MedRLM Framework Uses Recursive Agents for Long-Context Clinical Reasoning

MedRLM coordinates specialized agents to recursively inspect, decompose, and synthesize heterogeneous patient data including EHRs, medical images, and sensor streams, rather than compressing everything into a single prompt. It targets community-to-tertiary referral optimization and sensor-guided screening as concrete clinical workflows.