MyMoai: Reimagining insurance to create stronger communities

MyMoai is a decentralized insurance app inspired by Okinawa’s Moai system. Users form groups, pool funds, and manage claims, reducing premiums and increasing transparency.

Comment

Inspiration

In recent years the Insurance industry has received pushback from customers for various reasons spanning inefficient claim speeds to unfair premium rates. Most, if not all, issues arise from the industries' inherited “centralized” system putting insurance companies at the forefront of plans as opposed to the people.

Our inspiration to this dilemma stems from Okinawa, a prefecture island in the Southern Japanese archipelago. Anthropologists have researched this area due to its “blue zone” which indicates a significant population. Among diet, lifestyle decisions, and exercise, a tradition unique to this area is an ancient practice called a “Moai”.

A Moai is a tightly-knit group of individuals (traditionally formed at birth) who are connected for life in terms of career goals, support, and social events. One of their nuanced use cases is also as an insurance system. For example, in a recent documentary on blue zones, Moai described how they helped pay for one member’s surgeries at a time when she could not afford it. Moai’s effectively replaced traditional insurance systems in the region. This community-based insurance format inspired the foundations behind our insurance technology (InsureTech), MyMoai, a novel decentralized insurance technology application.

What it does

MyMoai is split into two offerings allowing users to create Moai’s online.

A Moai is a collection of users with a shared interest or mutual connection, pooling money together to act as an “insurance fund”. Without the presence of a middleman, insurance premiums drastically decrease. At the moment, claims are managed by the users in the Moai themselves, just like in real life. MyMoai employs blockchain technology to track claims and purchases, offering secure protection against fraud. This decentralized ledger records every transaction, providing a transparent and immutable history. The blockchain network is exclusive to each Moai, ensuring privacy and security. Only members of a specific Moai can access and engage with their respective blockchains.

Users are also allowed to join several Moais for different purposes and can manage their memberships within the platform. Users are also able to analyze Moai groups from several statistics such as credit scores, members, employment, and Moai transaction histories.

The most probable use cases for Moais are local, which can be created for families or friend groups.

There is a built-in search function for users to find and join public Moais, this is somewhat less likely to be used, unless in scenarios where users have a common interest/goal (pet lovers -> pet insurance). Users are recommended Moais based on their demographics and various metrics in addition to a karma system that prioritizes active members who are less at risk.

Finally, our most novel use case for Moais is for short-term, one-time, group trips for popular events such as college spring break. Here, users can create a limited Moai to cover risks associated with the trip.

How we built it

UI/UX: The front end of MyMoai was developed through Flutter in Android Studio, allowing us to work with web, iOS, and Android apps all at once. Flutter allowed us to have better compatibility between our screens and our backend. Our data storage was made through Firebase, utilizing Firestore Databases to keep track of both user and Moai data.

The user collection contains basic profile information as well as financial and lifestyle data that is used to recommend Moais for the user to join with our back-end algorithm. The Moai collection

Recommendation System: Using mostly Python and Scikit the ML algorithm harnesses the power of K-Means clustering, augmented with a sophisticated data preprocessing pipeline that includes feature scaling, one-hot encoding, and dimensional reduction techniques. This ensures a robust foundation for high-dimensional data analysis, optimizing the algorithm's capacity to discern nuanced user patterns and preferences.

The system's advanced predictive modeling is further enhanced by custom transformers like FeatureWeighter and CategoricalFeatureWeighter, employing weighted feature engineering to emphasize critical user attributes. This, combined with strategic use of euclidean distances for cluster proximity analysis, enables the algorithm to deliver hyper-personalized moai group recommendations. By integrating techniques such as inverse transformations for interpretability, our system not only recommends with precision but also elucidates the logic behind its suggestions.

This recommendation system was then leveraged and adapted to recommend users to different Moais based on their interests, financial standing, and mutual networks.

Challenges we ran into

At the outset of the college hackathon, we were immediately daunted by the task of brainstorming ideas. The challenge of building a mobile app with a machine learning (ML) backend within the hackathon’s tight timeframe posed several significant hurdles. The time constraints were a major obstacle, with the looming deadline leaving little room for extensive model training and optimization. The technical complexity of integrating ML models into our mobile app required a deep understanding of both mobile development frameworks and ML libraries, which was a tough ask for us as novice hackers. Additionally, acquiring and preprocessing data for training ML models was challenging within the hackathon’s timeframe, as it often involved cleaning and annotating datasets, which was time-consuming. Despite these challenges, we believe that building a mobile app with an ML backend offers immense potential. Addressing these challenges demanded strategic planning, effective teamwork, and adaptability, and we feel that our final product was very successful.

Accomplishments that we're proud of

*Successful Integration of Complex Algorithms into a User-friendly Interface: One of our most notable achievements was the seamless integration of advanced machine learning algorithms into an accessible and engaging user interface. This bridging of complex backend processes with a straightforward frontend experience represents a significant accomplishment, showcasing our ability to make cutting-edge technology approachable for users.

*Cross-disciplinary Collaboration and Innovation: We're particularly proud of the collaborative spirit that characterized our project, bringing together expertise from data science, software development, and design. This multidisciplinary approach not only enriched our project but also fostered an environment of innovation, allowing us to explore novel solutions and technologies.

*Adaptation and Application of APIs for Language Interoperability: Successfully leveraging APIs to convert Python code to Dart was a milestone that demonstrated our team's adaptability and technical skill. This accomplishment not only allowed us to extend the functionality of our Python-based algorithms but also opened up new avenues for app development, enhancing our project's scalability and reach.

  • Creation of a Dynamic and Responsive App Using Dart and HTML: Developing an app and UI with Dart and HTML that is both dynamic and responsive was an achievement that we take particular pride in. This accomplishment underscores our commitment to delivering an exceptional user experience, characterized by a sleek design and intuitive navigation.

*Tailoring Complex Recommendations to Individual Users: Perhaps our most significant technical achievement was the development of a recommendation system that not only analyzes complex user data but also provides personalized moai group suggestions. This system's ability to distill intricate data patterns into actionable recommendations reflects our project's core innovation and our dedication to enhancing user engagement.

What we learned

In the journey of developing our recommendation system, we learned the immense value of integrating diverse technological tools and languages to create a seamless, user-focused application. The transition from Python's powerful data processing capabilities to Dart for app development underscored the importance of cross-platform fluency and the potential of APIs in bridging different programming environments. This process not only enhanced our proficiency in both languages but also deepened our understanding of how to effectively translate complex data science models into intuitive, user-friendly applications. Crafting the UI with Dart and HTML, we appreciated the nuanced art of UI/UX design, striving to ensure that our sophisticated backend algorithms translated into a simple, engaging user experience. This project was a testament to the synergy between backend intelligence and frontend elegance, teaching us that the true power of technology lies in its ability to disappear into the fabric of a seamlessly delivered service.

What's next for MyMoai: Reimagining insurance to create stronger communities

The future of MyMoai involves several key developments. Firstly, to facilitate financial transactions, MyMoai is planning to establish connections with credit card, banking, or digital wallet services. This will involve partnering with various financial institutions and service providers to build the necessary financial infrastructure. The team is also considering the implementation of a Stripe-powered solution to enable swift digital transactions.

Secondly, MyMoai will need to define legal restrictions for each Moai and for users joining the network. This will necessitate the development of an efficient and secure digital contract system. Such a system will ensure that payments to each Moai are made as agreed and that Moai members can create clear claims. These claims will define their entitlement to different amounts of money within each group.

Lastly, MyMoai plans to incorporate social media information. This data will be used to inform Moai suggestions, enhancing the user experience and fostering a sense of community within the platform. In summary, MyMoai’s next steps involve significant advancements in financial transaction capabilities, legal frameworks, and social media integration. These developments aim to enhance the functionality and user experience of the platform.

Feel free to reach out to any developers with questions or comments. We are extremely passionate about this project and would love to know any opinions on this. Furthermore, feedback never hurts! Thank you Hoohacks 2024

Share this project:

Updates