Development of an Interoperable and Easily Transferable Clinical Decision Support System Deployment Platform: System...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision Support Systems

Background:

• Clinical decision support systems (CDSS) improve healthcare efficacy and safety but face challenges due to data standards and interoperability.
• Existing platforms often lack the flexibility to fully integrate diverse CDSS algorithms and data formats.

Purpose of the Study:

• Introduce the Common Data Model-based Intelligent Algorithm Network Environment (CANE) platform.
• Facilitate the implementation and deployment of CDSS by addressing interoperability and data standardization issues.

Main Methods:

• Deployed CDSS reasoning engines (R/Python objects) into the CANE platform, converting them to C#.
• Processed patient data in Health Level 7 Fast Healthcare Interoperability Resources (FHIR) format within the EHR system.
• Utilized preprocessing, reasoning, and integration modules for data conversion, algorithm execution, and inter-institutional data sharing.

Main Results:

• Developed a CANE system allowing direct algorithm calls via RESTful APIs when integrated with EHR systems.
• Successfully deployed eight algorithms, including knowledge-based sepsis screening and treatment guidance, and non-knowledge-based acuity/prognosis prediction for triage.
• Demonstrated enhanced clinical decision support for conditions like sepsis and improved emergency resource allocation.

Conclusions:

• Successfully developed a common data model-based platform adhering to medical informatics standards.
• The CANE platform supports the deployment of artificial intelligence models (R/Python) within healthcare systems.
• The platform enhances CDSS capabilities by enabling seamless integration and interoperability.