AI and machine learning are enabling insurers to automate more underwriting decisions, optimize risk selection and improve the customer experience, innovators in the life and health space told Insurtech Connect Asia.
Asia is a rapidly maturing market for artificial intelligence (AI) and machine learning (ML) development – and these tools are transforming underwriting in Asia’s life and health insurance space, Insurtech Connect Asia delegates heard in a webinar from Lee Sarkin, Chief Analytics Officer (Life and Health) – APAC, Middle East and Africa for Munich Re, and Dr. Deepak Gandhi, Head of New Business Underwriting for Great Eastern Life Insurance.
Going forward, insurers will be able to leverage a growing range of data sources to drive value through AI and ML, from their own auto-underwriting and back-office data to agent behaviour data and unstructured data such as customer photos and videos, augmented by an ever-expanding array of external and public datasets. This, they explained, will enable insurers to automate more of the underwriting process, streamline customer journeys and gather deeper insights, informing better risk selection, pricing and decision-making.
“AI and ML can assist with many of the pain points most life and health insurers currently deal with,” said Sarkin, highlighting the fact that, despite many insurers already having some form of rule-based automated underwriting engine in place, a high proportion of standard or good risks are still referred for manual underwriting before being accepted.
To tackle this issue, Munich Re and Great Eastern Life developed an AI-ML model which was trained on auto-underwriting data and the back-office data from insurers to predict the decisions made by underwriters – allowing many more standard decisions to be automated. This resulted in the proportion of cases Great Eastern Life straight through processes (STP) more than doubling from 30% to 65%, freeing up human underwriting capacity while shortening onboarding times and reducing requests for medical evidence.
“The model allows us to make offers at the point of sale. It’s a win-win for agent, customer and organization,” Gandhi explained.
Other pain points being addressed by AI and ML include reducing and personalizing the questions for customers and identifying anti-selective behaviour from agents and the mis-disclosure of factors such as body mass index, smoking or medical conditions by customers (through analysis of selfies, for example), meaning high-risk cases can be flagged earlier.
Further applications include underwriter quality checks, evaluating the effectiveness of medical evidence, identifying signals for early claims and optimizing the referral process workflow itself. “End-to-end, this is driving a more streamlined and robust selection process that transforms customer experience and the efficiency of insurance,” Sarkin said.
Crucially, claims data must also be fed back into these models to ensure their decision-making continues to improve. “Predicting back-office underwriting decisions is replicating the existing process – feeding claims experience back improves the accuracy,” Sarkin said.
However, applying these tools brings some risk that must be carefully managed. Any errors in the models, for example, will reflect in future claims and profitability, so underwriters must think carefully about how much of their business to STP. “It is important to quantify, minimize and accept the risk this introduces,” Sarkin said. “We also need to be mindful of the extent to which automation could affect the customer experience,” Gandhi added.
Nevertheless, there is no doubt AI and ML will continue to play a growing role in underwriting going forward – in life and health and beyond. “There are so many things that can be done,” said Gandhi. “It’s helping us reimagine and reinvent the whole underwriting process.”