claim into a supervised machine learning model and receiving as output from the supervised machine learning model a complexity of the claim with the teachings of Hayward regarding wherein the complexity of the claim is generated by reconciling data with different dimensionalities within the data of the claim.
The motivation to do so would be to facilitate training and use of a model in such a manner that it produces results according to the relevance of inputs for a task (Hayward, col. 20, lines 4-13: "each layer may have a discrete, recognizable, function with respect to input data ... where all dimensions are analyzing a distinct and unrelated aspect of the input data. For example, the dimensions may correspond to aspects of a health insurance considered strongly determinative" and col. 22, lines 63: "portions of a claim that are identified as most dispositive of payment may be identified and processed first. If, in that example, they are dispositive of payment, then processing of further claim elements may be abated to save processing resources").
The Galia/Machnicki/Hayward combination does not explicitly teach input data ... into a supervised machine learning model ...; in parallel to inputting the data ... into the supervised machine learning model, also input the data ... into an unsupervised machine learning model and receiving as output from the unsupervised machine learning model an identification of a cluster.
However, Liao teaches:
input data ... into a supervised machine learning model ...; in parallel to inputting the data ... into the supervised machine learning model, also input the data ... into an unsupervised machine learning model (Liao, Fig. 2, depicting supervised model 132 and unsupervised model 140 receiving data in parallel, and [0041]: "The data preprocessing module 124 provides application data to one or more models .... [A]pplication data is provided to one or more loan models 132 that generate data indicative of fraud based on application and applicant data. ... The data preprocessing module 124 can also provide application data to one or more entity models 140 that are configured to identify fraud based on data associated with entities involved in the processing of the application," where Luo's loan models 132 are supervised, as in Fig. 3, Supervised Model 170, and entity models 140 are unsupervised, as in [0063]: "FIG. 4 is a functional block diagram illustrating examples of the entity models 140 in the fraud detection system 100. ... [I]n one embodiment, an unsupervised model, e.g., a clustering model such as k-means, is applied to risk indicators for historical transactions for each entity") and receiving as output from the unsupervised machine learning model an identification of a cluster (Liao, [0063]: "FIG. 4 is a functional block diagram illustrating examples of the entity models 140 in the fraud detection system 100. ... [I]n one embodiment, an unsupervised model, e.g., a clustering model such as k-means, is applied to risk indicators for historical transactions for each entity. A score for each risk indicator, for each entity, is calculated based on the relation of the particular entity to the clusters").
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 Galia/Machnicki/Hayward combination regarding inputting data of the claim into a supervised machine learning model and an unsupervised machine learning model with those of Liao regarding input data into a supervised machine learning model and in parallel to inputting the data into the supervised machine learning model, also input the data into an unsupervised machine learning model and receive as output from the unsupervised machine learning model an identification of a cluster.
The motivation to do so would be to facilitate performing fraud analysis of accounts under circumstances where ensuring accuracy and scaling with large volumes of data are improved by the combination of results of the types of models (Liao, [0028]: "Existing fraud detection systems can use transaction data in addition to data related to the transacting entities to identify fraud. ... However, the fraud detection capabilities of existing systems have not kept pace with either the types of fraudulent activity that have evolved or increasing processing and storage capabilities of computing systems" and [0029]: "it has been found that ... fraud detection can be improved by using stored past transaction data in place of, or in addition to, summarized forms of past transaction data. In addition, in one embodiment, fraud detection can be improved by using statistical information that is stored according to groups of individuals that form clusters").
Regarding Claim 24, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the cluster of candidate claims is further identified by receiving a closest cluster identifier associated with the data of the claim from the clustering algorithm (Galia, [0089]: "Referring to FIG. 8C, a table is shown that represents the text mining results database 880 .... The table may include, for example, entries identifying results of a text mining operation. The table may also define fields ... for each of the entries. The fields ... may, according to some embodiments, specify: a text mining result identifier," where Galia's text mining result identifier may be a cluster identifier, such as a theme classification in: [0079]: "the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used to discover common patterns in data. For example, highly recurrent themes may be classified, and other concepts may then be highlighted based on a sense of adjacency to these recurrent themes. In some cases, cluster analysis and drilldown tools may be used to explore the business context of such themes").
Regarding Claim 25, the rejection of Claim 1 is incorporated.
Hayward further teaches:
wherein the baseline deep learning model is configured to reconcile data with different dimensionalities within the data of the claim (Hayward, col. 20, lines 4-13: "each layer may have a discrete, recognizable, function with respect to input data. For example, if n=3, a first layer may analyze one dimension of inputs, a second layer a second dimension, and the final layer a third dimension of the inputs, where all dimensions are analyzing a distinct and unrelated aspect of the input data. For example, the dimensions may correspond to aspects of a health insurance considered strongly determinative, then those that are considered of intermediate importance, and finally those that are of less relevance").
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 the Galia/Machnicki combination regarding inputting data of the claim into a supervised machine learning model with the further teachings of Hayward regarding wherein the baseline deep learning model is configured to reconcile data with different dimensionalities within the data of the claim.
The motivation to do so would be to facilitate training and use of a model in such a manner that it produces results according to the relevance of inputs for a task (Hayward, col. 19, 44-58: "the number of elements used by ANN 300 may change during the 55 training process, and some neurons may be bypassed or ignored if, for example, during execution of the ANN, they are determined to be of less relevance" and 22, lines 63: "portions of a claim that are identified as most dispositive of payment may be identified and processed first. If, in that example, they are dispositive of payment, then processing of further claim elements may be abated to save processing resources").
Regarding Claim 26, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the baseline deep learning model is modified using enterprise data (Galia, [0068]: "The predictive model 104 may be updated from time to time as an insurance company receives additional claim data to use as a baseline for building the predictive model 104. The updating includes retraining the model based on the updated data using the previously selected parameters," where Galia's insurance company claim data corresponds to the instant enterprise data, as in [0098]: "FIG. 10 illustrates an optional loss reserving strategy which an insurance enterprise may elect to implement based on volatility information yielded by a computerized predictive model").
Regarding Claim 27, the rejection of Claim 1 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein the combined result comprises a claim prediction (Galia, [0004]: "The computer server may then execute a computerized predictive model to predict a future performance characteristic indicator for the electronic record based on the at least one parameter and the indication received from the text mining platform, wherein the future performance characteristic indicator is indicative of a likelihood of an actual value of the electronic record differing from a predicted value of the electronic record," where Galia's future performance characteristic corresponds to the instant combined result
, where Galia's combined predictive result additionally includes partial prediction results from models, such as model 104 of [0047]: "The predictive model 104 may be a linear regression model, a neural network, a decision tree model, or a collection of decision trees, for example, and combinations" and [0048]: "The predictive model 104 preferably takes into account a large number of parameters, such as, for example, characteristics of electronic records").
Regarding Claim 28, the rejection of Claim 27 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
the claim prediction is generated based on a graph comprising a complexity axis and a cluster axis (Galia, Fig. 11, depicting a horizontal axis spanning eight cluster categories and a vertical axis spanning eight indemnity categories, and [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times (e.g., at initiation, at 30 days, etc.). For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation," where Galia's correlation corresponds to the instant claim prediction of future performance, as in [0004]: "the future performance characteristic indicator is indicative of a likelihood of an actual value of the electronic record differing from a predicted value of the electronic record"), and wherein the claim prediction corresponds to an intersection of the generated complexity and identified cluster on the graph (Galia, Fig. 11, depicting correlations of various strengths at the intersections of the indemnity-paid and cluster axes).
Claims 9 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Galia, et al. (US 2018/0107734 A1, hereinafter "Galia") in view of Machnicki, et al. (US 2015/0178850 A1, hereinafter "Machnicki") in view of in view of Hayward, et al. (US 11,783,422 B1, hereinafter "Hayward") in further view of Hoffberg, "Game Theoretic Prioritization System and Method" (US 2010/0235285 A1), hereinafter Hoffberg.
Regarding Claim 9, the rejection of Claim 8 is incorporated.
The Galia/Machnicki/Hayward/Liao combination teaches:
wherein each cell comprises ... a likelihood that a given claim matching that cell will have a given value (Galia, Fig. 11 and [0100]: "For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation.").
The Galia/Machnicki/Hayward/Liao combination does not explicitly teach each cell comprises a probability curve indicating a likelihood.
However, Hoffberg teaches:
each cell comprises a probability curve indicating a likelihood (Hoffberg, [0821]: "According to an embodiment of the invention, relevance of information and information reliability are represented as orthogonal axes. For each set of facts or interpretation (hypothesis) thereof, a representation is projected on the plane defined by these two axes. This representation for each event generally takes the form of a bell curve, although the statistics for each curve need not be Gaussian" where [0825]: "each event ... are represented as a distribution projected into a relevance-reliability plane. ... Since the determination of relevance is generally not exact nor precise, there is an associated reliability, that is, there is a range of possibilities and their likelihoods relating to a set of presumed facts").
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 the Galia/Machnicki/Hayward/Liao combination regarding wherein each cell comprising a likelihood that a given claim matching that cell will have a given value with the further teachings of Hoffberg regarding each cell comprising a probability curve indicating a likelihood.
The motivation to do so would be to facilitate accounting for user utility with respect to a given event among possible events in the display of risk data (Hoffberg, [0824]: "The present invention therefore provides a method, comprising ... determining, from the possible events, a relevance to a user and associated statistical distribution thereof.... The... ranking comprises an analysis of probability-weighted benefits from each event to an overall utility function for the user").
Claim 18 recites the limitations of Claim 9 in computer-readable medium form and is rejected under the same rationale.
Claims 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Galia, et al. (US 2018/0107734 A1, hereinafter "Galia") in view of Machnicki, et al. (US 2015/0178850 A1, hereinafter "Machnicki").
Regarding Claim 19, Galia teaches:
a system for combining output (Galia, Fig. 1 and Claim 1, "A system for predicting a future performance characteristic associated with an electronic record," where Galia's future characteristic corresponds to the instant combined output) of supervised (Galia, [0110]: "The predictive model(s), in various implementation, may include one or more of neural networks, Bayesian networks (such as Hidden Markov models), expert systems, decision trees, collections of decision trees, support vector machines, or other systems known in the art for addressing problems with large numbers of variables. Preferably, the predictive model(s) are trained on prior text data and outcomes known to the insurance company," where Galia's predictive model trained on known outcomes corresponds to the instant supervised model, and where a further example of training data includes: [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") and unsupervised machine learning models (Galia, [0079]: "In some embodiments, the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used to discover common patterns in data") the system comprising:
a communications module (Galia, Fig. 1, 103, "BACK-END APPLICATION COMPUTER SERVER"; Fig. 8A, 820, "COMMUNICATION DEVICE") for receiving, from a client device, an indication of a claim and a request ... specific to the claim (Galia, [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104. The client terminal 107 also includes an input module where a new claim may be filed, and where information pertaining to the claim may be entered, such as a notice of loss, for example," where Galia's request to administer the associated claim corresponds to the instant request specific to the claim, as in [0091]: "FIG. 9 is flowchart 900 illustrating a method of claim administration based upon a claim's predicted likelihood of exhibiting cost volatility, according to one embodiment of the invention. The method begins at step 901, when an insurance company receives a notice of loss");
a complexity determination module (Galia, Fig. 1, 111, "WORK FLOW PROCESSOR") for, responsive to receiving the request (Galia, [0093]: "After a period of time in which additional claim characteristics are collected by the employee assigned to process the claim ( e.g., 30, 45, 60, or 90 days after the notice of loss) the back-end application computer server 103 may access text mined data and other data at step 907 to calculate a claim likelihood of volatility," where Galia's claim initiation plus arbitrary window of time corresponds to the instant responsive to receiving), inputting data of the claim (Galia, [0046]: "The predictive model 104 is used by the back-end application computer server 103 to estimate the likelihood that a claim will exhibit increased volatility in comparison to other claims" and [0049]: "Some examples of characteristics that might be considered in connection with a volatility prediction include ... estimated incurred (reserved amount) at end of evaluation period, estimated total medical spend, ... estimated indemnity payment," where Galia's costs associated with a claim correspond to the instant data of the claim) into a supervised (Galia, [0050]: "In some embodiments, the predictive model 104 is trained on a collection of data known about prior insurance claims and their ultimate disposition, including, for example, and without limitation, the types of costs described above") machine learning model (Galia, [0047]: "The predictive model 104 may be a ... neural network") and receiving as output from the supervised machine learning model a complexity of the claim (Galia, [0063]: "FIG. 6 is a flowchart of a method 600 of using the predictive model generated in FIG. 5 to obtain a future volatility prediction on a particular test claim," where Galia's predicted claim volatility corresponds to the instant a complexity output from the model);
a cluster identification module for, responsive to receiving the request (Fig. 1, 150, "TEXT MINING PLATFORM"), inputting the data of the claim into an unsupervised machine learning model (Galia, [0079]: "the text mining described herein may be associated with insight discovery wherein unsupervised data mining techniques may be used" and [0073] "The pulled data may be associated with, for example, various insurance applications and/or data types 720, such as claim handler notes, loss descriptions, injury descriptions, FNOL [first notice of loss] statements, call transcripts, and/or OCR documents") and receiving as output from the unsupervised machine learning model an identification of a cluster of candidate claims to which the claim belongs (Galia, [0100]: "FIG. 11 is a claim volatility tool machine learning cluster analysis example display.... In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times") ... wherein the complexity of the claim is generated independent of the cluster of candidate claims (Galia, [0056]: "Defining the target variable begins ... by collecting monthly medical expenditure data for each of a group of claims. ... The monthly medical expenditure of each claim may take the form of a time series, such as the time series plotted in the graphs 300 of FIG. 3A" and [0057]: "At step 402, the expenditure data for each claim or for groups of claims is analyzed to produce values for volatility-indicative variables, i.e., statistical characteristics deemed to be evidence of volatility.... At step 404, the fuzzy variables are aggregated to determine a single variable that is representative of degree of volatility," where Galia's time-series expenditure data is used to calculate volatility degree); and
an integration module (Galia, [0100]: "FIG. 11 is a claim volatility tool machine learning cluster analysis example…. clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times (e.g., at initiation, at 30 days, etc.)") for, responsive to receiving the request: combining the complexity and the identification of the cluster into a combined result (Galia, [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times (e.g., at initiation, at 30 days, etc.). For each, the display 1100 graphically indicates if there is no strong correlation, a strong negative correlation, or a strong positive correlation," where Galia's depicted degree of correlation of volatility based according to reserved/paid status corresponds to the instant combined result), wherein the combined result is generated via escalation potential (Galia, Fig. 9, which depicts block 907: "Mine Text And Other Data (During Evaluation Period) And Calculate Claim Likelihood Of Volatility" followed optionally by block 909: "Reassign Claim Based On Likelihood" and by block 911: "Customized Processing Based Upon Likelihood," where Galia's volatility score corresponds to the instant combined result, and Galia's re-assignment based on volatility likelihood corresponds to via escalation potential, as in: [0071]: "the workflow processor 111 assigns more aggressive medical care and review to claims having higher likelihoods of being volatile claims, thereby applying resources to those that might benefit the most" and [0105]: "there is a population of volatile claims whose outcomes vary significantly from their expectations. It is these claims that can drive the total severity, and the predictive model described herein may deliver an ability to predict volatility (or lack of volatility) and facilitate early intervention via claim prioritization, escalation, and/or re-assignment");
identifying ... a matrix corresponding to the combined result (Galia, Fig. 11, depicting the user interface for a "Claim Volatility Tool," which comprises a matrix relating indemnity reserved/paid values to predicted claim volatility by a degree of correlation, and [0100]: "In the illustration of FIG. 11, eight clusters of claims 1110 are displayed for different indemnity reserved/paid values at various times"); and
providing, for display at the client device ... a display of the matrix (Galia, [0100]: "Fig. 11 is a claim volatility tool machine learning cluster analysis example display 1100.... According to some embodiments, a user might select a graphically displayed element to see more information about that element" where [0070]: "The client terminal 107 includes a computer that has a CPU, display, memory and input devices such as a keyboard and mouse. The client terminal 107 also includes a display and/or a printer for outputting the results of the analysis carried out by the predictive model 104").
Galia may not explicitly teach receiving, from a client device ... a request for an analysis specific to the claim; identifying a cell in a matrix ... having a probability curve corresponding to the claim; providing ... an identification of the cell, the cell to be emphasized to a user within a display of the matrix, the cell emphasized because it corresponds to the claim , other cells of the matrix having respective probability curves that are not specific to the claim.
However, Machnicki teaches:
receiving, from a client device (Machnicki, [0138]: "According to some embodiments, the request to analyze the third insurance claim is received via a GUI of a mobile computing device," where Machnicki's mobile device corresponds to the instant client device), ... a request for an analysis specific to the claim (Machnicki, [0138]: "the method may further comprise generating, by the processing device and based on the third liability data for the third insurance claim, the determination of the third insurance claim. According to some embodiments, the determination of the third insurance claim comprises an indication of whether the third insurance claim should be allowed or denied," where Machnicki's determination of allowability of the third claim corresponds to the instant analysis specific to the claim);
identifying a cell in a matrix (Machnicki, [0088]: "The total premium calculated for a potential insurance policy offering ... may, to continue the example, be graded between 'B' and 'C' (e.g., at 810 of FIG. 8) or between 'Fair' and 'Average'. The resulting combination of risk score and premium rating may be plotted on the risk matrix 900, as represented by a data point 904 shown in FIG. 9," where Machnicki's combination plotted on the matrix corresponds to the instant identifying, and where Machnicki's quadrant containing figure element 1004 from Fig. 9 corresponds to the instant cell) ... having a probability curve corresponding to the claim (Machnicki, [0085]: "One example of how the risk matrix 900 may be output and/or implemented with respect to VED [Virtual Engineering Data] of an account and/or group of objects will now be described. ... Typical risk metrics ... may be utilized to produce expected loss frequency and loss severity distributions," where Machnicki's loss severity distribution corresponds to the instant probability curve, and where the distribution corresponds to a claim by way of Machnicki's associated engineering data);
providing ... an identification of the cell (Machnicki, [0084]: "The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher," where quadrant 902b is identified in Fig. 9 as "Good Money Despite Risk"), the cell to be emphasized to a user within a display of the matrix, the cell emphasized because it corresponds to the claim (Machnicki, [0088]: "The data point 904, based on the VED-influenced risk score and the corresponding VED-influenced premium calculation, is plotted in the second quadrant 902b," where Machnicki's point plotted in the second quadrant corresponds to the instant cell emphasized as corresponding to the claim), other cells of the matrix having respective probability curves (Machnicki, [0084]: "the risk matrix 900 may comprise four (4) quadrants 902a-d .... The first quadrant 902a represents the most desirable situations where risk scores are low and premiums are highly graded. The second quadrant 902b represents less desirable situations where, while premiums are highly graded, risk scores are higher. ... The third quadrant 902c represents less desirable characteristics of having poorly graded premiums with low risk scores and the fourth quadrant 902d represents the least desirable characteristics of having poorly graded premiums as well as high risk scores") that are not specific to the claim (Machnicki, [0088]: "The data point 904 ... is plotted in the second quadrant 902b, in a position indicating ... the calculated premium is probably large enough to compensate for the level of risk").
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 Galia regarding receiving an indication of a claim and a request specific to the claim from a client device with those of Machnicki regarding receiving a request for an analysis specific to the claim; identifying a cell in a matrix having a probability curve corresponding to the claim; providing an identification of the cell, the cell to be emphasized to a user within a display of the matrix, the cell emphasized because it corresponds to the claim , other cells of the matrix having respective probability curves that are not specific to the claim.
The motivation to do so would be to facilitate evaluating a level of risk and other processing related to an insurance claim based on related engineering data (Machnicki, [0062]: "such engineering data 602a-n may comprise data indicative of a level of risk ... at the time of casualty or loss (e.g., as defined by the one or more claims). Information on claims ... to update, improve, and/or enhance these procedures and/or associated software and/or devices. In some embodiments, engineering data 602a-n may be utilized to determine, inform, define, and/or facilitate a determination or allocation of responsibility related to a loss").
The Galia/Machnicki combination does not explicitly teach inputting the data of the claim into an unsupervised machine learning model ... in parallel to inputting the data of the claim into the supervised machine learning model.
However, Liao teaches:
inputting the data of the claim into an unsupervised machine learning model ... in parallel to inputting the data of the claim into the supervised machine learning model (Liao, Fig. 2, depicting supervised model 132 and unsupervised model 140 receiving data in parallel, and [0041]: "The data preprocessing module 124 provides application data to one or more models .... [A]pplication data is provided to one or more loan models 132 that generate data indicative of fraud based on application and applicant data. ... The data preprocessing module 124 can also provide application data to one or more entity models 140 that are configured to identify fraud based on data associated with entities involved in the processing of the application," where Luo's loan models 132 are supervised, as in Fig. 3, Supervised Model 170, and entity models 140 are unsupervised, as in [0063]: "FIG. 4 is a functional block diagram illustrating examples of the entity models 140 in the fraud detection system 100. ... [I]n one embodiment, an unsupervised model, e.g., a clustering model such as k-means, is applied to risk indicators for historical transactions for each entity").
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 Galia/Machnicki combination regarding inputting data of the claim into a supervised machine learning model and an unsupervised machine learning model with those of Liao regarding inputting the data of the claim into an unsupervised machine learning model in parallel to inputting the data of the claim into the supervised machine learning model.
The motivation to do so would be to facilitate performing fraud analysis of accounts under circumstances where ensuring accuracy and scaling with large volumes of data are improved by the combination of results of the types of models (Liao, [0028]: "Existing fraud detection systems can use transaction data in addition to data related to the transacting entities to identify fraud. ... However, the fraud detection capabilities of existing systems have not kept pace with either the types of fraudulent activity that have evolved or increasing processing and storage capabilities of computing systems" and [0029]: "it has been found that ... fraud detection can be improved by using stored past transaction data in place of, or in addition to, summarized forms of past transaction data. In addition, in one embodiment, fraud detection can be improved by using statistical information that is stored according to groups of individuals that form clusters").
Regarding Claim 20, the rejection of Claim 19 is incorporated.
The Galia/Machnicki/Liao combination teaches:
wherein the system further comprises an escalation determination module (Galia, Fig. 1, 111, "WORK FLOW PROCESSOR") for determining the escalation potential of the claim (Galia, [0105]: "it is these claims that can drive the total severity, and the predictive model described herein may deliver an ability to predict volatility (or lack of volatility) and facilitate early intervention via claim prioritization, escalation, and/or re-assignment.") wherein the integration module is further for weighting the complexity based on the determined escalation potential (Galia, [0042]: " The insurance applications might be associated with, for example … volatile claim detection…. Note that the transmitted indication might be used to trigger an insurance application … and/or update an insurance application (e.g., by updating a variable or weighing factor of a predictive model).").
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ROBERT N DAY whose telephone number is (703)756-1519. The examiner can normally be reached M-F 9-5.
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, Kakali Chaki can be reached at (571) 272-3719. 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.
/R.N.D./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122