Architecting Synergy: Knowledge Graphs for Representing Complex, Multi-Domain Patient Information.

Area of Science:

• Health Informatics
• Information Architecture
• Clinical Data Management

Background:

• Despite increased healthcare data exchange, critical care gaps persist, including missed referrals, polypharmacy risks, and continuity issues.
• Existing information architectures struggle to represent the dynamic, multi-domain, and intent-driven nature of patient care.
• Empirical failures in primary care (scheduling, referrals, labs, medications) and broader care lapses highlight the need for improved data representation.

Purpose of the Study:

• To identify the information architecture requirements for representing patient care as a dynamic, multi-domain, intent-driven process.
• To evaluate candidate architectures based on empirical failure modes and derived functional and technical requirements.
• To determine if knowledge graphs can address the limitations of current healthcare data exchange systems.

Main Methods:

• Analysis of empirical failure modes in Canadian primary care (appointment scheduling, referrals, labs, medication processes).
• Identification of functional and technical requirements for an ideal healthcare information architecture.
• Evaluation of four candidate architectures against the derived requirements, focusing on semantic linkage, schema flexibility, state tracking, longitudinal reasoning, and explainability.

Main Results:

• Knowledge graphs uniquely satisfy the requirements for semantic linkage, schema flexibility, state tracking, longitudinal reasoning, and explainability.
• Other evaluated architectures failed to meet the comprehensive set of functional and technical requirements.
• Knowledge graphs demonstrate a strong potential for overcoming current data exchange limitations.

Conclusions:

• Knowledge graphs present a promising solution for achieving truly interoperable, safe, and patient-centered healthcare.
• Addressing regulatory, governance, and vendor implications is crucial for the successful adoption of knowledge graphs in healthcare.
• Implementing knowledge graphs can significantly enhance the representation and management of dynamic patient care processes.