Early Claim Prediction

The Challenge

Risk assessment is a crucial element in the life insurance business to classify the applicants. The pay out from life insurance claims are extremely high relative to the insurance premiums that companies collect from an individual customer. This means insurance companies need to be selective about whom they insure and what price they issue policies to make their business financially viable. AI and Machine Learning can help insurers to efficiently screen prospects, evaluate their risk with greater precision, and make accurate pricing recommendations.

About the Company

The Solution

The conventional approach to risk assessment is built on rule-based policies driven by business experience. These policies are inflexible and once the rules are written they tend to be applied equally to every case. CyborgIntell improves risk assessment and helps automate decision making. iTuring uses historic data to understand which customers are likely to make an early insurance claim and provides accurate predictions at an individual customer level in real time. Not only does iTuring accurately predict which customers are most likely to claim early, it provides detailed explanations on risk factors to enable insurers decide product programs and make risk-based price decisions.

For the Life Insurance Company, CyborgIntell used iTuring to build a “Early Claim Prediction Model” to predict & manage early claims and reduce their overall loss ratio. iTuring built a new model in only 70 minutes and improved the predictive accuracy over the company’s existing model by 15%. In addition, the false positive rate was reduced by ~7%, thereby reducing administrative cost and improving customer experience.

Highlights

At the start of the project - 80% predictive accuracy (AUC) for the existing early claim (pre-issuance risk verification) model
2 weeks from project start to production-ready application
400+ features created
AI model built in 70 minutes with iTuring AutoML+
92% model accuracy (AUC) vs. 80% of previously built model
Reduced false positive rate by ~7%

Impact

Predictive Accuracy

0 %↑

Time to Deployment

0 Weeks

False Positives

0 %↓

Why CyborgIntell

Not only does iTuring provide highly accurate predictions of which customers are likely to claim early, it also provides explanations for why each customer is likely to claim. This allows business users to better understand their client’s behaviour and provide personalized, risk-based pricing recommendations.

Early Claim Prediction