Insurers generally started using predictive modeling for the underwriting process. Claims predictive modeling is at the early momentum stages for insurers and a great place to start for self-insured employers.
Effective models are always about the amount of data and its quality. Self-insured and self-administered employers should be the easiest candidates to model. (To see an example, please click here and scroll to the sidebar.) Since insurance companies augment data to build their models, self-insured employers should consider doing the same.
Self-insured employers that do not self-administer will need to access their data from their third party administrators (TPAs). Some TPAs, like progressive self-insured employers, are applying predictive modeling to the claims process. Employers can also install claim processing systems that are incorporating predictive models. I would imagine that self-insuring groups have the same advantage of similar data just as insurers that specialize in specific industries.
Specific data to use depends on the model’s extent and purpose.
Claims predictive modeling covers several facets. Models can be crafted to indicate possible factors for fraud, determine how well claims are being handled, the effectiveness of medical case management or even to identify the most effective doctors.
Specific data to use depends on the model’s extent and purpose. Determining which data to model starts with using known risk factors that can affect overall claim costs. Many of these factors employers should be tracking anyway. We know, for example, that filing claims immediately improves the chances of positive outcomes. So including the injury, filing and treatment dates is important.
Including claim examiner response times also provide insight into how much the process itself impedes progress. Other factors to include incident location; doctor(s); diagnosis; employee’s length of employment; overall workforce employment length; employee age; educational level; and, co-morbidities. Since the employee-supervisor relationship can reveal a lot about a claim’s destiny, consider including it too.
Some factors can also be based on research. WCRI’s study on attorney involvement revealed that injured workers hired attorneys based on the perception they were not getting paid when really, the claim had not been accepted yet. This should inspire keeping track of the not just if an attorney is involved but the why and when it happened. Safety and workplace incident prevention also benefit from predictive modeling as well.
- Predictive modeling will more accurately reflect an employer’s specific experience. Click Here
- Predictive Modeling Is Here to Stay Click Here
- Predictive modeling and premium auditing Click Here
- “Predictive Modeling Is Here to Stay” Discussion on the LinkedIn Group: Workers Compensation Roundtable Click Here
- Predictive modeling and Your Insurance Agent and Broker. Click Here
- Predictive Modeling ‘s Impact on Agents and Brokers—and Their Clients Click Here
- Workers’ Compensation Predictive Modeling Click Here
- Workers’ Comp Predictive Modeling Comes of Age Click Here