That challenge? To quickly and accurately identify the small percentage of workers compensation claims at risk to drive the majority of total claim costs.
Some of these high-cost claims are obvious from the start, such as a roofer tragically falling and becoming paralyzed. But most emerge over time.
In 2004, Liberty Mutual developed one of the industry’s first workers compensation predictive models to pinpoint slow emerging claims in response to soaring medical costs.
"Today, our next generation predictive model works with significantly more data, enables more sophisticated multivariate analysis, incorporates Red and Yellow flags and supports better decision making," notes George Neale, general manager for claims, Liberty Mutual's Commercial Markets strategic business unit. “As early as claim intake, we now have the ability to identify claims with the potential to be high cost. In addition, we continuously assess the risk of escalation of our entire inventory through several modeling stages. Arming our claim handlers with this information gives us a unique opportunity to apply strategies and expert resources at a point in time where they maximize our ability to achieve the best possible outcome for injured workers and our policyholders."
To enhance its predictive model, Liberty Mutual analyzed its workers compensation claims and medical billing database – the largest in the industry - to find highly predictive variables and to understand the interplay of these in order to identify claims whose costs were likely to escalate. It evaluated more than 825,000 lost time claims and 140 million individual medical billing transactions. To validate the accuracy of the updated model, the insurer ran more than 200,000 lost time claims through it.
Liberty Mutual's next generation predictive model is effective because it includes:
• Broader data. The model captures additional data - beyond just medical and billing - on individual claimants to better predict how a claim will develop. This new data is based in part on research from the Liberty Mutual Research Institute’s Center for Disability Research that identified the psychological and social issues and co-morbid medical conditions that impact a workers compensation claim’s duration and cost
• Sophisticated multivariate analysis. Not only does the model capture more data, but it looks at that data holistically to more accurately predict high-cost claims. Rather than measure the impact of each data point, the model looks at the impact of all data points together and over time in order to identify likely changes in the claim and their potential impact
• More frequent model runs. The predictive model will look at claims monthly and pick up changes in each claim’s profile that can negatively impact that claim’s development, such as emerging medical and non-medical factors
• Tools that enable claims professionals to take action sooner. Capturing more data and analyzing it carefully is important. But to be truly effective in lowering claim costs and improving outcomes, the model must give claim professionals more insight more quickly into how a claim is developing so that they can take action to improve its outcome. The next generation predictive model does this through several new tools:
- An Early Alert tool flags potential high-cost claims and those at risk for escalation
- A Medical Referral tool triggers referrals to specialized medical resources
- A Dashboard presents relevant claims information, including flags to indicate importance
"The next generation predictive model is the latest addition to Liberty Mutual’s VantageComp, our integrated approach to claims management," notes Neale. "This approach seamlessly blends technology, data analytics, predictive modeling and superior talent to better manage policyholders’ total workers compensation claim costs. It enables us to close workers compensation claims faster and to produce lower Average Paid Costs per Claim than our competitors, according to data from A.M. Best."