Patent Application 17485448 - Method and System for Medical Malpractice - Rejection

Patent Application 17485448 - Method and System for Medical Malpractice

Title: Method and System for Medical Malpractice Insurance Underwriting Using Value-Based Care Data

Application Information

• Invention Title: Method and System for Medical Malpractice Insurance Underwriting Using Value-Based Care Data
• Application Number: 17485448
• Submission Date: 2025-05-16T00:00:00.000Z
• Effective Filing Date: 2021-09-26T00:00:00.000Z
• Filing Date: 2021-09-26T00:00:00.000Z
• National Class: 706
• National Sub-Class: 012000
• Examiner Employee Number: 99801
• Art Unit: 3626
• Tech Center: 3600

Rejection Summary

• 102 Rejections: 0
• 103 Rejections: 3

Cited Patents

No patents were cited in this rejection.

Office Action Text

    DETAILED ACTION
Notice of Pre-AIA  or AIA  Status
1.	The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
2.	This nonfinal rejection is in response to claims filed on 09/26/2021. Claims 1-20 are pending and are examined herein.
Priority
3.	The examiner acknowledges the applicant’s claim for continuation in part status to parent application 16/595,864. 
The later-filed application must be an application for a patent for an invention which is also disclosed in the prior application (the parent or original nonprovisional application or provisional application). The disclosure of the invention in the parent application and in the later-filed application must be sufficient to comply with the requirements of 35 U.S.C. 112(a) or the first paragraph of pre-AIA  35 U.S.C. 112, except for the best mode requirement. See Transco Products, Inc. v. Performance Contracting, Inc., 38 F.3d 551, 32 USPQ2d 1077 (Fed. Cir. 1994). Support for claims 9-20 have been found in the parent application, therefore claims 9-20 have the effective priority date of 10/08/2019.
However, the disclosure of the prior-filed application, Application No. 16/595,864, fails to provide adequate support or enablement in the manner provided by 35 U.S.C. 112(a) or pre-AIA  35 U.S.C. 112, first paragraph for one or more claims of this application. Claims 1-8 do not have support in the parent application 16/595,864, since the parent application fails to disclose at least:
-the number of training data points
-the number of iterations
-a third data set
-cleaning by normalizing said provider data set.
Therefore, claims 1-8 have the effective priority date of the instant application, which is 09/26/2021. 
Specification
4.	The disclosure is objected to because of the following informalities:
-Paragraph [000114] recites “At Step 620,” which does not seem to match the actual sequence of steps. The examiner presumes the applicant intended for this paragraph to read “At step 618.” Appropriate correction is required.
-The disclosure is objected to because it contains an embedded hyperlink and/or other form of browser-executable code in at least paragraphs [0003], and [0085]. Applicant is required to delete the embedded hyperlink and/or other form of browser-executable code; references to websites should be limited to the top-level domain name without any prefix such as http:// or other browser-executable code. See MPEP § 608.01.
Drawings
5.	The drawings are objected to as failing to comply with 37 CFR 1.84(p)(5) because they include the following reference character(s) not mentioned in the description:
Fig. 3 uses reference number “308” to label “patient satisfaction.” However, [0082] uses 310 to indicate patient satisfaction scores, however 310 is already used to indicate the outcome data. The examiner presumes that 308 is the correct reference to indicate patient satisfaction scores in Fig. 3.  Corrected drawing sheets in compliance with 37 CFR 1.121(d), or amendment to the specification to add the reference character(s) in the description in compliance with 37 CFR 1.121(b) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Objections
6.	Claim 1 is objected to because of the following informalities:
-Line 7 reads “second data points,” however, the examiner presumes the applicant intended for it to read “second training data points,” which would be more consistent with the “said second training data points” in the paragraphs following it.
-Line 10 reads, “wherein said… score is based hospital readmissions…” The examiner presumes that the applicant intended for the line to read “is based on…” which would be more grammatically correct. 
-Line 20 reads, “patient complaints, from medial staff…” The examiner presumes that the applicant intended for the line to read “patient complaints from medical staff,” which would be more grammatically correct.

Appropriate correction is required.
Claim Rejections – 35 USC § 101
7.	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.


8.	Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
	Step 1: Is the claim to a Process, Machine, Manufacture, or Composition of Matter?
	Claims 1-8: A computer-implemented method comprising:
	Claims 9-14: A system…comprising: a processor; and a memory storing computer program instructions, which when executed by the processor cause the processor to perform operations comprising:
	Claims 15-20: A non-transitory computer-readable medium storing computer program instructions, which when executed by a processor cause the processor to perform operations comprising:
	Claims 1-8 recite a method which falls under the potentially eligible subject matter category, “process.” Claims 9-14 recites a system comprising processors and memory, which is an apparatus claim and falls under the category “machine.” Claims 15-20 recite a non-transitory computer-readable medium which falls under the category at least “manufacture.”
Step 2a Prong 1: Is the claim directed to a Judicial Exception(A Law of Nature, a Natural Phenomenon (Product of Nature), or An Abstract Idea?)
The claims under the broadest reasonable interpretation in light of the specification are analyzed herein. Representative claims 1, 9 and 15 are marked up, isolating the abstract idea from additional elements, wherein the abstract idea is in bold and the additional elements have been italicized as follows:
Claim 1: A computer-implemented method comprising: securing at least 16 first training data points wherein said first training data points are related to medical malpractice claims; securing at least 16 second training data points wherein said second training data points are related to known payment outcomes related to said medical malpractice claims; combining said first training data points and second data points into a first data set; cleaning said first data set; securing at least 16 value-based care data points wherein said value-based care score is based hospital readmissions patient satisfaction scores, outcome data and billing/coding/staging data; and said hospital readmissions patient, said satisfaction scores, said outcome data and said billing/coding/staging are directly associated with said first training data points and said second training data points; combining said value-based care data points into a second data set; cleaning said second data set; securing at least 16 social factor data points wherein said social factor data points are related to credit score, change in income, change in personal spending habits, civil actions, criminal actions, regulatory actions, patient complaints, from medial staff and patient complaints from medical administration staff; wherein said credit score, said change in income, said change in personal spending habits, said civil actions, said criminal actions, said regulatory actions, said patient complaints from medical staff and said patient complaint from medical administration staff are directly associated with first and second data points; combining said social value data points into a third data set; cleaning said third data set; training a machine-learning based predictive model to predict a risk of a medical malpractice claim by having a computer update said machine-learning based predictive model by iterative training sessions using said first data set, said second data set, and said third data set; wherein a computer will run no less than three iteration sessions of said machine- learning based predictive model, and no more than 200 iteration sessions of said machine-learning based predictive model for said training; wherein said iteration is an update to an algorithmic parameter; wherein said computer will stop running said training when said risk of a medical malpractice claim for the last two iteration sessions differs by two percent or less; retrieving a provider data set, wherein said provider data set includes provider data points, value-based care data points, and provider social data points; cleaning said provider data set wherein said cleaning normalizes said provider data set to be compatible with said first data set, said second data set, and said third data set; inputting said provider data set into said trained machine-learning based predictive model; predicting, using said trained machine-learning based predictive model, a risk score indicating said risk of a medical malpractice claim for the provider based on the input provider data set; and determining a premium for medical malpractice insurance for the provider based on said risk score predicted using said trained machine-learning based predictive model.
Claim 9: A system for determining a premium for medical malpractice insurance for a provider based upon a predicted risk score using a trained machine-learning based predictive model, comprising: a processor; and a memory storing computer program instructions, which when executed by the processor cause the processor to perform operations comprising: training said machine-learning based predictive model to predict a risk of a medical malpractice claim based on training cases with known outcomes and associated training provider data sets including value-based care data and social factor data; retrieving said provider data set including value-based care data and social data for said provider; inputting said provider data set into said trained machine-learning based predictive model; predicting, using said trained machine-learning based predictive model, a risk score indicating said risk of said medical malpractice claim for said provider based on said provider data set input; and determining said premium for medical malpractice insurance for said provider based on said predicted risk score using said trained machine-learning based predictive model.

Claim 15: A non-transitory computer-readable medium storing computer program instructions, which when executed by a processor cause the processor to perform operations comprising: training a machine-learning based predictive model to predict a risk of a medical malpractice claim based on training cases with known outcomes and associated training provider data sets including value-based care data and social factor data; retrieving a provider data set including value-based care data and social data for a provider; inputting the provider data set into the trained machine-learning based predictive model; predicting, using the trained machine-learning based predictive model, a risk score indicating a risk of a medical malpractice claim for the provider based on the input provider data set; and determining a premium for medical malpractice insurance for the provider based on the risk score predicted using the trained machine-learning based predictive model.
When evaluating the bolded limitations of the claims under the broadest reasonable interpretation in light of the specification, it is clear that representative claims 1, 9, and 15 recite an abstract idea within the category “certain methods of organizing human activity.” More specifically, the present invention falls under the sub-grouping “fundamental economic principles or practices (including hedging, insurance, mitigating risk) which is outlined in MPEP 2106.04(a)(2)(II)(A). The bolded limitations above recite the process of calculating risk and determining an insurance premium based on that risk evaluation. The steps which involve training a model are still part of the abstract idea because the model is used to calculate risks, and the steps are no more than mere data gathering, data processing, or data manipulation steps to perform the abstract idea of calculating and predicting risk. The examiner notes that the MPEP states, “The term "fundamental" is not used in the sense of necessarily being "old" or "well-known." Therefore, despite the use of a newer technology such as machine learning, this considered an additional element and is evaluated in the next step. The overall concept at hand is still fundamentally tied to predicting risk and calculating an insurance premium. Whether this is done with the intended use of calculating malpractice insurance or done using machine learning, it still recites a fundamental economic practice. 

Step 2A Prong 2: Does the claim recite additional elements that integrate the judicial exception into a practical application?
Claims 1, 9, and 15 recite the following additional elements:
-computer in claims 1, 9, and 15
-machine learning in claims 1, 9, and 15
-processor in claims 9, 15
-memory in claims 9, 15
-non-transitory computer readable medium in claim 15
The additional elements listed above, when considered individually and in combination with the claim as a whole, no more than a recitation of the words “apply it” (or an equivalent) or mere instructions to implement an abstract idea or other exception on generic computing components as outlined in MPEP 2106.05(f). In this case, the abstract idea of “calculating risk and determining an insurance premium” is being performed on generic computing components such as a computer, machine, processor, memory, and non-transitory computer readable medium. It is clear in the specification in at least paragraphs [0059] and [0060] that the computing infrastructure described is not an improved or specialized computing infrastructure, but encompasses a wide array of generic computing devices that perform the abstract idea as software functions. Therefore, the claims are no more than a mere recitation of apply it, which does not provide integration into a practical application.
In addition, the use of machine learning to perform the risk calculation is merely a general link to particular technological environment or field of use as outlined in MPEP 2106.05(h). In this case, the claims merely recite data collection, processing, and output steps in order to develop a predictive model to calculate risk, merely limiting the model to be a “machine learning” model. Simply reciting the field of use that the abstract idea is to be performed on is merely a general link. Furthermore, no particular algorithm or improvements to the field of machine learning have been recited, as required in MPEP 2106.05(a) as a consideration to possibly being an integration into a practical application. All the steps are merely data gathering, processing, and outputting steps without a specific implementation of machine learning. Specifying the types of data to be used, the amount of iterations to be performed, or that the data goes through “cleaning” does not describe the claims in any meaningfully way in order to consider it a practical application. Therefore, the claims recite an abstract idea without integration into a practical application.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception?
Claims 1, 9, and 15 recite the following additional elements:
-computer in claims 1, 9, and 15
-machine learning in claims 1, 9, and 15
-processor in claims 9, 15
-memory in claims 9, 15
-non-transitory computer readable medium in claim 15

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, the additional element of using a computer, machine, processor, memory, and non-transitory computer readable medium to perform the steps associated with fundamental economic practices such as calculating risk and determining an insurance premium, amounts to no more than mere instructions to apply the exception using generic computer components. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Furthermore, restricting the predictive model to be a machine learning model is no more than a general link to the technological environment or field of use of “machine learning.” No improvements to computer, the computing infrastructure, or to the field of “machine learning” have been shown as outlined in MPEP 2106.05(f) and MPEP 2106.05(h). Accordingly, even when viewed as a whole, nothing in the claim adds significantly more (i.e. an inventive concept) to the abstract idea. Thus claims 1, 9 and 15 are not patent eligible because the claims are directed to an abstract without significantly more.

Regarding dependent claims 2-8, 10-14, and 16-20:
Claims 2-4, 10-12, and 16-18 merely further limit the abstract idea by restricting elements of the abstract idea into further specified categories that are still part of the abstract idea. For example, claims 2, 10, 16 limit “value based care data in the provider data set” to include “patient satisfaction scores, quality metrics, procedure outcome data, hospital readmission data, and utilization data.” Similarly claims 3, 11, and 17 limit the social factor data to be associated with the provider or patients of the provider and claims 4, 12, 18 limit the data to include credit score data, income data…etc. This is more of the same abstract of “calculating risk and determine an insurance premium” because it is merely reciting the types of data or data sources used to calculate the risk. Furthermore, there are no additional elements to consider therefore the claims are directed to an abstract without integration into a practical application or significantly more. 
Claims 5, 6, 13, and 19 further define the abstract idea by adding steps to the training aspect of the abstract idea. These steps include using training the model based on known outcomes and labeling them as positive cases or negative cases, cleaning the data sets to perform imputation. These steps are more of the same abstract idea of “calculating risk and determine an insurance premium” because they merely add more data processing and data gathering steps which are still steps towards performing risk calculations and insurance premium underwriting. The additional element “machine learning” even when considered individually or in combination with the claims they depend on, is still a general link because the steps in the dependent claims are still steps found within generic machine learning calculations. Using positive and negative data sets to train a model, and perform data cleaning on the model are steps that are inherent to the field of machine learning, therefore it is still a general link with no improvement to the field (see MPEP 2106.05(h) and MPEP 2106.05(a)). Therefore the claims are directed to an abstract without integration into a practical application or significantly more.
Claims 7, 14, and 20 further limit the abstract idea because the claims do not add any additional steps to the abstract idea therefore the claims still recite the same abstract idea of “calculating risk and determine an insurance premium.” All that is being done within these claims is further limiting an additional element “machine learning” to “deep neural networks” which is still another field of use within machine learning. When considering the claims individually, or in combination with the claims they depend on, substituting machine learning for “deep neural networks” is still a general link to a field of use(outlined in MPEP 2106.05(h)). Similarly, the steps are still generic steps inherent to the implementation of deep neural networks to perform the abstract idea, without an improvement to the field of “deep neural networks.” Therefore the claims are still directed to an abstract without integration into a practical application or significantly more.
Claim 8 further defines the abstract idea by adding the following steps(also outlined to indicate the abstract idea and additional elements):
-The method of claim 1, further comprising: training a second machine-learning based predictive model to predict a risk of a stop loss claim based on training cases with known outcomes and associated training provider data sets including value-based care data and social factor data; inputting a second provider data set, including value-based care data and social data for the provider, to the trained second machine-learning base predictive model; and predicting, using the trained second machine-learning based predictive model, a second risk score indicating a risk of a stop loss insurance claim for the provider based on the input second provider data set; wherein determining a premium for medical malpractice insurance for the provider based on the risk score predicted using the trained machine-learning based predictive model comprises: determining a combined premium for medical malpractice insurance and stop loss insurance for the provider based on the risk score predicted using the trained machine-learning based predictive model and the second risk score predicted using the trained second machine-learning based predictive model.
These steps, when considered individually, or combination with the abstract idea are still an abstract idea within the categories of “certain methods of organizing human activity” further subcategorized into “fundamental economic practices” including mitigating risks and insurance. Adding a second predictive model to predict risk of stop loss claims is still more of the same abstract idea because it is part of the risk calculation and insurance premium determination process. This second model being machine learning is still a general link to the field of use of “machine learning” since it is just using two different machine learning models in tandem, which is not considered an improvement to the field of machine learning. Therefore, the claims are still directed to an abstract idea without integration into a practical application or significantly more. 

Claim Rejections - 35 USC § 103
9.	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.

10.	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.
11.	Claims 9-12 and 15-18 are rejected under 35 U.S.C. 103 as being unpatentable over Besman et al. (US 20160171619 A1) hereinafter Besman in view of Michael S Woods (US 20060074708 A1) hereinafter Woods. 

Regarding Claim 9:
Besman discloses a calculating an insurance premium based on a prediction of the risk using neural networks. Besman teaches:
-A system for determining a premium for insurance for a provider (Besman [0003] According to some embodiments, systems, methods, apparatus, computer program code and means may facilitate underwriting decisions. In some embodiments, account information may be received in connection with a potential insurance policy. An account score matrix may be received for the potential insurance policy, including grade values comparing the account information to other insured parties, along with a benchmark premium value calibrated to a target return on equity based on the account information and information in a risk database. A set of guide indication adjustments may then be received from an underwriter for the potential insurance policy. A premium indication model application may dynamically calculate, in substantially real time, an adjusted premium value for the potential insurance policy calibrated to the target return on equity based at least in part on associated guide indication adjustments, and an indication associated with the dynamically calculated adjusted premium value may be transmitted. [0038] At 304, an account score “matrix” for the potential insurance policy may be received from a risk score model application. As used herein, the term “matrix” may refer to any set of scores, including a score card listing one or more account grades, categories, and/or numeric values. The account score matrix may, for example, include grade values comparing the account information to other insured parties.) The broadest reasonable interpretation for provider is any insured party.

- based upon a predicted risk score using a trained machine-learning based predictive model, (Besman [0027] According to some embodiments, the risk score platform 110 and/or premium indication portal 150 may retrieve information from an insurance policy database, an underwriter database, and/or a claim database. In some embodiments, the risk score platform 110 and/or premium indication portal 150 may also receive information from a third party platform (e.g., when a potential insurance policy is associated with automobile insurance, some information may be copied from a state department of motor vehicles platform). [0031] According to some embodiments, an “automated” risk score platform 110 and/or premium indication portal 150 may facilitate underwriting decisions. For example, the risk score platform 110 and/or premium indication portal 150 may automatically output account scores, score matrixes, premium values, etc. to an underwriter device. As used herein, the term “automated” may refer to, for example, actions that can be performed with little (or no) intervention by a human. Moreover, any of the embodiments described herein may be “dynamically” performed by monitoring parameters and/or automatically updating the risk score platform 110 and/or premium indication portal 150 in substantially real time. [0069] Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications. The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled. [0081] A function of the output device 1922 may be to provide an output that is indicative of (as determined by the trained predictive model application component 1918) particular risk evaluation data (account scores and score matrixes) and/or underwriting data (e.g., benchmark or guide premium values).) Besman’s platform includes a “risk score” platform, which is taught to be generated by a predictive model in [0081], which can be trained using neural networks ([0069]). Therefore, the limitation has been taught. 

-comprising: a processor; and a memory storing computer program instructions, which when executed by the processor cause the processor to perform operations comprising: (Besman [0063] The processor 1710 also communicates with a storage device 1730. The storage device 1730 may comprise any appropriate information storage device, including combinations of magnetic storage devices (e.g., a hard disk drive), optical storage devices, mobile telephones, and/or semiconductor memory devices. The storage device 1730 stores a program 1712 and/or an underwriting engine 1714 for controlling the processor 1710. The processor 1710 performs instructions of the programs 1712, 1714, and thereby operates in accordance with any of the embodiments described herein.)

-training said machine-learning based predictive model to predict a risk of a claim based on training cases with known outcomes and (Besman [0069] Thus, embodiments described herein may facilitate underwriting decisions. According to some embodiments, one or more predictive model applications may be used in connection with the underwriting processes. As used herein, the phrase “predictive model application” might refer to, for example, any of a class of algorithms that are used to understand relative factors contributing to an outcome, estimate unknown outcomes, discover trends, and/or make other estimations based on a data set of factors collected across prior trials. Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications. The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled. [0071] As described in more detail below, the historical claim transaction data 1904 is employed to train a predictive model application to provide an output that indicates how a claim should by assigned to claim handler, and the current claim transaction data 1906 is thereafter analyzed by the predictive model application. Moreover, as time goes by, and results become known from processing current claim transactions, at least some of the current claim transactions may be used to perform further training of the predictive model application) Since Besman trains the model based on “results that have become known” which is mapped to “training cases with known outcomes,” the limitation has been taught. 

-associated training provider data sets including data and social factor data; (Besman [0053] The account score matrix 950 might be based on various risk factors depending on the line of business associated with the potential insurance policy. For example, a workers’ compensation insurance policy might have an account score matrix 950 listing payroll size, average wage, prior indemnity claim frequency, geographic, industry classification, prior medical claim frequency, an overall number of locations, and or business credit risk factors. Other types of insurance policies may have account score matrixes including other risk factors, such as: a number of years in business, a fleet size, driver age information, vehicle weight information, an overall exposure size, an exposure type, a building age, and weather related data. Note that an account score matrix 950 might assign different weights to different risk factors to determine an overall account score. [0072] Either the historical claim transaction data 1904 or the current claim transaction data 1906 might include, according to some embodiments, determinate and indeterminate data. As used herein and in the appended claims, “determinate data” refers to verifiable facts such as the date of birth, age or name of a claimant or name of another individual or of a business or other entity; a type of injury, accident, sickness, or pregnancy status; a medical diagnosis; a date of loss, or date of report of claim, or policy date or other date; a time of day; a day of the week; a vehicle identification number, a geographic location; and a policy number. [0073] As used herein and in the appended claims, “indeterminate data” refers to data or other information that is not in a predetermined format and/or location in a data record or data form. Examples of indeterminate data include narrative speech or text, information in descriptive notes fields and signal characteristics in audible voice data files. Indeterminate data extracted from medical notes or accident reports might be associated with, for example, an amount of loss and/or details about how an accident occurred.) These training data sets associated with the policy holder(claimant) include social factor data which is defined by the instant application to include factors such as income, family circumstances, demographics, etc. 

-retrieving said provider data set including data and social data for said provider; (Besman [0038] At 304, an account score “matrix” for the potential insurance policy may be received from a risk score model application. As used herein, the term “matrix” may refer to any set of scores, including a score card listing one or more account grades, categories, and/or numeric values. The account score matrix may, for example, include grade values comparing the account information to other insured parties. For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential insurance policy.) “Receiving” a matrix, which includes risk variables from the risk database, is mapped to “retrieving said provider data set.”

-inputting said provider data set into said trained machine-learning based predictive model; (Besman [0063] For example, the processor 1710 may receive account information in connection with a potential insurance policy. The processor may also receive, from a risk score model application, an account score matrix for the potential insurance policy, including grade values comparing the account information with other insured policies in a risk database, along with a benchmark premium value calibrated to a target return on equity based on the account information and information in the risk database. The account score matrix may be displayed by the processor 1710 on an underwriter device, and guide indication adjustments may be received from the underwriter device for the potential insurance policy. The processor 1710 may then automatically calculate an adjusted premium value calibrated to the target return on equity based at least in part on the guide indication adjustments. [0078] A function of the predictive model application component 1918 may be to facilitate risk evaluation and/or underwriting decisions. The predictive model application component may be directly or indirectly coupled to the data storage module 1902.) In Besman, the processor receives account information, which eventually is fed to the predictive model through the data storage module to facilitate risk evaluation.

-predicting, using said trained machine-learning based predictive model, a risk score indicating said risk of said claim for said provider based on said provider data set input; (Besman [0059] An underwriter analyst or other party may then use the account score application to determine, at 1420, claim frequency, a summary risk score, an account score matrix, and a benchmark input factor by line of business. [0069] The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled. Both the historical claim transaction data and data representing the current claim transactions might include, according to some embodiments, indeterminate data or information extracted therefrom. For example, such data/information may come from narrative and/or medical text notes associated with a claim file. [0081] In addition, the computer system 1900 may include an output device 1922. The output device 1922 may be coupled to the computer processor 1914. A function of the output device 1922 may be to provide an output that is indicative of (as determined by the trained predictive model application component 1918) particular risk evaluation data (account scores and score matrixes) and/or underwriting data (e.g., benchmark or guide premium values). [0069] Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications. The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled.) 

-and determining said premium for insurance for said provider based on said predicted risk score using said trained machine-learning based predictive model. (Besman [0068] The account score 1806 may reflect a level of risk associated with the potential insurance policy (as determined by a pricing model application based on the account information). The benchmark premium to achieve desired return on equity 1808 may be automatically calculate based on the account score 1806 (and/or other account information). The guide indication adjusted premium to achieve desired return on equity 1810 may represent benchmark premium adjusted in view of the underwriter’s expertise (as reflected by the guide adjustments he or she provided). See also [0081] for predictive model) The determination of a benchmark premium based on account score(which includes risk score), anticipates this limitation. 

However, Besman fails to teach:
-Determining a premium for medical malpractice insurance
-predict a risk of medical malpractice claim
-associated training provider data sets include value based care data.
-the “provider” is a healthcare provider.

Alternatively, Woods discloses a method for grouping healthcare professionals having a low risk of being charged with medical malpractice by using patient satisfaction data, hence qualifying for a reduced malpractice insurance premium. Woods teaches:
-A premium for medical malpractice insurance. (Woods [0010] Medical malpractice insurance premiums constitute a second component of current medical professional liability risk management programs. The threat of having a higher premium increases if a physician has had to settle a malpractice claim made against him or if a judgment had been entered against the physician for malpractice. [0065] A candidate seeking inclusion as a member of a group of physicians having relatively low probability of being charged with medical malpractice and hence qualifying for a reduced malpractice insurance premium contacts an insurer, which only insures physicians having a relatively low probability of being charged with malpractice. To qualify for inclusion as a member of the group of physicians having relatively low probability of being charged with medical malpractice and hence qualifying to be insured by the insurance carrier, the physician arranges for placement of an electronic terminal in the physician's offices whereby that physician's patients may be polled as to their satisfaction with a physician's services.)

-predict a risk of medical malpractice claim(Woods [0046] Once the personality profile has been developed for a particular subject, the particular subject’s profile is compared to profiles for successful people in the particular profession of interest, which in the preferred practice of the invention is medicine. The comparison of the subject’s profile with the profiles of persons who have been demonstrably successful in the profession and the determination of the amount of deviation of the subject's profile from one or more profiles constructed from the average or median scores of persons who have been successful in the profession indicates whether the subject individual is of low risk or high risk, in accordance with the practice of the invention.   [0135] An institution develops a cognitive behavioral profile predictive of risk according to institutionally predefined parameters proceeds by first defining a target population of interest to the institution. Development of the profile proceeds by identifying, within the target population of interest, mutually exclusive groups of high risk and low risk individuals according to said institutionally predefined parameters. The institution then causes the individuals in said high and low risk groups from said target population to undergo standardized cognitive behavioral profiling evaluation, as described above in Prophetic Example D for FIG. 2, in order to delineate and document cognitive behavioral group profiles of the institutionally defined high risk and low risk groups from the target population of interest. The procedure then randomly selects, from the target population of interest, a group of sufficient number that selection of some individuals from both of the institutionally defined high and the low risk groups is statistically assured. [0049] As a result, the physician is initially included as a member of the group of physicians having relatively low probability of being charged with medical malpractice.) 

-associated training provider data sets include value based care data.(Woods [0052] Upon electing one of the two and completing the elected course, patient satisfaction is again monitored preferably via the Internet or other interactive electronic telecommunication means as to satisfaction of that physician’s patients over forward-going time periods. The patient satisfaction levels are compared to predetermined criteria. If the satisfaction level of the candidate-physician’s patients is then high according to the predetermined criteria, the candidate physician is admitted as a member of the group. This new member of the group, like all members of the group, is thereafter monitored preferably via the Internet or other interactive electronic telecommunication means as to the satisfaction of patients over forward-going time periods and the patent satisfaction level is compared to predetermined criteria.) According to the specification, in at least [0071], value based care includes patient satisfaction data, therefore, the limitation has been taught.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by substituting the data sets of Besman with data related to medical malpractice, and value based care by Woods. By performing this simple substitution, the combined invention would result in the predictable outcome of determining a risk for medical malpractice and determining the associated insurance premiums. This is considered obvious to combine because medical malpractice is merely the intended use, and by substituting the training data to relate to medical malpractice, the system would perform the same function of predicting a risk and determining an insurance premium. One would have been motivated to perform this combination as it would provide the benefit of incentivizing low risk medical behavior by offering reduced premiums. (Woods [0014] [0049])

Regarding Claim 15:
Besman teaches:
A non-transitory computer-readable medium storing computer program instructions, which when executed by a processor cause the processor to perform operations comprising: (Besman [0036] FIG. 3 illustrates an underwriting method 300 that might be performed by some or all of the elements of the systems 100, 200 described with respect to FIGS. 1 and 2 according to some embodiments of the present invention. The flow charts described herein do not imply a fixed order to the steps, and embodiments of the present invention may be practiced in any order that is practicable. Note that any of the methods described herein may be performed by hardware, software, or any combination of these approaches. For example, a computer-readable storage medium may store thereon instructions that when executed by a machine result in performance according to any of the embodiments described herein.)
-training a machine-learning based predictive model to predict a risk of a claim based on training cases with known outcomes and  (Besman [0069] Thus, embodiments described herein may facilitate underwriting decisions. According to some embodiments, one or more predictive model applications may be used in connection with the underwriting processes. As used herein, the phrase “predictive model application” might refer to, for example, any of a class of algorithms that are used to understand relative factors contributing to an outcome, estimate unknown outcomes, discover trends, and/or make other estimations based on a data set of factors collected across prior trials. Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications. The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled. [0071] As described in more detail below, the historical claim transaction data 1904 is employed to train a predictive model application to provide an output that indicates how a claim should by assigned to claim handler, and the current claim transaction data 1906 is thereafter analyzed by the predictive model application. Moreover, as time goes by, and results become known from processing current claim transactions, at least some of the current claim transactions may be used to perform further training of the predictive model application) Since Besman trains the model based on “results that have become known” which is mapped to “training cases with known outcomes,” the limitation has been taught.

-associated training provider data sets including data and social factor data; (Besman [0053] The account score matrix 950 might be based on various risk factors depending on the line of business associated with the potential insurance policy. For example, a workers’ compensation insurance policy might have an account score matrix 950 listing payroll size, average wage, prior indemnity claim frequency, geographic, industry classification, prior medical claim frequency, an overall number of locations, and or business credit risk factors. Other types of insurance policies may have account score matrixes including other risk factors, such as: a number of years in business, a fleet size, driver age information, vehicle weight information, an overall exposure size, an exposure type, a building age, and weather related data. Note that an account score matrix 950 might assign different weights to different risk factors to determine an overall account score. [0072] Either the historical claim transaction data 1904 or the current claim transaction data 1906 might include, according to some embodiments, determinate and indeterminate data. As used herein and in the appended claims, “determinate data” refers to verifiable facts such as the date of birth, age or name of a claimant or name of another individual or of a business or other entity; a type of injury, accident, sickness, or pregnancy status; a medical diagnosis; a date of loss, or date of report of claim, or policy date or other date; a time of day; a day of the week; a vehicle identification number, a geographic location; and a policy number. [0073] As used herein and in the appended claims, “indeterminate data” refers to data or other information that is not in a predetermined format and/or location in a data record or data form. Examples of indeterminate data include narrative speech or text, information in descriptive notes fields and signal characteristics in audible voice data files. Indeterminate data extracted from medical notes or accident reports might be associated with, for example, an amount of loss and/or details about how an accident occurred.) These training data sets associated with the policy holder(claimant) include social factor data which is defined by the instant application to include factors such as income, family circumstances, demographics, etc. 

-retrieving a provider data set including value-based care data and social data for a provider; (Besman [0038] At 304, an account score “matrix” for the potential insurance policy may be received from a risk score model application. As used herein, the term “matrix” may refer to any set of scores, including a score card listing one or more account grades, categories, and/or numeric values. The account score matrix may, for example, include grade values comparing the account information to other insured parties. For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential insurance policy.) “Receiving” a matrix, which includes risk variables from the risk database, is mapped to “retrieving said provider data set.”

- inputting the provider data set into the trained machine-learning based predictive model; (Besman [0063] For example, the processor 1710 may receive account information in connection with a potential insurance policy. The processor may also receive, from a risk score model application, an account score matrix for the potential insurance policy, including grade values comparing the account information with other insured policies in a risk database, along with a benchmark premium value calibrated to a target return on equity based on the account information and information in the risk database. The account score matrix may be displayed by the processor 1710 on an underwriter device, and guide indication adjustments may be received from the underwriter device for the potential insurance policy. The processor 1710 may then automatically calculate an adjusted premium value calibrated to the target return on equity based at least in part on the guide indication adjustments. [0078] A function of the predictive model application component 1918 may be to facilitate risk evaluation and/or underwriting decisions. The predictive model application component may be directly or indirectly coupled to the data storage module 1902.) In Besman, the processor receives account information, which eventually is fed to the predictive model through the data storage module to facilitate risk evaluation.

- predicting, using the trained machine-learning based predictive model, a risk score indicating a risk of a claim for the provider based on the input provider data set; (Besman [0059] An underwriter analyst or other party may then use the account score application to determine, at 1420, claim frequency, a summary risk score, an account score matrix, and a benchmark input factor by line of business. [0069] The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled. Both the historical claim transaction data and data representing the current claim transactions might include, according to some embodiments, indeterminate data or information extracted therefrom. For example, such data/information may come from narrative and/or medical text notes associated with a claim file. [0081] In addition, the computer system 1900 may include an output device 1922. The output device 1922 may be coupled to the computer processor 1914. A function of the output device 1922 may be to provide an output that is indicative of (as determined by the trained predictive model application component 1918) particular risk evaluation data (account scores and score matrixes) and/or underwriting data (e.g., benchmark or guide premium values).)

-and determining a premium for insurance for the provider based on the risk score predicted using the trained machine-learning based predictive model. (Besman [0068] The account score 1806 may reflect a level of risk associated with the potential insurance policy (as determined by a pricing model application based on the account information). The benchmark premium to achieve desired return on equity 1808 may be automatically calculate based on the account score 1806 (and/or other account information). The guide indication adjusted premium to achieve desired return on equity 1810 may represent benchmark premium adjusted in view of the underwriter’s expertise (as reflected by the guide adjustments he or she provided). See also [0081] for predictive model) The determination of a benchmark premium based on account score(which includes risk score), anticipates this limitation.

However, Besman fails to teach:
-Determining a premium for medical malpractice insurance
-predict a risk of medical malpractice claim
-associated training provider data sets include value based care data.
-the “provider” is a healthcare provider.

Alternatively, Woods discloses a method for grouping healthcare professionals having a low risk of being charged with medical malpractice by using patient satisfaction data, hence qualifying for a reduced malpractice insurance premium. Woods teaches:
-A premium for medical malpractice insurance. (Woods [0010] Medical malpractice insurance premiums constitute a second component of current medical professional liability risk management programs. The threat of having a higher premium increases if a physician has had to settle a malpractice claim made against him or if a judgment had been entered against the physician for malpractice. [0065] A candidate seeking inclusion as a member of a group of physicians having relatively low probability of being charged with medical malpractice and hence qualifying for a reduced malpractice insurance premium contacts an insurer, which only insures physicians having a relatively low probability of being charged with malpractice. To qualify for inclusion as a member of the group of physicians having relatively low probability of being charged with medical malpractice and hence qualifying to be insured by the insurance carrier, the physician arranges for placement of an electronic terminal in the physician's offices whereby that physician's patients may be polled as to their satisfaction with a physician's services.)

-predict a risk of medical malpractice claim(Woods [0046] Once the personality profile has been developed for a particular subject, the particular subject’s profile is compared to profiles for successful people in the particular profession of interest, which in the preferred practice of the invention is medicine. The comparison of the subject’s profile with the profiles of persons who have been demonstrably successful in the profession and the determination of the amount of deviation of the subject's profile from one or more profiles constructed from the average or median scores of persons who have been successful in the profession indicates whether the subject individual is of low risk or high risk, in accordance with the practice of the invention.   [0135] An institution develops a cognitive behavioral profile predictive of risk according to institutionally predefined parameters proceeds by first defining a target population of interest to the institution. Development of the profile proceeds by identifying, within the target population of interest, mutually exclusive groups of high risk and low risk individuals according to said institutionally predefined parameters. The institution then causes the individuals in said high and low risk groups from said target population to undergo standardized cognitive behavioral profiling evaluation, as described above in Prophetic Example D for FIG. 2, in order to delineate and document cognitive behavioral group profiles of the institutionally defined high risk and low risk groups from the target population of interest. The procedure then randomly selects, from the target population of interest, a group of sufficient number that selection of some individuals from both of the institutionally defined high and the low risk groups is statistically assured. [0049] As a result, the physician is initially included as a member of the group of physicians having relatively low probability of being charged with medical malpractice.) 

-associated training provider data sets include value based care data.(Woods [0052] Upon electing one of the two and completing the elected course, patient satisfaction is again monitored preferably via the Internet or other interactive electronic telecommunication means as to satisfaction of that physician’s patients over forward-going time periods. The patient satisfaction levels are compared to predetermined criteria. If the satisfaction level of the candidate-physician’s patients is then high according to the predetermined criteria, the candidate physician is admitted as a member of the group. This new member of the group, like all members of the group, is thereafter monitored preferably via the Internet or other interactive electronic telecommunication means as to the satisfaction of patients over forward-going time periods and the patent satisfaction level is compared to predetermined criteria.) According to the specification, in at least [0071], value based care includes patient satisfaction data, therefore, the limitation has been taught.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by substituting the data sets of Besman with data related to medical malpractice, and value based care by Woods. By performing this simple substitution, the combined invention would result in the predictable outcome of determining a risk for medical malpractice and determining the associated insurance premiums. This is considered obvious to combine because medical malpractice is merely the intended use, and by substituting the training data to relate to medical malpractice, the system would perform the same function of predicting a risk and determining an insurance premium. One would have been motivated to perform this combination as it would provide the benefit of incentivizing low risk medical behavior by offering reduced premiums. (Woods [0014] [0049])

Regarding Claims 10 and 16:
The combination of Besman and Woods teach the system of claim 9/ The non-transitory computer-readable medium of claim 15,
Besman fails to teach:
- wherein the value-based care data in the provider data set includes one or more of patient satisfaction scores, quality metrics, procedure outcome data, hospital readmission data, or utilization data.
However, Woods teaches:
- wherein the value-based care data in the provider data set includes one or more of patient satisfaction scores. (Woods [0108] These behavioral profiling/patient satisfaction monitoring/physician training/education steps of the process are repeated, preferably via the Internet or other interactive electronic telecommunication means, until either the physician qualifies for the low risk group as a result of high patient satisfaction scores or the process is terminated and the physician is rejected for membership in the group.) 
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by including patient satisfaction scores in the calculation of risk. This is considered obvious to combine because medical malpractice is merely the intended use, and by substituting the training data to relate to medical malpractice, the system would perform the same function of predicting a risk and determining an insurance premium. One would have been motivated to perform this combination as it would provide the benefit of incentivizing low risk medical behavior by offering reduced premiums, and because patient satisfaction data is a good indicator of risk. (Woods [0014] [0017] [0049])

Regarding Claims 11 and 17:
The combination of Besman and Woods teach the system of claim 9/ The non-transitory computer-readable medium of claim 15,
Furthermore, Besman teaches:
-wherein the social factor data in the provider data set includes one or more of social factor data.(Besman [0038] For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential insurance policy. The risk variables might be associated with, by way of examples only, wage information, prior indemnity claim frequency data, geographic information, an industry classification, prior medical claim frequency, business credit, a payroll size, and/or a location count.)

However, Besman fails to teach:
-wherein the social factor data in the provider data set includes one or more of social factor data associated with the (healthcare) provider or social factor data associated with patients of the provider.  
	
	Alternatively, Woods teaches:
-wherein the social factor data in the provider data set includes one or more of social factor data associated with the provider or social factor data associated with patients of the provider. (Woods [0161] It is further to be understood that the while the invention has been disclosed principally discussing determining average levels of customer satisfaction by averaging determined satisfaction levels of responding customers for a given candidate, in certain instances average of level of customer satisfaction may not be the appropriate parameter. For example, older customers may be less easily satisfied by the candidate but may also be less likely than the younger customer to bring an action for professional malpractice. It is within the purview of the invention to apply algorithms to account for such differences in ease of satisfaction among respective age groups, demographic groups, geographic groups and the like and also to apply algorithms to adjust for the ease of satisfaction among, for example in the case of physicians, general practitioners, internists, gastrointerologists, neurosurgeons, neurologists, psychiatrists, dentists, oral surgeons, periodontists, endodontists, dermatologists, orthopedic surgeons, allergists and other professional specialties.) Social factor data associated with patients of the provider includes demographic/age/geographic groups that are patients of the provider, which is taught by Woods.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by accounting for social factors amongst the patients of the healthcare providers as taught by Woods. Modifying the customer satisfaction thresholds that indicate the risk based on such factors would provide the expected result of more accurately determining qualification for the lower premiums associated with a lower risk score. One of ordinary skill in the art would have been motivated to perform this combination as it would provide the benefit of accurately assessing risk. (Woods [0017-0018])

Regarding Claims 12 and 18:
The combination of Besman and Woods teach the system of claim 11/ The non-transitory computer-readable medium of claim 17,
Furthermore, Besman teaches:
-wherein the social factor data associated with the provider includes one or more of credit score data, income data.(Besman [0038] At 304, an account score “matrix” for the potential insurance policy may be received from a risk score model application. As used herein, the term “matrix” may refer to any set of scores, including a score card listing one or more account grades, categories, and/or numeric values. The account score matrix may, for example, include grade values comparing the account information to other insured parties. For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential Insurance policy. The risk variables might be associated with, by way of examples only, wage information, prior indemnity claim frequency data, geographic information, an industry classification, prior medical claim frequency, business credit, a payroll size, and/or a location count.)
- and wherein the social factor data… includes socio-economic data… including one or more of income or assets. (Besman [0038] For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential Insurance policy. The risk variables might be associated with, by way of examples only, wage information, prior indemnity claim frequency data, geographic information, an industry classification, prior medical claim frequency, business credit, a payroll size, and/or a location count.) 

However, Besman fails to teach:
-and wherein the social factor data associated with the patients of the provider includes socio-economic data associated with the patients of the provider

Alternatively, Woods teaches:
-social factor data associated with patients of the providers including socio-economic data associated with the patients of the providers. (Woods [0039] For the exemplary physician patient satisfaction survey results illustrated in Diagrams 1 and 2, the survey was conducted utilizing seventy-eight (78) adult respondent-patients replying to one hundred sixty-one (161) distributed surveys. Of the respondents thirty-six percent (36%) were male, fifty-five percent (55%) were female while the gender of nine percent (9%) of the respondents was not known. The demonstrated higher response rate in females is believed usual for surveys since women tend to have a higher response rate than men. Age range of the respondents was from twenty-five (25) to seventy-four (74) years. A specific ethnic break-down is not available, but it is know that African Americans, Asians, Hispanics and Caucasians participated in this survey, with the population being highly skewed towards Caucasian. Educational backgrounds ranged from high school to doctorate degrees.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by including socio-economic data of the patients of the providers as taught by Woods. One of ordinary skill in the art would have been motivated to perform this combination as it would result in the predictable benefit of using multiple factors to more accurately assess malpractice risk. (Besman [0036])

12.	Claims 13 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Besman et al. (US 20160171619 A1) in view of Woods (US 20060074708 A1) further in view of Kirshenbaum et al. (US 20060248054 A1) hereinafter Kirshenbaum.
Regarding Claims 13 and 19:

The combination of Besman and Woods teach the system of claim 9/ The non-transitory computer-readable medium of claim 15, wherein training a machine-learning based predictive model to predict a risk of a medical malpractice claim based on training cases with known outcomes and associated training provider data sets including value- based care data and social factor data comprises:

Furthermore Besman teaches:
-identifying cases in which providers were subject to claims. (Besman [0027] the risk score platform 110 and/or premium indication portal 150 may retrieve information from an insurance policy database, an underwriter database, and/or a claim database [0080] Still further, the computer system 1900 includes a model application training component 1920. The model application training component 1920 may be coupled to the computer processor 1914 (directly or indirectly) and may have the function of training the predictive model application component 1918 based on the historical claim transaction data 1904. (As will be understood from previous discussion, the model application training component 1920 may further train the predictive model application component 1918 as further relevant claim transaction data becomes available.) In Besman, retrieving data from a claim database, or using historical claim transaction data is an example of identifying cases in which providers were subject to claims. 
-retrieving a training provider data set including data and social factor data (Besman [0038] At 304, an account score “matrix” for the potential insurance policy may be received from a risk score model application. As used herein, the term “matrix” may refer to any set of scores, including a score card listing one or more account grades, categories, and/or numeric values. The account score matrix may, for example, include grade values comparing the account information to other insured parties. For example, the account score matrix might include grade values for each of a plurality of risk variables in the risk database, each grade reflecting a percentage of other insured parties having a level of risk, for the associated risk variable, worse than the potential insurance policy.) “Receiving” a matrix, which includes risk variables from the risk database, is mapped to “retrieving said provider data set.”
-training the machine-learning based predictive model based on the training provider data sets and known outcomes. (Besman [0071] As described in more detail below, the historical claim transaction data 1904 is employed to train a predictive model application to provide an output that indicates how a claim should by assigned to claim handler, and the current claim transaction data 1906 is thereafter analyzed by the predictive model application. Moreover, as time goes by, and results become known from processing current claim transactions, at least some of the current claim transactions may be used to perform further training of the predictive model application. Consequently, the predictive model application may thereby adapt itself to changing claim patterns. [0069] Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications. The predictive model application might be trained with historical claim transaction data, and be applied to current claim transactions to determine how the current claim transactions should be handled.)

However, Besman fails to teach:
-identifying positive training cases in which providers were subject to medical malpractice claims and negative training cases in which providers were not subject to medical malpractice claims;

-retrieving a training provider data set including value-based care data and social factor data for each of the positive training cases and for each of the negative training cases; and 

-training the machine-learning based predictive model based on the training provider data sets and known outcomes of the positive training cases and negative training cases.

Alternatively, Woods teaches:
-medical malpractice claims (Woods [0010] Medical malpractice insurance premiums constitute a second component of current medical professional liability risk management programs. The threat of having a higher premium increases if a physician has had to settle a malpractice claim made against him or if a judgment had been entered against the physician for malpractice.)
-value-based care data (Woods [0052] Upon electing one of the two and completing the elected course, patient satisfaction is again monitored preferably via the Internet or other interactive electronic telecommunication means as to satisfaction of that physician’s patients over forward-going time periods. The patient satisfaction levels are compared to predetermined criteria. If the satisfaction level of the candidate-physician’s patients is then high according to the predetermined criteria, the candidate physician is admitted as a member of the group. This new member of the group, like all members of the group, is thereafter monitored preferably via the Internet or other interactive electronic telecommunication means as to the satisfaction of patients over forward-going time periods and the patent satisfaction level is compared to predetermined criteria.) According to the specification, in at least [0071], value based care includes patient satisfaction data, therefore, the limitation has been taught.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by substituting the data sets of Besman with data related to medical malpractice, and value based care by Woods. By performing this simple substitution, the combined invention would result in the predictable outcome of determining a risk for medical malpractice and determining the associated insurance premiums. This is considered obvious to combine because medical malpractice is merely the intended use, and by substituting the training data to relate to medical malpractice, the system would perform the same function of predicting a risk and determining an insurance premium. One would have been motivated to perform this combination as it would provide the benefit of incentivizing low risk medical behavior by offering reduced premiums. (Woods [0014] [0049])

However, neither Besman nor Woods teach:
-identifying positive training cases in which providers were subject to medical malpractice claims and negative training cases in which providers were not subject to medical malpractice claims;

-retrieving a training provider data set including value-based care data and social factor data for each of the positive training cases and for each of the negative training cases; and 

-training the machine-learning based predictive model based on the training provider data sets and known outcomes of the positive training cases and negative training cases.

Alternatively, Kirshenbaum discloses training a categorizer to categorize positive training cases and negative training cases. Kirshenbaum teaches:
-identifying positive training cases and negative training cases. (Kirshenbaum [0083] As additional categories are added, positive training cases and negative training cases are identified for the additional categories, using the search and confirm processes described above. Also, categories can be modified and deleted.)
-retrieving training data for each of the positive training cases and for each of the negative training cases (Kirshenbaum [0042] The display frame 216 displays a summary (e.g., title) of each of the cases identified by the search based on the query entered in the search frame 212. Note that each case is associated with several pieces of information, with the title being one of the pieces, for example. In other implementations, other pieces of information associated with the cases can be displayed. In some embodiments, the user may separately select which pieces of information are to be displayed, to be used for matching queries, and to be used for training the categorizer. In the example of Fig. 2, the leftmost column 218 of the display frame 216 indicates the category (in text form) of each of the corresponding cases.)
-training the predictive model based on the training data sets and known outcomes of the positive training cases and negative training cases.(Kirshenbaum [0063] Additionally, the displayed information includes the category (or categories) that a user (or the categorizer) has associated with the case (either based on an earlier training set or based on a prediction by the categorizer [0066] In an embodiment, the categorizer can determine whether a matching case should be indicated as belonging to a category by computing a score indicating a confidence level. The score indicating a confidence level is compared to a predefined threshold, and if the score is greater than the predefined threshold, the categorizer identifies the matching case as belonging to the category.) Kirshenbaum’s categorizer is an example of a predictive model trained based on known outcomes of positive training cases and negative training cases.
	Therefore it would have been obvious to one of ordinary skill in the art to further modify Besman in view of Woods by adding the labeling of the training data as either positive or negative training cases as taught by Kirshenbaum. The combination would yield the predictable outcome of identifying positive and negative training cases associated with the insurance claim outcomes, and training the machine learning predictive model based on provider data sets that are labelled as positive or negative training cases. Adding Kirshenbaum’s labelling process, as opposed to using unlabeled training data, provides the advantage of constituting a “correct answer” for each training case, and enable the validation of the training data. (Kirshenbaum [0002])

13.	Claims 14 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Besman et al. (US 20160171619 A1) in view of Woods (US 20060074708 A1) further in view of Lei et al. (US 20190005586 A1) hereinafter Lei.
Regarding Claims 14 and 20:
The combination of Besman and Woods teach the system of claim 9/ The non-transitory computer-readable medium of claim 15,

Furthermore, Besman teaches:
-wherein the machine- learning based predictive model is a neural network.(Besman [0069] Note that a predictive model application might refer to, but is not limited to, methods such as ordinary least squares regression, logistic regression, decision trees, neural networks, generalized linear model applications, and/or Bayesian model applications)

However, neither Besman nor Woods teach:
-wherein the machine- learning based predictive model is a deep neural network.

Alternatively, Lei discloses a method of predicting vehicle insurance risk. Lei teaches:
-wherein the machine- learning based predictive model is a deep neural network.(Lei [0073] In another implementation of the method provided in the present disclosure, a deep neural network can be used for modeling of the target. The vehicle insurance risk prediction algorithm is constructed in the following manner: [0074] S50: Collect a preset type of personal attribute samples; [0075] S52: Classify the personal attribute samples into feature samples in different value ranges based on corresponding feature types; [0076] S54: Extract feature data from the feature samples based on N specified feature types, and generate an N-dimensional discrete feature vector; [0077] S56: Map a single discrete feature vector of the feature samples to an M-dimensional continuous feature vector in a preset manner; and [0078] S58: Concatenate continuous feature vectors corresponding to the N-dimensional discrete features to form an (N*M)-dimensional continuous feature vector X, and use the continuous feature vector X as input of a selected deep neural network to construct a vehicle insurance risk prediction model.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Besman by using deep neural networks as taught by Lei to create the predictive model. Substituting the Besman’s neural network based model with deep neural networks would provide the benefit of providing a more comprehensive and accurate risk assessment. (Lei [0003])

Allowable Subject Matter
14.	Claims 1-8 would be allowable if rewritten or amended to overcome the rejection(s) under 35 U.S.C. 101. The following is a statement of reasons for the indication of allowable subject matter: 
The combination of Besman and Woods, as well as any of the prior art of record fail to teach or suggest the limitations of representative claim 1 including:
- securing at least 16 value-based care data points wherein said value-based care score is based hospital readmissions patient satisfaction scores, outcome data and billing/coding/staging data; and said hospital readmissions patient, said satisfaction scores, said outcome data and said billing/coding/staging are directly associated with said first training data points and said second training data points;
- securing at least 16 value-based care data points wherein said value-based care score is based hospital readmissions patient satisfaction scores, outcome data and billing/coding/staging data; and said hospital readmissions patient, said satisfaction scores, said outcome data and said billing/coding/staging are directly associated with said first training data points and said second training data points;
- wherein a computer will run no less than three iteration sessions of said machine- learning based predictive model, and no more than 200 iteration sessions of said machine-learning based predictive model for said training
- wherein said computer will stop running said training when said risk of a medical malpractice claim for the last two iteration sessions differs by two percent or less;
	The prior art of record does provide value based data points and social factor points, but does not specify that there are at least 16 value based care data points based on every single one of “hospital readmissions patient satisfaction scores, outcome data and billing/coding/staging data; and said hospital readmissions patient, said satisfaction scores, said outcome data and said billing/coding/staging.” In addition, the prior art of record does not provide a minimum amount of 3 iteration sessions, a maximum of 200 iterations, and does not teach that the computer will stop running when the risk differs by 2 percent or less. Since the prior art of record does not teach or suggest each and every single claim limitation above, claim 1 is considered allowable over prior art if rewritten or amended to overcome the 101 rejection. Furthermore, by virtue of their dependency on claim 1, claims 2-8 are also considered allowable over prior art if rewritten or amended to overcome the rejections under 101. 

Conclusion
15.	The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure:
-Robertson et al. (US 20040024620 A1) discloses an insurance risk classification methodology that classifies insured individuals into risk groups based on personality traits such as impulsivity, control, self-esteem, etc.
-Kenneth Neumann (US 20210201417 A1) discloses a method for using a machine learning model to calculate an insurance coverage and premium based on input data such as health information.
-Ceulemans et al. (US 20190102670 A1) discloses an insurance broker-mediated data analysis that includes machine learning models running a limited amount of iterations based on a threshold amount of error change per iteration.
16.	Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICO LAUREN PADUA whose telephone number is (703)756-1978. The examiner can normally be reached Mon to Fri: 8:30 to 5:00pm.
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, Jessica Lemieux can be reached at (571) 270-3445. 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.

/NICO L PADUA/               Junior Patent Examiner, Art Unit 3626                                                                                                                                                                                         
/JESSICA LEMIEUX/               Supervisory Patent Examiner, Art Unit 3626