State-of-the-Art Review of Life Insurtech: Machine learning for underwriting decisions and a Shift Toward Data-Driven, Society-oriented Environment

Machine learning has been used by insurance companies for nearly a decade to identify potential risks and improve underwriting decisions. Nonetheless, there is a lack of systematic survey articles on state-of-the-art (SoTA) machine learning techniques, especially with respect to Society-oriented Environment models, developed to tackle these problems. This article begins by outlining the limitations of current systems used in this domain, including interpretability and explainability constraints, privacy issues, and the credibility of machine learning constructions along with their solutions. It then provides an extensive review of state-of-the-art machine learning algorithms such as Explainable AI, Privacy-Preserving Techniques, Federated and Transfer Learning, Sharpley, as well as others developed and applied for various Life Insurtech problems. The article also examines the existing challenges and future trends in developing machine-learning-based underwriting decision approaches. We believe that this survey can offer practical guidance for building next-generation insurance risk management systems.