Optimizing MRI Scheduling in High-Complexity Hospitals: A Digital Twin and Reinforcement Learning Approach

Area of Science:

• Healthcare Operations Research
• Artificial Intelligence in Medicine
• Medical Imaging Informatics

Background:

• High-complexity hospitals face MRI operational inefficiencies: low machine utilization, long patient waits, and unfair priority access.
• Intelligent scheduling is crucial for dynamic waitlist management based on clinical urgency and resource optimization.

Purpose of the Study:

• To develop and evaluate a novel framework integrating a digital twin (DT) with a reinforcement learning (RL) agent for optimizing MRI operational scheduling.
• To enhance MRI machine utilization, reduce patient waiting times, and ensure equitable service delivery based on clinical priority.

Main Methods:

• A digital twin simulated MRI operational dynamics using real-world parameters (patient arrivals, exam durations, machine reliability, priority levels).
• A reinforcement learning agent, trained via Deep Q-Networks (DQN), learned scheduling policies to optimize key performance indicators.
• The framework was evaluated against traditional scheduling heuristics like First-Come-First-Served (FCFS).

Main Results:

• Achieved a 14.5% increase in MRI machine utilization.
• Reduced average patient waiting times by 44.8%.
• Demonstrated significant improvements in priority-weighted fairness compared to baseline methods.

Conclusions:

• The proposed DT-RL framework offers a scalable and adaptable solution for optimizing MRI services in complex hospital settings.
• This approach enhances patient satisfaction and clinical outcomes by improving operational efficiency and fairness.
• The study highlights the potential of advanced AI and simulation techniques in transforming healthcare operations.