Objective:

To propose a digital twin model that analyzes patient arrivals, treatment procedures, and resource allocation using data-driven techniques, specifically predictive analytics.

Approach:

• Framework Development: A multi-level AI-enhanced digital twin framework is developed, incorporating temporal forecasting, clinical robustness, and outcome grounding.
• Data Utilization: The framework utilizes three real-life datasets to analyze patient flow and clinical scheduling.
• Modeling Techniques: An LSTM-based model is employed for temporal intelligence, achieving competitive prediction performance.

Key Findings:

• The temporal intelligence level achieved R² = 0.6785, MAE = 0.0895, RMSE = 0.1103.
• Maximum congestion in emergency departments occurs between 11:00 and 14:00, primarily due to moderate-acuity patients (ESI-3).
• The mean waiting time for patients is approximately 35 minutes, with a negative association between waiting time and patient satisfaction.

Interpretation:

The developed framework reflects representative features of hospital dynamics and emphasizes the importance of AI-enabled predictions in decision-making.

Limitations:

• The study does not implement a functioning system in real-time.
• The analysis is based on historical data and may not account for future changes in hospital operations.

Conclusion:

The proposed digital twin model provides a systematic approach to understanding and optimizing patient flow and scheduling in healthcare systems.