Insurance | Wallaroo
  • White Twitter Icon
  • White LinkedIn Icon

© 2019 by Wallaroo Labs, Inc.
Made with  in NYC.

Insurance Claims

The insurance business relies heavily on risk mitigation. 

 

Effectively managing risks requires complex algorithms applied against a backdrop of vast amounts of data to find and alert the business to abnormal patterns.

We can help.

Opportunity

Challenge

There are three main technical challenges to overcome:

  1. Complexity. The analysis has to be able to join data in many formats from various sources to get a complete picture of the lifecycle of a claim and find correlations such as the same individual being involved in multiple claims.
     

  2. Agility. The system has to be agile enough to adapt to new forms of fraud, so new fraud models can be quickly developed and put into production.
     

  3. Versatility. The system has to support the SIU (Special Investigations Unit) with tools that humans can use to identify higher probability incidents, clusters, etc.

How Wallaroo Helps

Wallaroo provides easy solutions to each of these challenges.

 

Wallaroo allows for deployment at the necessary scale cluster in production, while forensics scientists can develop new models as needed. Wallaroo allows for multiple types of source of data, and the ability to transform and combine that data, and then perform map-reduce or other forms of data analysis on it.

 

Finally, the output, whether it's flagged events, individuals, or other statistical information can be integrated with standard methods for data visualization, alerting, and reporting.

 

An added benefit of Wallaroo is that, because it supports stream-processing, as insurers move towards real-time analysis it's straightforward to incorporate that.

Sample Insurance Use Case

Fraud Detection

Insurance fraud can cost individual insurers billions every year, and occurs across automotive, disability, homeowner's, and healthcare sectors.

 

There is normally two categories of fraud: "hard", and "soft". In the former, for example, a group of individuals my conspire to create a claims event, such as staging an accident. In the later, for example, a repair shop may adjust the cost of a repair beyond a reasonable amount.

 

To detect fraud proactively means analyzing huge amounts of structured and unstructured data, & finding patterns and correlations between them in way that an army of humans could not do.