US 11,783,422 B1
Implementing machine learning for life and health insurance claims handling
Gregory L Hayward, Bloomington, IL (US); Meghan Sims Goldfarb, Bloomington, IL (US); Nicholas U. Christopulos, Bloomington, IL (US); and Erik Donahue, Normal, IL (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 Sep. 20, 2018, as Appl. No. 16/136,387.
Claims priority of provisional application 62/652,121, filed on Apr. 3, 2018.
Claims priority of provisional application 62/646,740, filed on Mar. 22, 2018.
Claims priority of provisional application 62/646,735, filed on Mar. 22, 2018.
Claims priority of provisional application 62/646,729, filed on Mar. 22, 2018.
Claims priority of provisional application 62/632,884, filed on Feb. 20, 2018.
Claims priority of provisional application 62/625,140, filed on Feb. 1, 2018.
Claims priority of provisional application 62/622,542, filed on Jan. 26, 2018.
Claims priority of provisional application 62/621,797, filed on Jan. 25, 2018.
Claims priority of provisional application 62/621,218, filed on Jan. 24, 2018.
Claims priority of provisional application 62/618,192, filed on Jan. 17, 2018.
Claims priority of provisional application 62/617,851, filed on Jan. 16, 2018.
Claims priority of provisional application 62/610,599, filed on Dec. 27, 2017.
Claims priority of provisional application 62/580,713, filed on Nov. 2, 2017.
Claims priority of provisional application 62/580,655, filed on Nov. 2, 2017.
Claims priority of provisional application 62/564,055, filed on Sep. 27, 2017.
Int. Cl. G06Q 40/08 (2012.01); G06Q 10/10 (2023.01); G06N 3/08 (2023.01); G06N 20/00 (2019.01)
CPC G06Q 40/08 (2013.01) [G06N 3/08 (2013.01); G06N 20/00 (2019.01); G06Q 10/10 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method of claims handling, comprising:
receiving a set of labeled historical claims, each labeled historical claim of the set of labeled historical claims corresponding to a respective adjusted settlement amount and a label, the label being one of a plurality of predetermined labels;
training a plurality of first artificial neural networks using a first subset of the set of labeled historical claims;
training a second artificial neural network using a second subset of the set of labeled historical claims and the respective adjusted settlement amount for each labeled historical claim of the second subset of the labeled historical claims, the first subset of the set of labeled historical claims being different from the second subset of the set of labeled historical claims;
receiving, from a user, a life claim, the life claim comprising at least one selected from a group consisting of image data and audio data; and
analyzing the life claim using the plurality of trained first artificial neural networks and the trained second artificial neural network to determine a claim settlement prediction by at least:
extracting text-based content from the at least one selected from a group consisting of image data and audio data in the life claim using at least a natural language processing model;
selecting a trained first artificial neural network from the plurality of trained first artificial neural networks based on the extracted text-based content;
inputting the extracted text-based content to the selected trained first artificial neural network;
determining a claim label representing a category of the life claim using the selected trained first artificial neural network based at least in part on the extracted text-based content, the claim label being one of the plurality of predetermined labels;
inputting the extracted text-based content and the determined claim label to the trained second artificial neural network; and
determining the claim settlement prediction using the trained second artificial neural network based at least in part on the extracted text-based content and the determined claim label.