System and method for evaluating images to support multiple risk applications

Patent metadata

Field	Value
Publication number	US-11250364-B2
Application number	US-201715833008-A
Country	US
Kind code	B2
Filing date	Dec 6, 2017
Priority date	Dec 6, 2017
Publication date	Feb 15, 2022
Grant date	Feb 15, 2022

Abstract

In some embodiments, image input data is received from multiple sources. The received image input data may then be aggregated and mapped to create a set of image input data. An event in the set of image input data may be automatically detected, such as by being triggered by a rule and an associated tag. An image mining result database may be updated by adding an entry to the database identifying each detected event and the triggering rule. An indication associated with the image mining result database may then be transmitted to a plurality of risk applications.

First claim

What is claimed:

1. A system for evaluating image data to support multiple insurance applications, comprising:
a communication device to receive image input data from multiple sources, the sources including a medical report, a police report, social network data, Internet of Things (IoT) data, Global Positioning System data, satellite photo data, activity tracking data, loss descriptions, injury descriptions, a Final Notice of Loss statement and a video teleconference;
a processor coupled to the communication device;
a storage device in communication with said processor and storing instructions adapted to be executed by said processor to:
aggregate and map the received image input data to create a set of image input data;
automatically detect an insurance claim in the set of image input data triggered by a rule and an associated tag, said detecting including processing the image input data via an analytics engine and applying, via machine learning, at least one pre-determined condition to the image input data reducing a number of messages transmitted via a communication network, wherein the analytics engine is trained via machine learning processes with training data and non-training data resulting from a filter applied to a prediction set of image input data,
flag the detected insurance claim as meeting the pre-determined condition,
update an image mining result database, responsive to the flag, by adding an entry to the database identifying each detected insurance claim and the triggering rule,
transmit an indication associated with the image mining result database to a plurality of insurance applications, at least one of the plurality of insurance applications associated with a predictive model, the predictive model including a neural network and a support vector machine, said plurality of insurance applications including a subrogation software application, a fraud detection software application and a recovery factor analysis software application;
display an image from the database on a handheld computing device, the image showing an object with associated tags, the associated tags presented in the image as a data hierarchy tree, the hierarchy tree formed of a plurality of nodes and branches, and the associated tags ranked in the hierarchy tree according to a relative status of the associated tags,
adjust the data hierarchy tree in response to input from a user of the handheld computing device,
automatically update the triggering rule based on corrections of image data evaluations manually inputted into the system, and
update at least one variable or weighing factor of the predictive model;

an email alert server in communication with the processor to receive results of image data processing performed by the processor;
a workflow application in communication with the processor to receive results of image data processing performed by the processor; and
a server executing reporting and calendar functions, said functions in communication with the processor to receive results of image data processing performed by the processor; and
automatically train the machine learning processes using the received results of image data processing.

2. The system of claim 1 , wherein execution of the stored instructions further causes the processor to:
receive, from at least one of the insurance applications, feedback information associated with the detected insurance claim, and based on the received feedback information, automatically update at least one of the rule and the associated tag.

3. The system of claim 1 , wherein the insurance applications are associated with at least two of. (i) a workers' compensation claim, (ii) a personal insurance policy, (iii) a business insurance policy, (iv) an automobile insurance policy, (v) a home insurance policy, (vi) a sentiment analysis, (vii) risk event detection, (viii) a cluster analysis, (ix) a predictive model, (x) large loss and volatile claim detection, (xi) a premium evasion analysis, (xii) an insurance policy comparison, (xiii) an underwriting decision, and (xiv) indicator incidence rate trending.

4. The system of claim 1 , wherein the transmitted indication is to: (i) trigger an insurance application, or (ii) update an insurance application.

5. The system of claim 1 , wherein the transmitted indication is associated with a variable or weighing factor of a predictive model.

6. The system of claim 1 , wherein multiple versions of the rule are maintained along with multiple versions of the image mining result database.

7. The system of claim 1 , wherein the insurance claim is associated with at least one of: (i) an item present within the image, and (ii) an activity associated with the image.

8. The system of claim 1 , wherein the triggering rule was defined by an administrator using a graphical user interface.

9. The system of claim 1 , wherein the entry added to the image mining result database includes all of: (i) an image identifier, (ii) an insert date, (iii) an image source, (iv) a rule identifier, and (v) a line of business.

10. The system of claim 1 , wherein the storage device further stores instructions adapted to be executed by said processor to:
identify at least a portion of the image input data as being connected with an image domain, and
associate that portion of the image input data with a custom tag based on the image domain.

11. The system of claim 1 , wherein said automatic detection is performed via cloud analytics associated with at least one of: (i) object storage, (ii) a data catalog, (iii) a data lake store, (iv) a data factory, (v) machine learning, and (vi) artificial intelligence services.

12. The system of claim 1 , wherein: the object shown in the displayed image is an automobile; and the tags shown with the object in the displayed image include an object type tag, an object owner tag, an automobile type tag, and a year of manufacture tag.

13. The system of claim 1 , wherein the at least one pre-determined condition is at least one of:
a pre-determined number of times an item is detected; and
a pre-determined proximity between a first item and a second item.

14. A computer-implemented method for evaluating image data to support multiple insurance applications, comprising:
receiving, at an image mining platform in a system, image input data from multiple sources including a source internal to an insurance enterprise and a source external to the insurance enterprise, the sources including a medical report, a police report, social network data,, Internet of Things (IoT) data, Global Positioning System data., satellite photo data, activity tracking data, loss descriptions, injury descriptions, a Final Notice of Loss statement and a video teleconference;
aggregating and mapping the received image input data to create composite input image data;
automatically detecting, by a computer processor of the image mining platform, an insurance claim in the set of image input data triggered by a rule and an associated tag, said detecting including processing the image input data via an analytics engine and applying, via machine learning, at least one pre-determined condition to the image input data reducing a. number of messages transmitted via a communication network, wherein the analytics engine is trained via machine learning processes with training data and non-training data resulting from a. filter applied to a prediction set of image input data;
updating, by the computer processor, an image mining result database by adding an entry to the database, the entry including: (i) an image identifier, (ii) an insert date, (iii) an image source, (iv) a rule identifier, and

Assignees

• Beznos Vladislav Michael
• Boles Brenda K
• Kazmi Saira A
• Steger Tracey Ellen
• Cultrera Steven J
• Hartford Fire Insurance Comp

Inventors

• Beznos Vladislav Michael
• Boles Brenda K
• Kazmi Saira A
• Steger Tracey Ellen
• Cultrera Steven J

Classifications

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/23

  Clustering techniques · CPC title

• G06F18/254

  of classification results, e.g. of results related to same input data · CPC title

• G06V20/00

  Scenes; Scene-specific elements (control of digital cameras H04N23/60) · CPC title

• G06F2216/03

  Data mining · CPC title