Multi-Agent Reinforcement Learning for Intraday Operating Rooms Scheduling under Uncertainty

One-line Summary

This study presents a multi-agent reinforcement learning framework for scheduling operating rooms, outperforming traditional methods in balancing elective and urgent surgeries under uncertainty.

Plain-language Overview

Scheduling surgeries in operating rooms is a complex task due to the unpredictable nature of urgent cases and the need to balance elective procedures. This research introduces a novel approach using multi-agent reinforcement learning, where each operating room acts as an 'agent' making decisions based on shared policies. The system is designed to optimize various factors like surgery throughput, delay penalties, and staff workload, and it has been shown to perform better than traditional scheduling methods. The approach also provides insights into scheduling priorities, such as prioritizing emergencies and grouping similar surgeries to minimize setup times.

Technical Details