CSE498, Collaborative Design, Spring 2023
Computer Science and Engineering
Michigan State University

Roosevelt Innovations, LLC is the first technology solution to deliver a simple, seamless, and smart platform for health insurance companies. With industry-leading claims processing capabilities, Roosevelt has transformed operations, enabling a total treatment cost savings of $972 million. Additionally, Roosevelt has more than 22 million users on its platform and an industry-leading auto-adjudication rate of 95%.

When processing many insurance claims, there is a large amount of data generated. Data that is anomalous for any reason is cause for further investigation.

Our Provider Analysis Toolkit gives data analysts of Roosevelt Innovations a streamlined, efficient, and verbose view of the data that comes from processing claims. Users can view real-time data, exchange their thoughts with other analysts on specific data, and share reports with other analysts.

Our solution identifies statistical anomalies within the data that can help them further investigate the causes for these anomalies. Data is filtered for the analysts so that the most anomalous data is easily accessible to them in a simple, quick, and efficient manner, requiring no further mathematical calculations on the analysts’ end.

Data visualizations are also provided. Whether the data is represented in tables or charts, everything is laid out in a way that is streamlined and easily digestible.

Our solution automatically flags anomalous data and provides a suite of tools for further analysis, increasing productivity of analysts.

The web application is written in Python with utilization of the Streamlit library for the front-end framework, as well as FastAPI are used for construction of the web application’s APIs. The anomaly detection machine learning models were developed using the PyTorch framework and scikit-learn library, while the data is stored within a Snowflake database.