Event-Driven Clinical Decision Support: Securing Medical Logic Through Structural Fault Isolation.

Area of Science:

• Computer Science
• Health Informatics
• Software Engineering

Background:

• Monolithic clinical decision support architectures create complex regulatory validation challenges by tightly coupling IT infrastructure with medical logic.
• Existing systems face difficulties in integrating diverse data and adapting clinical logic due to intertwined components.

Purpose of the Study:

• To present an event-driven microservice architecture for clinical decision support systems.
• To streamline the clinical validation process and reduce regulatory overhead.
• To enhance the reliability and adaptability of clinical decision support.

Main Methods:

• Implemented an event-driven microservice architecture orchestrated by Kafka and a Drools rule engine.
• Designed the system to intercept and shed structurally invalid data before processing by the medical logic core (ArdenSuite).
• Conducted empirical testing to evaluate fault tolerance and performance under high-stress scenarios.

Main Results:

• Demonstrated zero state loss during node failures, ensuring system resilience.
• Confirmed that shedding malformed data at the perimeter prevents computational bottlenecks.
• Reduced mean evaluation time per valid input during high-stress scenarios.

Conclusions:

• The proposed architecture decouples IT reliability from clinical safety through strict architectural boundaries.
• This approach confines clinical validation to an isolated interpretation engine, accelerating integration and adaptation.
• Provides a foundation for agile, compliant, and evolving clinical decision support systems.