Drivers of Mortality – A Study Using Artificial Intelligence and Machine Learning Techniques

Life expectancy is the average number of years that a person can expect to live. Mortality and life expectancy have important implications in pricing and valuation of life insurance products, particularly for products with periods of premium guarantees. Understanding drivers of life expectancy and mortality also plays an important role in product design, underwriting and risk management of life insurance products. Advancements in the field of Artificial intelligence (AI) have led to the adoption of machine learning models in a wide domain of applications. While machine learning is known for its predictive capabilities, most of the models suffer from a relative lack of explainability and are based on correlation of parameters instead of causation. One example of the distinction between correlation and causation is the observation that when ice cream sales go up, drowning increases. While the two variables can be positively correlated, there is a lack of causal dependencies. Causal reasoning algorithm is one of the latest advancements in AI and machine learning techniques that attempts to establish causal relationship and provide a basis for sophisticated human-like analysis of interventions and counterfactuals. This has led to greater explanatory power delivered by causal models over other models based on statistical correlation alone. This paper uses the latest algorithm of causal reasoning in discrete probability trees to analyse causes of mortality, and looks at effect of potential interventions on enhancing life expectancy. The paper also compares other models to exam their predictive and explanatory capacity. Lastly, the paper discusses the application of causal reasoning and implication for life insurance. Traditional methods of assumption setting and pricing rely on study of historical claims experience for statistically significant correlation, with judgement applied on the choices and sequence of drivers in shaping the decrement tables. Greater explanatory power delivered by causal models can improve understanding of real causes of claims experience and deliver fairer pricing outcomes for customers. It can also inform product designs and potential interventions to improve claims outcome.

Find the Q&A here: Q&A on 'The Digital World and Life Insurance'