Physics-Informed Multi-Agent Coordination for Hospital Patient Flow Optimization

2026-09-29

AI Summary

This study addresses the limitations of traditional models in handling dynamic changes and the unsuitability of centralized reinforcement learning for decentralized governance in hospital patient flow coordination. We propose a physics-informed multi-agent coordination framework that embeds an empirically calibrated BCMP queueing network as a physical prior within a Dec-POMDP architecture. By integrating spatially decomposed rewards with local action fingerprint exchange, the method enables autonomous inter-departmental routing and dynamic resource scaling, effectively mitigating non-stationarity while reducing communication overhead. Experiments on MIMIC-IV data demonstrate that the proposed framework significantly reduces cumulative system delay while satisfying clinical safety constraints, outperforming both static approximation and independent baseline methods.

Abstract

Efficient patient flow coordination across autonomous hospital departments is critical for mitigating overcrowding and balancing resource utilization. While classical queueing theory, specifically open Baskett--Chandy--Muntz--Palacios (BCMP) networks, provides an interpretable mathematical topology for healthcare operations, analytical models rely on stationary assumptions and fixed routing matrices that degrade under state-dependent real-world dynamics. Conversely, centralized reinforcement learning approaches struggle to accommodate the decentralized structure of hospital governance, where individual clinical departments function with localized observations, heterogeneous resources, and divergent operational objectives. In this paper, we present a Multi-Agent Systems (MAS) framework titled \emph{Physics-Informed Multi-Agent Coordination}, which embeds empirically calibrated BCMP queueing topologies as physical priors within a decentralized multi-agent reinforcement learning architecture. Formulated as a Decentralized Partially Observable Markov Decision Process (Dec-POMDP) under coupled resource constraints, our method enables autonomous departmental agents to cooperatively negotiate patient routing and dynamic service scaling. To mitigate environmental non-stationarity without inducing excessive communication overhead, agents exchange localized action fingerprints along network edges and optimize a spatially decomposed reward structure. Empirical evaluations driven by real-world MIMIC-IV patient trajectories indicate that this cooperative multi-agent approach substantially reduces cumulative system delay compared to static Markovian approximations, heuristic dispatching, and independent multi-agent baselines, while maintaining clinical safety constraints.

Problem

• Patient Flow Optimization
• Multi-Agent Coordination
• Hospital Overcrowding
• Dec-POMDP
• Resource Utilization

Innovation

• Physics-Informed Multi-Agent Reinforcement Learning
• Dec-POMDP
• BCMP Queueing Networks
• Patient Flow Optimization
• Action Fingerprints