Our client is a leading US-based provider of insurance and financial services products including life insurance, retirement plans, and mutual funds.
The underwriting team was overloaded with health screening of incoming applications through a traditional invasive process requiring extensive lab tests to identify high-risk applicants. The client wanted to create an automated process that enables straight-through processing of applications of high-risk individuals.
- The process needed to be non-invasive (without blood testing) for cases where good risk prediction is feasible with non-invasive evidence data such as MVR, MIB, RX
- Automated risk assessment required developing white box interpretable predictive models and triangulation from multiple data sources including external data
- Reconciling and deriving KPIs definition from basic evidence data sources ( MVR, MIB, RX, etc. ) required working widely with the business teams
Developed an augmented underwriting solution framework comprising:
- Robust KPIs from various sources of basic evidence data (MVR, MIB, RX) to augment traditional application data
- Submodels to feed into overall risk triaging process:
— Smokers model to identify smokers who do not declare their habit in the application.
— Cholesterol/HDL model to predict cholesterol and HDL levels of applicants.
- Acquired additional data from Commercial Auto Bureau to understand various characteristics such as speed violations, and other forms of traffic violations, indicating propensity of risky on-road behavior.
- Predictive classification models to triage applications through the use of the above-mentioned submodels as well as derived KPIs from MVR, MIB, RX data
- Leveraging the models and derived KPIs to predict high and low-risk applicantsto enable STP (straight-through processing) approval and rejection
- ~30% of applications processed straight through post implementation of the automated framework
- Lab test cost reduced by about 25%
- Over $2M annual savings realized from avoided lab tests.