VISTA Architect: A graph database-oriented health AI system demonstrated in multidisciplinary tumor boards

2026-06-21

AI Summary

This study addresses the inefficiency in reconstructing longitudinal patient histories and the frequent loss of temporal information during multidisciplinary tumor board consultations. To this end, the authors propose a graph database–oriented health AI architecture featuring a novel dual-layer design: the MEDS knowledge graph and the Timeline Object Architecture (TOA). This framework transforms electronic health records into a provenance-aware, structured knowledge graph in a single pass and hierarchically abstracts it into clinical timelines, enabling efficient querying and validation. Leveraging graph-guided large language model–based event extraction and an intelligent agent interface, the system substantially enhances both the efficiency and accuracy of clinical information synthesis. Evaluated on 1,180 patient records, the approach achieves an average accuracy of 96.4% (9.75/10) for key variables—outperforming a BM25 RAG baseline—and reduces case preparation time to approximately 2.2 minutes per consultation without compromising accuracy.

Abstract

We introduce VISTA Architect, a database-oriented AI architecture for integrating large language models (LLMs) with longitudinal electronic health records (EHRs). At ingestion, it transforms complex clinical documentation into a persistent, provenance-linked knowledge graph, eliminating repeated reprocessing of raw records at query time. The architecture has two layers: a source-faithful MEDS Graph preserving granular EHR structure with full provenance, and a clinically abstracted Timeline Object Architecture (TOA) that uses graph-guided LLM extraction to synthesize a concise timeline of deduplicated, temporally coherent clinical events. This addresses key limitations of direct long-context prompting and retrieval-augmented generation (RAG), which often miss temporal relationships and incur high cost and latency from repeated raw-text processing. By precomputing clinical synthesis once, downstream queries access an organized patient state and traverse to source documentation only when detailed verification is needed. We demonstrate the system in multidisciplinary thoracic oncology tumor boards at Stanford Medicine, where precise reconstruction of patient histories is critical. Across 1,180 patients, VISTA Architect achieved 96.4% accuracy (mean 9.75/10) on 15 tumor board-salient variables (17,700 evaluations; 95% CI 96.1-96.7%), surpassing a matched BM25 RAG baseline and recent benchmarks for LLM-based clinical extraction. An agentic interface reduced preparation for a 30-patient held-out cohort to about 2.2 minutes without sacrificing accuracy. While configured here for thoracic oncology, the modular design adapts to other specialties through customizable event definitions, episode structures, and agentic tools; validation beyond thoracic oncology remains future work.

Problem

• electronic health records
• temporal relationships
• clinical timeline reconstruction
• multidisciplinary tumor boards
• longitudinal data

Innovation

• knowledge graph
• large language models
• electronic health records
• temporal coherence
• graph-guided extraction