Capital generation and return metrics in Europe
Despite its ease of calculation, capital generation and return—measured as a change in eligible own funds—may not be the most useful metric.
The market-leading predictive analytics platform for insurance optimizes claims performance, identifies at-risk claims, and reduces claims cost.
Learn how to maximize your ROI by integrating your claims data with AI.
Claims triage is a well-established component of insurers’ and self-insurers’ claims processes. Often a key determinant in the outcome of a claim, the claims triage process aims to segment high-cost from low-cost claims and then allocate the appropriate resources to optimally manage them. But the process has increasingly been met with challenges as more and more complex claims arise, costs increase, and claims resources come under pressure. These challenges, however, are occurring at a time when advances in artificial intelligence (AI) technology have made it possible for organizations to gain access to the rich trove of data buried in their claims files. This access has given insurers and self-insurers actionable insights into the claims triage process and allowed them to more efficiently allocate their resources and lower claims costs.
While catastrophic injuries, such as multiple traumas and brain injuries, are easy targets for any claims triage process, it is much harder to identify claims that initially seem innocuous but will increase in cost over time. These types of claims can evade detection, causing them to languish until a time when cost containment strategies may no longer be effective. Identifying these hard-to-detect claims early in the process can help insurers improve claims performance and shift the balance on costs.
While some claims organizations rely on automated rules to triage claims, many still triage them using a claims professional’s judgment based on limited information taken at first notice of loss (FNOL). As time passes, more information on the claim will emerge, which can change the severity and outlook of the claim. But the sheer volume of claims information that comes in each day on a portfolio of claims makes it impractical for claims professionals to notice and act on changes that might signal a deterioration. The inability to systematically rereview all open claims as time passes leads to claims slipping through the cracks.
The traditional ways of allocating claims and resources are now giving way to a modern approach that uses AI and predictive analytics to mimic human thought process and triage claims. This technology can identify potentially hard-to-detect high-cost claims as new claims information becomes available and alert claims professionals of the potential need to redirect them to the appropriate resources. On the other end of the spectrum, low-cost claims can be auto-adjudicated and fast-tracked to resolution.
Advances in a type of AI known as natural language processing (NLP) can intelligently read through adjusters’ notes and other text documents, extracting data to create high-quality data sets that support and inform the claims triage process. Information that may have been inadvertently passed over during a first report such as a comorbidity—or that would not be available like a scheduled MRI—is identified early in the process.
NLP unlocks a trove of information that was hard to efficiently access without considerable time and effort, and it overcomes many of the limitations of missing or inconsistent structured data. Advanced claims triage models can effectively track the evolution of a claim, uncovering valuable information much sooner than was previously possible. This approach closes the gap between the time when a potentially significant development occurs and when an insurer can bring in more appropriate resources to manage the claim.
Automated models run daily—scanning for characteristics or terms that signal high cost—making claims triage a continual process. References to a possible surgery, MRI, attorney involvement, or increasing pain are no longer buried in adjusters’ notes but can now systemically inform the claims triage process, transforming it from a subjective, time-consuming, intermittent, and often inaccurate procedure to an objective, efficient, continuous, and informed best practice.
A predictive model typically provides a score or dollar amount for each claim based on its complexity and potential cost. High-cost claims are flagged shortly after the information is entered into the claims system, allowing for the earliest possible intervention to control costs. Scoring is based on an objective set of criteria that are consistently applied to the process.
The output from the predictive model needs to be supported by a rationale that is accessible to the claims adjusters. Was the high score the result of attorney involvement? References to a surgery or costly procedure? Or some combination? Claims professionals need to feel the predictive analytics used in an automated claims triage approach are interpretable for the output to be effective.
Benchmarking costs is an essential part to monitoring the performance of an automated claims triage system. Shifts in the population of claims over time can distort an equitable evaluation of changes in claims costs. An example can be seen with the onset of COVID-19, which turned the mix of workers’ compensation claims on its head, as scores of office workers began working from home and medical-only claims turned into lost time. Other factors such as social inflation and law changes can also cause shifts in claims trends that can skew benchmarking analyses. These shifts need to be understood and reflected in any benchmarking analysis to determine the effectiveness of a newly implemented claims triage process and overall claims performance.
The data generated by AI can also provide strategic value by tapping into claims trends that are typically unavailable in most structured data. A prime example is attorney representation, which is typically not tracked in many traditional claims systems but can be among the output of an automated claims triage process. Knowledge about changes in attorney representation or other claims cost drivers can add specificity to an insurers’ understanding of its own claim trends.
The ability to triage claims early in the process has both financial and operational benefits. From a financial standpoint, an automated claims triage process can:
Other operational benefits typically include:
Developing a predictive model is not a trivial effort. Development time is only one consideration. The project also requires enormous human capital that can not only keep pace with rapidly emerging advances in technology but also has the breadth of experience in claims, data science, actuarial science, and clinical expertise. Having the expertise of an interdisciplinary team helps to ensure that the information is accurate, actionable, and interpretable.
The importance of good communication between claims adjusters and the predictive modeling team—not only during development but also during and after implementation—cannot be overemphasized. Feedback from adjusters is a key element to improving the usefulness and effectiveness of the model, and ultimately the claims triage process.
Advanced technologies have made it possible to significantly improve efficiency at a time when insurers and self-insurers are facing pressures on several fronts. The loss of expertise in claims departments, whether through retirements or turnover, has elevated the need for continuity and efficiency in the claims triage process. And claims costs, impacted by social inflation and litigation financing, are under pressure. A data-driven claims triage process can give organizations control to manage volatility and provide the actionable insights to better combat the uncertainties that lie ahead.
The complete guide to claims triage: Lowering workers’ compensation costs with predictive analytics
As the claims triage process experiences more complex pressures, advances in artificial intelligence are giving insurers rich data and actionable insights to efficiently allocate resources and lower claims costs.