US 12,393,987 B1
Methods and apparatus for automated insurance claim processing using historical data
Holly Lambert, Roswell, GA (US); Michelle Delynn Adler, Sugar Hill, GA (US); Song Zheng, Duluth, GA (US); Joshua Zaks, Alpharetta, GA (US); Orr Epstein, Atlanta, GA (US); and Ryan Monroe, Dacula, GA (US)
Assigned to State Farm Mutual Automobile Insurance Company, Bloomington, IL (US)
Filed by STATE FARM MUTUAL AUTOMOBILE INSURANCE COMPANY, Bloomington, IL (US)
Filed on Nov. 22, 2019, as Appl. No. 16/693,031.
Claims priority of provisional application 62/815,711, filed on Mar. 8, 2019.
Int. Cl. G06Q 40/08 (2012.01); G06F 16/2458 (2019.01); G06N 20/00 (2019.01); G06Q 10/10 (2023.01)
CPC G06Q 40/08 (2013.01) [G06F 16/2462 (2019.01); G06N 20/00 (2019.01); G06Q 10/10 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method of estimating damage to a vehicle, the method comprising:
receiving, by one or more processors, one or more images illustrating damage to an area of a particular vehicle, wherein the one or more images are:
received, via a network, from an electronic device separate from the one or more processors, and
digital images captured by an imaging sensor of the electronic device;
selecting, by the one or more processors, based on a vehicle type of the particular vehicle, and from a plurality of machine learning algorithms trained using respective sets of digital images illustrating damaged vehicles of a same vehicle type, a first machine learning algorithm configured to identify similarities between digital images illustrating damaged vehicles of the vehicle type;
identifying, by the one or more processors, using the first machine learning algorithm, a plurality of stored images that are matching in appearance with an image of the one or more images, each image of the plurality of stored images illustrating a damaged vehicle of the vehicle type;
identifying, by the one or more processors, based on previously-processed insurance claims associated with damaged vehicles illustrated in the plurality of stored images, a subset of the plurality of stored images, wherein:
each image of the subset illustrates a vehicle having damage to a same vehicle component,
the same vehicle component is identified as damaged by the insurance claims, and
the same vehicle component is obscured from view in the received one or more images illustrating damage to the area of the particular vehicle;
determining, by the one or more processors, a proportion indicating a likelihood that the same vehicle component of the particular vehicle is damaged, wherein determining the proportion comprises:
determining a first quantity that indicates a total number of individual vehicles illustrated in the subset of the plurality of stored images, and
determining a second quantity that indicates a total number of individual vehicles illustrated in the plurality of stored images;
determining, by the one or more processors, that the proportion is greater than a threshold value;
determining, by the one or more processors and based on the proportion being greater than the threshold value, a repair estimate including a cost associated with repair or replacement of the component;
generating, by the one or more processors, a user interface indicating the repair estimate;
providing, by the one or more processors, the user interface such that the repair estimate is output via the user interface;
determining, by the one or more processors, that the repair estimate is within a threshold amount of an actual cost of repair; and
based on determining that the repair estimate is within the threshold amount of the actual cost of repair, augmenting, by the one or more processors, a set of training data to include:
the received one or more images illustrating damage to the area of the particular vehicle, and
the repair estimate, wherein
the augmented set of training data is configured to train a second machine learning algorithm, different from the first machine learning algorithm, to generate a repair estimate.