Linus Health: Transitioning to Durable Workflows in Cognitive Healthcare

Abstract

Linus Health’s AI-enhanced digital cognitive assessment platform enables early detection of impairment, while actionable clinical guidance and personalized patient action plans inform next steps.

Prior to using Temporal, our workflow architecture was completely custom-built over time and resulted in increasing concerns over performance, consistency, traceability, and maintenance. With our growing customer base and addition of new complex analytics patterns to our core workflows, we decided to start using Temporal to address these architectural gaps.

To transition legacy workflow patterns into Temporal, we designed new infrastructure components as well as code-level paths and abstraction layers, which allowed us to translate our analytics engines and subsequent processing into Temporal activities, signals, and parent-child workflows. We also explored supporting cross-language compatibility within our platform, monorepo configuration, and running legacy and Temporal workflows in parallel.

About the Presenter

Diane Nguyen is a software architect at Linus Health, where the team introduced Temporal as our new workflow engine to replace our homegrown orchestration system. She collaborates with various engineering teams when it comes to new architecture and product features. She is passionate about reliability, state management, orchestration, and improving the developer experience.