The challenge

A leading U.S. healthcare payer required the capability to provide analytical services to both internal and external customers.

Atos was engaged to perform the following objectives:

• Implement Analytics as-a-Service (AaaS) for diabetes treatment, Rx rejections and readmission risk
• Build a scalable, independent deployment service platform
• Expose the analytics API to internal teams as well as external stakeholders
• Enable fast time to market for this new service line

The solution

Atos worked with the health insurer to understand its business areas and the purpose of the machine learning models, in order to develop the architecture for development and deployment. Atos was involved with:

• Designing and implementing analytics services using a microservices architecture
• Building a platform to deploy and evaluate machine learning models with R and Python
• Exposing the endpoints to be used by microservices
• Managing APIs for consumption by external customers and user engagement via web and mobile devices
• Microservices testing and deployment on cloud

Business benefit

After just two months, Atos was able to get the health insurer’s machine learning API up and running.

Benefits include:

• Algorithms and business rules produce patient predictions based on health records
• Easy plug-and-play analytics across the organization
• Improved business agility for accessing predictive models
• Increased revenue potential with a new analytics service line
• Automated model deployment and productivity improvements
• Robust platform to build, deploy and evaluate machine learning models
• Open source technology with no license cost