MSU – Mount Sinai Hospital – University Health Network Academic Medical Organization
Agent-Based Simulation of COVID-19 Transmission and Prevention Measures in Hemodialysis Units at University Health Network/Mt. Sinai (UHN/MSH) for Management, Rapid Training, and Education (HD-COVID-SIM)
The COVID-19 pandemic is particularly threatening to patients with end-stage Renal disease (ESRD) on intermittent hemodialysis and their care providers. Reports from Canadian healthcare facilities confirmed that maintenance hemodialysis (MHD) patients are highly susceptible to COVID-19 infections and vulnerable to their severe consequences. Many in-centre dialysis facilities experienced outbreaks among dialysis patients, physicians, and nursing staff leading to nursing and physician shortages in an already overburdened healthcare system. Therefore, It was important to understand the potential impacts of an outbreak and the pathways through which the virus could propagate in the dialysis unit environment.
Physical interaction in close proximity is one of the major contributing factors in the transmission of respiratory diseases e.g. the SARS-CoV-2 spread. Thus, understanding the characteristics of contacts such as count, duration, and distance can inform planning and policies to reduce the probability of disease transmission. Most of the available methods such as survey-based, device-based, mathematical model-based methods for estimating the contacts’ characteristics have limitations in terms of cost and accuracy. These methods are often sensitive to the period, situation, and setting of measurements and reproducing the measurement for other conditions is time and cost-consuming and in many cases risky and impossible.
In this study we used a hybrid simulation approach by developing a high fidelity multi-simulation method by combining agent-based and discrete event simulations of the Toronto General Hospital Dialysis wards to generate granular contact matrices and integrating their results into a very detailed agent-based disease transmission model for this setting. In doing so, we first developed and implemented a novel method for calculating the contact characteristics by using our high fidelity simulation which mimics the actual workflow and movements of healthcare providers and patients in time and space in 2D and 3D views. We used discrete event simulation methods to capture the movements and operations of the healthcare providers and patients in the dialysis ward and used the agent based simulation methods to account for behavior and characteristics of the agents in the dialysis setting. This allowed us for the first time to generate synthetic contact matrices for a large dialysis ward without using costly, sensitive, and risky field measurements. Subsequently, we developed another agent-based model that uses the contact matrices generated by the simulations to predict disease transmission in the dialysis unit under different public health measures scenarios such as testing, physical distancing, as well as changes in the daily schedule and operations.
This study provided very promising results for using agent-based simulation at a micro-scale to develop various contact matrices that can be used for detailed disease modeling and investigation of various mitigation measures. We aim to continue further assessment and further validate our model and expand its applications in other settings. This research has led to new avenues of research and research projects in using agent-based modeling in generating contact matrices.
Finally, we have started to develop a novel virtual reality application by creating a platform that integrates our agent-based simulations in Unity that can be used for training and decision making purposes. A team consisting of researchers of this project, several postdoctoral fellows and graduate students have been working very closely together to successfully achieve the project’s goals and objectives.
The results of this project have been presented in many different conferences and publications including: Modeling COVID-19 transmission in a hemodialysis center using simulation generated contacts matrices.
PLoS ONE 16(11). 2021: e0259970. https://doi.org/10.1371/journal.pone.0259970 Agent-based Distributed Modeling and Simulation for COVID-19 transmission in a Perioperative unit”, Submitted to Journal of Simulation. 2021.
Agent-Based Simulation of COVID-19 Transmission & Prevention Measures in HD Units at UHN/MSH for Management, Rapid Training, & Education”, Presentation at Hemo Quality Meeting, July 21.
A Distributed Simulation Approach to Integrate AnyLogic and Unity for Virtual Reality Applications: Case of COVID-19 Modeling and Training in a Dialysis Unit”, Submitted to IEEE/ACM DS-RT 2021 conference.
Generating Simulation based Contacts matrices for Disease Transmission Modeling at Special Settings”, Submitted to Journal of Simulation, http://220.127.116.11/abs/2101.10224
Risks and Complications
Primary Project Lead for Contact
Dr. Mohammad A. Shafiee
Secondary Project Lead for Contact
Dr. Ali Asgary