Node Graph Data Storage for Real-Time Cross-Clinic Patient Records

The patent segments patient data into discrete nodes (patient nodes, visit nodes, diagnosis nodes, etc.) that can be independently stored and managed. Each node contains specific attributes related to a particular aspect of patient care, allowing the system to handle complex medical data through modular, manageable units rather than monolithic database entries.

The patent transitions from traditional tabular database structures to a graph-based data model with multiple dimensions. Nodes can have multiple attributes and relationships, enabling the system to represent complex medical relationships (patient-visit-diagnosis-treatment) in a multi-dimensional space that improves retrieval efficiency while managing complexity through structured relationships.

The patent merges data from multiple clinics into a unified graph structure where related data elements are consolidated into shared nodes. When the same patient visits multiple clinics, their demographic information, medical history, and other static data are stored once in a single patient node, while visit-specific data are stored in separate visit nodes that reference the patient node, eliminating redundancy and reducing overall storage requirements.

The patent uses reference copying rather than data duplication. Instead of copying patient demographic data into each visit record, the system creates visit nodes that contain references (foreign keys) to the patient node. This allows comprehensive patient history to be accessed from any visit node without actually duplicating the underlying patient data, significantly reducing memory requirements.

The patent performs preliminary organization of data into a graph structure during data ingestion, creating nodes and edges that represent relationships between patients, visits, diagnoses, and treatments. This preliminary structuring enables efficient querying later, as the system can traverse pre-established relationships rather than performing repetitive full-table scans or complex joins across multiple clinic databases.

The patent introduces visit nodes as intermediaries between patient nodes and diagnosis/treatment nodes. These intermediary nodes capture the temporal and contextual relationships of specific clinic visits, allowing the system to efficiently retrieve visit-specific information while maintaining connections to the broader patient history, thus reducing retrieval time without excessive structural complexity.