EURO 2025 Leeds
Abstract Submission

2857. Optimizing ICU Operations with Digital Twins and Machine Learning: Enhancing Decision-Making and Resource Management

Invited abstract in session TD-34: Advancements of OR-analytics in statistics, machine learning and data science 5, stream Advancements of OR-analytics in statistics, machine learning and data science.

Tuesday, 14:30-16:00
Room: Michael Sadler LG10

Authors (first author is the speaker)

1. Margarida Parrado
Department of Mechanical Engineering, NOVA School of Science and Technology
2. Luís Lapão
Intelligent Decision Support Systems Laboratory, Universidade Nova de Lisboa

Abstract

Backgroud: Digital Twins (DTs), virtual representations that replicate real-world entities, are emerging as transformative tools in healthcare operations, particularly in Intensive Care Units (ICUs). ICUs are high-stakes environments requiring continuous patient monitoring and advanced medical interventions. However, operational inefficiencies and data overload often delay decision-making, affecting patient outcomes and resource utilization.
Methods: This study developed a DT platform, in collaboration with ICU director from Francisco Xavier Hospital in Lisbon, integrating Machine Learning (ML) to enable decision-making. The model incorporates real-time patient and operational data to predict sepsis.
Results: The DT model demonstrated 78.8% accuracy in predicting Sepsis onset, supporting early detection and timely intervention. Furthermore, the system successfully forecasted systolic arterial blood pressure (ABP) and heart rate (HR) responses to vasopressor administration, achieving a mean absolute error (MAE) of 4.9 mmHg for ABP and 4.1 bpm for HR. These insights enable proactive treatment adjustments, improving patient outcomes and care efficiency.
Conclusion: Merging DTs with ML enables a data-driven approach to optimizing ICU workflows and reducing medical errors. Beyond ICUs, this technology holds promise for operational research, providing intelligent, real-time decision support for optimizing critical care processes, resource management and patient care.

Keywords

Status: accepted

Back to the list of papers