PrivMeSA: Privacy-Aware Self-Evolving Multi-Agent System for Medicine via Local-Remote LLM Collaboration

Researchers introduced PrivMeSA, a privacy-aware self-evolving multi-agent system that lets a local clinical LLM agent consult remote specialists while learning to control disclosure, according to an arXiv paper (arXiv:2609.38458v1). On an emergency-department benchmark built from MIMIC-IV-ED records, PrivMeSA improved mean task accuracy over delegation by up to 15.8 percentage points while cutting disclosure of personal details from 98.0% to 0.2% of cases and reducing the share of cases where a patient could be narrowed to ten or fewer registry patients from 74% to 0%. The system uses reinforcement learning to balance accuracy against direct disclosure and registry-based re-identification risk, and a local lesson memory that distills completed consultations into generalized clinical guidance for reuse without further remote exchanges.

arXiv:2609.38458v1 Announce Type: new Abstract: Clinical large language model LLM agents deployed locally can consult more capable remote models, but doing so risks exposing patient information. Privacy-conscious delegation places disclosure decisions with a local agent, yet removing explicit identifiers is insufficient: quasi-identifiers can accumulate across multi-turn consultations and repeated patient visits to enable re-identification. We introduce PrivMeSA, a privacy-aware self-evolving multi-agent system that learns to control disclosure and retains remote expertise for local reuse. A local agent manages each encounter and consults remote specialists that may request additional information. Reinforcement learning balances task accuracy against direct disclosure and registry-based re-identification risk, with privacy evaluated over the complete outbound transcript of each encounter. A local lesson memory distills completed consultations into generalized clinical guidance and retrieves relevant lessons before transmission, allowing subsequent cases to reuse expertise without another remote exchange. Memory grows without additional outcome labels or parameter updates. On an emergency-department benchmark built from MIMIC-IV-ED records, PrivMeSA improves mean task accuracy over delegation by up to 15.8 percentage points. In the same setting, PrivMeSA reduces the disclosure of personal details from 98.0% to 0.2% of cases and the share of cases in which the patient can be narrowed to ten or fewer registry patients from 74% to 0%.