Helps your team identify risky claims and improve claim handling outcomes
For a claims manager, severity scoring could be their most useful and used tool. Having the ability to quickly look at an overview of a claim and determine how urgent it is can increase the productivity and efficiency of the claims lifecycle. The length of treatment, disability duration, and the overall cost can vary by a claim by claim basis. We aim to pinpoint claims that have favorable and unfavorable outcomes and highlight problematic claims so claim managers can intervene in earlier stages.
Our severity scoring models predict the severity of claims with respect to incurred costs at the various stages of the claims lifecycle. This process can help claims managers keep track and monitor the progress of each of their claims and see if a claim is increasing or decreasing in severity. Our models categorize claims into 3 groups- low, medium, and high-risk bands. Generally, we have claims that have a total incurred cost of less than $15k as low risk, $15k to $95k as a medium, and $100k and above as high-risk claims. These thresholds can be changed depending on the user. The severity level is time-updated by new dynamic information as the claim continues through its lifecycle.
Our models collect a plethora of different data inputs ranging from the injury incident, claimant demographics, employment details, medical information, and much more. These data inputs are then run through our severity engine which uses machine learning algorithms and artificial intelligence to score each and every claim on a predetermined scale. Our models then output its predictions into a clean interface that shows its predicted severity level, current severity level, the confidence it has in its outputs, and the top drivers of a severity score.
Primarily, our models get their data through text mining and adjuster input from the first report of injuries (FROIs) and with claim notes. Our text mining engine mines the textual documents, notes from adjusters, and any other documentation related to a claim to discern meaning and assign scores. Another source of information that we mine from is social media analytics. We analyze social media to find possible collusion between industry players by measuring their level of interaction with each other. These inputs are then put through our predictive modeling which is based on logistic regression, decision trees, and neural networks. We also have business rule-driven models to which outline industry-standard rules and anomaly detection rules to evaluate from a clinical or business logic standpoint to create better scorecards.
Why Klear.ai ?
In a quick demonstration of the Klear.ai platform, we can show you how our models efficiently output data into our dashboards and learn how this data can lead to actions that can help claims managers with triaging difficult claims and tracking the progress of these claims throughout their entire lifecycle.
Our severity scoring engine could be one of the most useful tools not only for a claim manager but also for independent adjusters and adjudicators. With these severity scores, claims managers can more effectively assign higher risk claims to senior-level adjusters to assure that the claim is properly managed, medium risk claims to tier 2 adjusters, and low severity claims to fast track processing workflow. Adjusters can also benefit from our severity scoring platforms because it helps in the daily workflow and helps streamline the claims process.
How is the severity score calculated?
Our severity scoring models predict the severity of claims with respect to incurred costs at the various stages of the claims lifecycle. Our models take into account the injury incident, claimant demographics, employment details, medical information, and much more.
What are the top drivers of severity and mode confidence percentages?
How often is the severity score updated?
As a claim continues through its lifecycle, its severity level is time-updated by new dynamic information. Our models also have the ability to display a claim’s current severity level, and its predicted level. This can help claim managers determine if a claim is increasing or decreasing in severity
Where does this information come from? How is the data collected?
Data is collected from your databases and is sent over to our servers to be run through our engines. Most of the claimant information is taken from the FROI and is also automatically taken from adjuster notes through our text mining algorithms.
How can I schedule a demo?
You can book a free comprehensive demo here