InsuranceRiskModel

This hackathon project is a part of our larger life insurance protocol. The overarching goal here is to build a life insurance protocol that eliminates the need for regulators just like Uniswap has done for exchanges.

Needless to say, it is a lot more difficult to do this for insurance. Our aim here has been to satisfy the needs for two regulators; data regulator and insurance regulator

With this in mind All sensitive data given to us by policyholder is encrypted using Lit and is stored in a decentralised database Subsequent access to this data is programmatically controlled with open source code. If a claim is made (less than 2% of the issued policies), the data is decrypted and made visible to the claim assessment team only (managing the life cycle of this decrypted data is beyond the scope of this project). If there is no claim i.e. for 98% of the policies, the data is never seen by any human To calculate solvency requirements, the regulator runs a stochastic simulation on the above policy data to determine the loss scenario at the 99.5th centile. Given our commitment to openness, this simulation has been built on Lit actions. The code is open and available for audit for regulators. Moreover, they can run it themselves with different sets of assumptions to ensure the protocol’s solvency. As an aside, this kind of functionality would also be a great candidate for Lit's upcoming general purpose compute within TEE In our usecase, no code needs to remain private

We believe this ability to run code on top of encrypted data with decentralised key management opens up disruption of other industries/businesses that currently need regulatory oversight and hence special access to their offices.

Tech Stack Overview

In developing the insurance application, we employed a robust and versatile tech stack to ensure data privacy, scalability, and verifiability in our processes.

Encryption and Security Security is a central focus of the application, especially given the sensitive nature of the data involved. We integrated Lit Protocol to enable advanced encryption and decryption mechanisms using shared partial private keys. This ensured that all PII, as well as specific organizational weights used in calculations, were stored in an encrypted form. Lit Protocol's distributed key management allowed for secure, decentralized encryption, ensuring compliance with privacy standards while providing granular control over access to sensitive data.

Front-end Development The application’s user interface was built using React.js. The front end facilitates the purchase of insurance policies, guiding users through a set of personal questions. Additionally, it allows users to run simulations by inputting values such as portfolio size and the number of simulations. React.js was chosen for its component-based architecture, enabling efficient development of reusable UI components that offer real-time data visualizations and a seamless user experience.

Database and Storage Solutions For data management, we explored a few options before settling on a hybrid approach. Initially, we evaluated OrbisDB for its potential as a relational database on Ceramic, but encountered data management challenges that impacted our ability to meet our deadlines. As a result, we opted to use Irys Database for encrypted data storage. SQLite was used as an in-memory database to manage transient data during operations. This combination allowed us to store sensitive data, including personally identifiable information (PII) and proprietary calculations, in a secure and manageable way.

Simulation and Computation To run the core simulations, we leveraged Lit Action, which allows for executing verifiable, auditable code on encrypted cohort data. These simulations focus on calculating various insurance-related metrics such as mortality rates for different age groups, projecting claims scenarios, and estimating associated claims costs. The ability to run these calculations in a privacy-preserving manner ensures both transparency and security, as all actions can be audited without exposing sensitive data.

Business Intelligence and Analysis A crucial component of our application is understanding the relation between portfolio size and profitability. The simulation results, after being processed by Lit Actions, are visualized to demonstrate how changes in portfolio size impact the organization’s profitability. This analysis is essential for strategic decision-making and is rendered in simple graph that allow stakeholders to easily interpret complex data sets.

Acknowledgments Finally, we would like to extend our sincere thanks to Charles from Orbis, Eli from Lit, and others we collaborated with throughout this development process. Their guidance and insights were invaluable in shaping our final implementation.