# US 12,602,508 B1 · Machine learning based authentication and real time data access with disparate communication networks > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,602,508 B1 - Document ID: US12602508B1 - Status: Granted - Kind: B1 - Application number: 18/424,621 - Filed: 2024-01-26 - Granted: 2026-04-14 - Earliest claimed: 2023-08-18 - Assignees: Forward Lending, Inc. - Applicants: Forward Lending, Inc. - Inventors: Jose Bethancourt; Mit Shah; Jesus Marco del Carmen; Mustafa Ali; Christopher Ayre Boggs - Web page: https://patentgrep.com/patent/US12602508B1 ## CPC - G06F 21/6227: {where protection concerns the structure of data, e.g. records, types, queries} - G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning - G06N 3/09: Supervised learning - G06N 20/00: Machine learning - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Systems, methods, and non-transitory computer readable media are configured to perform operations comprising obtaining formatted account data maintained by an entity and associated with an account of a user; accessing in real time account data associated with the account from an institution after the entity has facilitated authentication of the user with the institution; and based on a machine learning model, generating synchronized data that reflects selection of a data field of a data type from the formatted account data maintained by the entity or a data field of the data type from the account data from the institution. ## Claims ### Claim 1 A computer-implemented method comprising: obtaining, by a computing system, formatted account data maintained by an entity and associated with an account of a user; accessing, by the computing system, in real time account data associated with the account from an institution after the entity has facilitated authentication of the user with the institution; and based on a machine learning model, generating, by the computing system, synchronized data that reflects selection of a data field of a data type from the formatted account data maintained by the entity or a data field of the data type from the account data from the institution, wherein the machine learning model is a large language model, the generating synchronized data comprising: generating a first example of training data comprising i) account data from an institution including a data field associated with a first data type and account data from formatted data including a data field associated with the first data type and ii) synchronized data including the data field in the account data from the institution and excluding the data field in the account data from the formatted data, generating a second example of training data comprising i) account data from an institution including a data field associated with a second data type and account data from formatted data including a data field associated with the second data type and ii) synchronized data including the data field in the account data from the formatted data and excluding the data field in the account data from the institution, and fine tuning the large language model based on training data comprising the first example of training data and the second example of training data. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the generating synchronized data comprises: standardizing account data from the institution to be consistent with formatted data maintained by the entity, wherein a label associated with the data field of the account data from the institution is determined to be equivalent to the data type associated with the data field of the formatted account data. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein the selection of the data field from the formatted account data or the data field from the institution is based on recency or availability of the data field from the formatted account data and the data field from the institution. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, wherein the synchronized data excludes the data field from the institution based on a determination that the data field from the institution is inaccurate or unreliable. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field associated with a data type and ii) synchronized data associating the data field from the institution with a standardized data type specified by the entity that is equivalent to the data type. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field associated with a data type and account data from formatted data maintained by the entity including a data field associated with the data type and ii) synchronized data including the data field in the account data from the institution and excluding the data field in the account data from the formatted data, the data type relating to at least one of next due amount and current balance. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field associated with a data type and account data from formatted data maintained by the entity including a data field associated with the data type and ii) synchronized data including the data field in the account data from the formatted data and excluding the data field in the account data from the institution, the data type relating to at least one of account closure, account number, interest rate, and original loan amount. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, wherein the machine learning model is trained with the first example at a first time, and the machine learning model is trained with the second example at a second time. ### Claim 9 (depends on 1) The computer-implemented method of claim 1, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field that is determined to be unreliable or incorrect and ii) synchronized data excluding the data field. ### Claim 10 (depends on 1) The computer-implemented method of claim 1, wherein the training data is generated by the entity. ### Claim 11 A system comprising: at least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the system to perform operations comprising: obtaining formatted account data maintained by an entity and associated with an account of a user; accessing in real time account data associated with the account from an institution after the entity has facilitated authentication of the user with the institution; and based on a machine learning model, generating synchronized data that reflects selection of a data field of a data type from the formatted account data maintained by the entity or a data field of the data type from the account data from the institution, wherein the machine learning model is a large language model, the generating synchronized data comprising: generating a first example of training data comprising i) account data from an institution including a data field associated with a first data type and account data from formatted data including a data field associated with the first data type and ii) synchronized data including the data field in the account data from the institution and excluding the data field in the account data from the formatted data, generating a second example of training data comprising i) account data from an institution including a data field associated with a second data type and account data from formatted data including a data field associated with the second data type and ii) synchronized data including the data field in the account data from the formatted data and excluding the data field in the account data from the institution, and fine tuning the large language model based on training data comprising the first example of training data and the second example of training data. ### Claim 12 (depends on 11) The system of claim 11, wherein the generating synchronized data comprises: standardizing account data from the institution to be consistent with formatted data maintained by the entity, wherein a label associated with the data field of the account data from the institution is determined to be equivalent to the data type associated with the data field of the formatted account data. ### Claim 13 (depends on 11) The system of claim 11, wherein the selection of the data field from the formatted account data or the data field from the institution is based on recency or availability of the data field from the formatted account data and the data field from the institution. ### Claim 14 (depends on 11) The system of claim 11, wherein the synchronized data excludes the data field from the institution based on a determination that the data field from the institution is inaccurate or unreliable. ### Claim 15 (depends on 11) The system of claim 11, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field associated with a data type and ii) synchronized data associating the data field from the institution with a standardized data type specified by the entity that is equivalent to the data type. ### Claim 16 A non-transitory computer-readable storage medium including instructions that, when executed by at least one processor of a computing system, cause the computing system to perform operations comprising: obtaining formatted account data maintained by an entity and associated with an account of a user; accessing in real time account data associated with the account from an institution after the entity has facilitated authentication of the user with the institution; and based on a machine learning model, generating synchronized data that reflects selection of a data field of a data type from the formatted account data maintained by the entity or a data field of the data type from the account data from the institution, wherein the machine learning model is a large language model, the generating synchronized data comprising: generating a first example of training data comprising i) account data from an institution including a data field associated with a first data type and account data from formatted data including a data field associated with the first data type and ii) synchronized data including the data field in the account data from the institution and excluding the data field in the account data from the formatted data, generating a second example of training data comprising i) account data from an institution including a data field associated with a second data type and account data from formatted data including a data field associated with the second data type and ii) synchronized data including the data field in the account data from the formatted data and excluding the data field in the account data from the institution, and fine tuning the large language model based on training data comprising the first example of training data and the second example of training data. ### Claim 17 (depends on 16) The non-transitory computer-readable storage medium of claim 16, wherein the generating synchronized data comprises: standardizing account data from the institution to be consistent with formatted data maintained by the entity, wherein a label associated with the data field of the account data from the institution is determined to be equivalent to the data type associated with the data field of the formatted account data. ### Claim 18 (depends on 16) The non-transitory computer-readable storage medium of claim 16, wherein the selection of the data field from the formatted account data or the data field from the institution is based on recency or availability of the data field from the formatted account data and the data field from the institution. ### Claim 19 (depends on 16) The non-transitory computer-readable storage medium of claim 16, wherein the synchronized data excludes the data field from the institution based on a determination that the data field from the institution is inaccurate or unreliable. ### Claim 20 (depends on 16) The non-transitory computer-readable storage medium of claim 16, wherein an example of training data to train the machine learning model comprises i) account data from an institution including a data field associated with a data type and ii) synchronized data associating the data field from the institution with a standardized data type specified by the entity that is equivalent to the data type.