Design and implementation of serverless architecture for i2b2 on AWS cloud and Snowflake data warehouse
Metadata[+] Show full item record
Informatics for Integrating Biology and the Beside (i2b2) is an open-source medical tool for cohort discovery that allows researchers to explore and query clinical data. The i2b2 platform is designed to adopt any patient-centric data models and used at over 400 healthcare institutions worldwide for querying patient data. The platform consists of a webclient, core servers and database. Despite having installation guidelines, the complex architecture of the system with numerous dependencies and configuration parameters makes it difficult to install a functional i2b2 platform. On the other hand, maintaining the scalability, security, availability of the application is also challenging and requires lot of resources. Our aim was to deploy the i2b2 for University of Missouri (UM) System in the cloud as well as reduce the complexity and effort of the installation and maintenance process. Our solution encapsulated the complete installation process of each component using docker and deployed the container in the AWS Virtual Private Cloud (VPC) using several AWS PaaS (Platform as a Service), IaaS (Infrastructure as a Service) services. We deployed the application as a service in the AWS FARGATE, an on-demand, serverless, auto scalable compute engine. We also enhanced the functionality of i2b2 services and developed Snowflake JDBC driver support for i2b2 backend services. It enabled i2b2 services to query directly from Snowflake analytical database. In addition, we also created i2b2-data-installer package to load PCORnet CDM and ACT ontology data into i2b2 database. The i2b2 platform in University of Missouri holds 1.26B facts of 2.2M patients of UM Cerner Millennium data.