A Hybrid Neuro-Fuzzy Network for Underwriting of Life Insurance

Abstract

This paper is a study of life insurance underwriting for individuals and the various processes involved in providing cover to the insurance seeker. Underwriting is the process of classification, rating and selection of risks among the diverse insurance applications. The paper explores various steps taken in insurance underwriting to rate the insurance application, the various risk factors considered for taking an insurance underwriting decision and proposes a model for insurance underwriting. A robust method for automating the underwriting process through the hybrid methodology is presented, which comprises of Neural Network and Fuzzy Logic. Neural Network in the proposed model can be trained to traditionally perform the insurance underwriting for individuals, while the fuzzy rules are expected to take care of the fuzzy inputs. An insurance application is compared against various pre-defined standards and is classified into one of the risk classes. Based on the risk classes, the premium to be paid by the applicant is determined - higher the risk class, higher is the premium. The application of Artificial Neuro-fuzzy Network in insurance underwriting process can considerably reduce the time and insurance losses.