Systems and methods for detecting potentially inaccurate insurance claims

• US 10,089,691 B2
• Filed: 12/04/2013
• Issued: 10/02/2018
• Est. Priority Date: 12/04/2013
• Status: Active Grant

• Current Assignee: State Farm Mutual Automobile Insurance Company (State Farm)
• Original Assignee: State Farm Mutual Automobile Insurance Company (State Farm)
• Inventors: Cook, Michael K.
• Primary Examiner(s): Ziegle, Stephanie
• Application Number:
  US14/096,884
• Publication Number:
  US 20150154712A1
• Time in Patent Office: 1,763 Days
• Field of Search: 705 4
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

Methods and systems for detecting inaccurate insurance claims associated with a loss event are provided. An insurance provider may receive loss event data associated with the loss event from a variety of sources. The insurance provider may also receive an insurance claim submission from the claimant customer that indicates an insured property and a claimed amount of damage to the insured property. The insurance provider examines the loss event data to estimate an actual amount of damage experienced by the property as a result of the loss event, and compares the actual amount of damage to the claimed amount of damage. Based on the comparison, the insurance provider may approve the insurance claim, deny the insurance claim, or flag the insurance claim. The systems and methods thus aim to reduce fraudulent or otherwise inaccurate insurance claim filings.

Claims

1. A computer-implemented method of detecting inaccurate insurance claims associated with a loss event, the method comprising:

continuously receiving, via a network connection using dedicated stream processing hardware, loss event data related to the loss event, the loss event data indicating a property;

continuously receiving, via the network connection using the dedicated stream processing hardware, social networking data from a social networking service, the social networking data including a set of posts submitted by a set of users, the set of posts including (i) a first post including at least one image and a caption for the at least one image, and (ii) a second post including textual content;

continuously receiving, from a set of news sources via the network connection using the dedicated stream processing hardware, news data associated with the loss event;

routing the loss event data, the social networking data, and the news data to a high-speed memory store;

indexing, using at least one inverted index, the loss event data, the social networking data, and the news data;

semantically analyzing, in real-time as the social networking data and the news data are continuously received, routed to the high-speed memory store, and indexed using the at least one inverted index, the social networking data and the news data to identify a set of relevant data, including;

identifying the loss event depicted in the at least one image,

determining that the caption for the at least one image indicates the loss event,

determining that textual content of the second post is not relevant to the loss event,

including the first post in the set of relevant data, and

excluding the second post from the set of relevant data;

receiving an insurance claim from a claimant customer, the insurance claim associated with an insurance policy for the property and the insurance claim indicating a claimed monetary amount of damage to the property; and

automatically, with one or more processors;

examining the loss event data and the set of relevant data to estimate an actual monetary amount of damage to the property,

calculating a difference between the claimed monetary amount of damage and the actual monetary amount of damage, and

comparing the calculated difference to a threshold amount or percentage.

2. The computer-implemented method of claim 1, wherein continuously receiving the loss event data related to the loss event comprises continuously receiving sensor data from at least one sensor device associated with the property, and wherein examining the loss event data comprises examining the sensor data to estimate the actual monetary amount of damage.

3. The computer-implemented method of claim 1, wherein continuously receiving the loss event data related to the loss event comprises continuously receiving third-party data from at least one third-party server, and wherein examining the loss event data comprises examining the third-party data to (1) identify the property and (2) estimate the actual monetary amount of damage to the property.

4. A system for detecting inaccurate insurance claims associated with a loss event, comprising:

stream processing hardware;

a communication module adapted to continuously receive, via a network connection using the stream processing hardware, (i) loss event data related to the loss event, the loss event data indicating a property, (ii) news data associated with the loss event, and (iii) social networking data from a social networking service, the social networking data including a set of posts submitted by a set of users, the set of posts including (i) a first post including at least one image and a caption for the at least one image, and (ii) a second post including textual content;

a claim processing module adapted to receive an insurance claim from a claimant customer, the insurance claim associated with an insurance policy for the property;

a high-speed memory store;

a memory adapted to store non-transitory computer executable instructions; and

a processor interfaced with the communication module, the claim processing module, the high-speed memory store, and the memory, wherein the processor is configured to execute the non-transitory computer executable instructions to cause the processor to;

route the loss event data, the social networking data, and the news data to the high-speed memory store,

index, using at least one inverted index, the loss event data, the social networking data, and the news data,

semantically analyze, in real-time as the news data and the social networking data are continuously received, routed to the high-speed memory store, and indexed using the at least one inverted index, the news data and the social networking data to identify a set of relevant data, including;

identify the loss event depicted in the at least one image,

determine that the caption for the at least one image indicates the loss event,

determine that textual content of the second post is not relevant to the loss event,

include the first post in the set of relevant data, and

exclude the second post from the set of relevant data,

examine the insurance claim to identify a claimed monetary amount of damage to the property,

identify at least a portion of the loss event data that indicates the property, and

automatically;

examine at least the portion of the loss event data and the set of relevant data to estimate an actual monetary amount of damage to the property,

calculate a difference between the claimed monetary amount of damage and the actual monetary amount of damage, and

compare the calculated difference to a threshold amount or percentage.

5. The system of claim 4, wherein the communication module automatically receives the loss event data related to the loss event from at least one sensor device associated with the property, and wherein the processor is further configured to execute the non-transitory computer executable instructions to cause the processor to examine the loss event data from the at least one sensor device to estimate the actual monetary amount of damage.

6. The system of claim 4, wherein the communication module automatically receives the loss event data related to the loss event from at least one third-party server and wherein the processor is further configured to execute the non-transitory computer executable instructions to cause the processor to examine the loss event data from the at least one third-party server to estimate the actual monetary amount of damage to the property.