TRACE: Temporal Reasoning via Agentic Context Evolution for Streaming Electronic Health Records (EHRs)

TL;DR

TRACE introduces a novel framework for temporal reasoning in streaming EHRs by structuring context with a dual-memory system and agentic components, improving accuracy and safety without increasing computational overhead.

Contribution

It presents a new method for temporal clinical reasoning that maintains a structured memory and agentic components, avoiding fine-tuning or retrieval overhead.

Findings

Significantly improves next-event prediction accuracy.

Enhances protocol adherence and clinical safety.

Produces interpretable and auditable reasoning traces.

Abstract

Large Language Models (LLMs) encode extensive medical knowledge but struggle to apply it reliably to longitudinal patient trajectories, where evolving clinical states, irregular timing, and heterogeneous events degrade performance over time. Existing adaptation strategies rely on fine-tuning or retrieval-based augmentation, which introduce computational overhead, privacy constraints, or instability under long contexts. We introduce TRACE (Temporal Reasoning via Agentic Context Evolution), a framework that enables temporal clinical reasoning with frozen LLMs by explicitly structuring and maintaining context rather than extending context windows or updating parameters. TRACE operates over a dual-memory architecture consisting of a static Global Protocol encoding institutional clinical rules and a dynamic Individual Protocol tracking patient-specific state. Four agentic components, Router, Reasoner, Auditor, and Steward, coordinate over this structured memory to support temporal inference and state evolution. The framework maintains bounded inference cost via structured state compression and selectively audits safety-critical clinical decisions. Evaluated on longitudinal clinical event streams from MIMIC-IV, TRACE significantly improves next-event prediction accuracy, protocol adherence, and clinical safety over long-context and retrieval-augmented baselines, while producing interpretable and auditable reasoning traces.