Self-service claim automation using artificial intelligence
Embodiments are disclosed for automatically processing a claim provided by a user. Responsive to receiving a notice of loss associated with a claim of a user, a set of customer identity validation data are collected. The set of customer identity validation data may be determined to meet a pre-defined identity validation criteria. Responsive to determining that the set of customer identity validation data meets the pre-defined identity validation criteria, current claim evaluation data for the claim may be accessed. A set of predictive impact assessment scores associated with the current claim evaluation data may be determined using a predictive model. The set of current claim evaluation data may be determined to meet pre-defined claim data criteria by comparing the predictive impact assessment scores with a set of impact assessment thresholds. Responsive to determining that the set of current claim evaluation data meets the pre-defined claim data criteria, a reactive action to the claim may be determined.
1 . An apparatus comprising one or more processors and at least one non-transitory computer storage medium storing instructions that, when executed by the one or more processors, cause the apparatus to:
receive a first set of claim evaluation data originating from a first computing device and a second set of claim evaluation data originating from an Internet of Things (IoT) computing device;
generate, based at least in part on one or more trained machine learning models, a plurality of predictive impact assessment scores, each predictive impact assessment score associated with a unique intake field and associated response value of current claim evaluation data for a claim of a user, wherein each intake field of the current claim evaluation data comprises a response value associated with one or more of the first set of claim evaluation data or the second set of claim evaluation data;
responsive to determining, based at least in part on the plurality of predictive impact assessment scores associated with the current claim evaluation data, that an aggregated predictive impact assessment score satisfies an impact assessment threshold, activate, via one or more application programming interfaces (APIs), an augmented reality view interface configured for overlaying on a screen display of the first computing device;
apply the one or more trained machine learning models to one or more captured images received via the one or more application programming interfaces (APIs);
assign a machine learning classifier providing an image subject description to the one or more captured images;
validate, according to a fraud threshold being met, one or more of the one or more captured images for fraud; and
responsive to validating each of the one or more captured images for fraud, determine a coverage for the claim using a coverage mapping for a set of policy data corresponding to an insurance policy of the user.
2 . The apparatus of claim 1 , further caused to:
responsive to receiving, originating from a first computing device, a notice of loss associated with the claim of the user, collect, originating from one or more external computing devices, customer identity validation data.
3 . The apparatus of claim 2 , further caused to:
determine whether the customer identity validation data satisfies a first threshold prior to receiving the first set of claim evaluation data and the second set of claim evaluation data.
4 . The apparatus of claim 3 , wherein the customer identity validation data comprises a plurality of fields each comprising an associated data value.
5 . The apparatus of claim 4 , further caused to:
assign one or more weights to one or more fields of the customer identity validation data.
6 . The apparatus of claim 5 , wherein determining whether the customer identity validation data satisfies the first threshold comprises:
evaluating each field for a level of adequacy of its associated data value;
generating an aggregated weighting based at least in part on the one or more weights assigned to the one or more fields; and
comparing the aggregated weighting to the first threshold.
7 . The apparatus of claim 1 , further caused to:
generate, using the plurality of predictive impact assessment scores associated with current claim evaluation data for a claim of the user, the aggregated predictive impact assessment score.
8 . The apparatus of claim 1 , wherein validating, according to the fraud threshold being met, each of the one or more captured images for fraud, comprises:
extracting metadata associated with the one or more captured images;
comparing the metadata against third party image sources; and
determining that time stamp data of the metadata is captured before a date of loss associated with a claim of the user.
9 . A computer-implemented method, comprising:
receiving, by one or more processors, a first set of claim evaluation data originating from a first computing device and a second set of claim evaluation data originating from an Internet of Things (IoT) computing device;
generating, by the one or more processors and based at least in part on one or more trained machine learning models, a plurality of predictive impact assessment scores, each predictive impact assessment score associated with a unique intake field and associated response value of current claim evaluation data for a claim of a user, wherein each intake field of the current claim evaluation data comprises a response value associated with one or more of the first set of claim evaluation data or the second set of claim evaluation data;
responsive to determining, based at least in part on the plurality of predictive impact assessment scores associated with the current claim evaluation data, that an aggregated predictive impact assessment score satisfies an impact assessment threshold, activating, by the one or more processors and via one or more application programming interfaces (APIs), an augmented reality view interface configured for overlaying on a screen display of the first computing device;
applying, by the one or more processors, the one or more trained machine learning models to one or more captured images received via the one or more application programming interfaces (APIs);
assigning, by the one or more processors, a machine learning classifier providing an image subject description to the one or more captured images;
validating, by the one or more processors and according to a fraud threshold being met, one or more of the one or more captured images for fraud; and
responsive to validating each of the one or more captured images for fraud, determining, by the one or more processors, a coverage for the claim using a coverage mapping for a set of policy data corresponding to an insurance policy of the user.
10 . The computer-implemented method of claim 9 , further comprising:
responsive to receiving, originating from a first computing device, a notice of loss associated with the claim of the user, collecting, by the one or more processors and originating from one or more external computing devices, customer identity validation data.
11 . The computer-implemented method of claim 10 , further comprising:
determining, by the one or more processors, whether the customer identity validation data satisfies a first threshold prior to receiving the first set of claim evaluation data and the second set of claim evaluation data.
12 . The computer-implemented method of claim 11 , wherein the customer identity validation data comprises a plurality of fields each comprising an associated data value.
13 . The computer-implemented method of claim 12 , further comprising:
assigning, by the one or more processors, one or more weights to one or more fields of the customer identity validation data.
14 . The computer-implemented method of claim 13 , wherein determining whether the customer identity validation data satisfies the first threshold comprises:
evaluating each field for a level of adequacy of its associated data value;
generating an aggregated weighting based at least in part on the one or more weights assigned to the one or more fields; and
comparing the aggregated weighting to the first threshold.
15 . The computer-implemented method of claim 9 , further comprising:
generating, by the one or more processors and using the plurality of predictive impact assessment scores associated with current claim evaluation data for a claim of the user, the aggregated predictive impact assessment score.
16 . The computer-implemented method of claim 9 , wherein validating, according to the fraud threshold being met, each of the one or more captured images for fraud, comprises:
extracting metadata associated with the one or more captured images;
comparing the metadata against third party image sources; and
determining that time stamp data of the metadata is captured before a date of loss associated with a claim of the user.
17 . A computer program product comprising at least one non-transitory computer readable storage medium storing instructions that, with one or more processors, cause the one or more processors to:
receive a first set of claim evaluation data originating from a first computing device and a second set of claim evaluation data originating from an Internet of Things (IoT) computing device;
generate, based at least in part on one or more trained machine learning models, a plurality of predictive impact assessment scores, each predictive impact assessment score associated with a unique intake field and associated response value of current claim evaluation data for a claim of a user, wherein each intake field of the current claim evaluation data comprises a response value associated with one or more of the first set of claim evaluation data or the second set of claim evaluation data;
responsive to determining, based at least in part on the plurality of predictive impact assessment scores associated with the current claim evaluation data, that an aggregated predictive impact assessment score satisfies an impact assessment threshold, activate, via one or more application programming interfaces (APIs), an augmented reality view interface configured for overlaying on a screen display of the first computing device;
apply the one or more trained machine learning models to one or more captured images received via the one or more application programming interfaces (APIs);
assign a machine learning classifier providing an image subject description to the one or more captured images;
validate, according to a fraud threshold being met, one or more of the one or more captured images for fraud; and
responsive to validating each of the one or more captured images for fraud, determine a coverage for the claim using a coverage mapping for a set of policy data corresponding to an insurance policy of the user.
18 . The computer program product of claim 17 , wherein the at least one non-transitory computer readable storage medium stores instructions that, with the one or more processors, further cause the one or more processors to:
responsive to receiving, originating from a first computing device, a notice of loss associated with the claim of the user, collect, by the one or more processors and originating from one or more external computing devices, customer identity validation data; and
determine whether the customer identity validation data satisfies a first threshold prior to receiving the first set of claim evaluation data and the second set of claim evaluation data.
19 . The computer program product of claim 18 , wherein the customer identity validation data comprises a plurality of fields each comprising an associated data value.
20 . The computer program product of claim 18 , wherein the at least one non-transitory computer readable storage medium stores instructions that, with the one or more processors, further cause the one or more processors to:
assign one or more weights to one or more fields of the customer identity validation data, wherein determining whether the customer identity validation data satisfies the first threshold comprises:
evaluating each field for a level of adequacy of its associated data value;
generating an aggregated weighting based at least in part on the one or more weights assigned to the one or more fields; and
comparing the aggregated weighting to the first threshold.