Vertically Integrated Project

Agent Based Simulation: Spatio-Temporal Modeling of COVID-19

Professor- Dr.Paola Gomez, Dr. Mathew Swartz

Partners & Sponsors: Georgia Tech Research Institute, Perkins and Will.

Term- Fall 2020

Team- Ofek Oshri, Charles Luo, Muskaan Vaishnav, Prerana Kamat,

The built environment plays a fundamental role in hour daily life. It impacts our health, performance, stress levels, social relations, and even enables the spread of the COVID-19 disease. This project explores expanding the notion of Building Performance Analytics towards a dynamic spatiotemporal framework, studying the characteristics of spaces combined with human activities and organizational processes that influence everyone’s lives.

This semester’s specific goal was to explore the spatiotemporal dynamics of the COVID-19 pandemic and its spread through the built environment. The dynamics of the spread have been modeled from a virus-centric perspective primarily at the regional and global. The current models do not incorporate spatial variables beyond the social distance. Our team worked on agent-based simulations, process simulations of the virus spread.

• Problem: Understanding human behavior in order to model the spread of the COVID-19 virus within varied settings

• Goal: Be able to have an accurate model with varying parameters that depicts the spread of the COVID-19 virus

• Parameters: Types of built environments, PPE usage, Social distancing policies

• Challenges: Model architecture, parameter reliability

Research

• Hospital demographic groups

•Nurses, patients, doctors, visitors

• Systematic actions for doctor/nurse movement

• Rates of transmission for Covid under various circumstances

• Hospital procedural changes due to the pandemic

Final Presenatation

Final

Report

Variables

• Number of patients

• Nurse-to-patient ratio

• Doctor-to-patient ratio

• Patient to bed/patient to room ratio

• Movement of doctors and nurses

• Rate of Transmission

• PPE Usage

• Visitors

Spatial Consideration

Reference: Queen Mary Hospital, Hong Kong by HKS

Key Design Elements:

• Departmental compartmentalization

• Dedicated patient transfer elevators

• Emergency observation unit conversion to pandemic isolation floors

Normal Flow

• Dedicated staff entrance and corridors

• All visitors and incoming patients- main entrance- waiting area-primary care-wards

• Admitted Patients-wards, critical care, dedicated elevator.

Pandemic Flow

• The main entrance acts as a thermal checkpoint to segregate incoming patients

• The wards are changed to isolation rooms and will only allow patient and staff movement

Agent Behavior

Doctors and Nurses

● Systematic

• Begin and end each shift in the workroom and through the entrance/exit

• Treat patient after patient almost entirety of shift

• Work in shifts

● Grouped

• Many nurses travel together with doctors assisting them, and leaving for the workroom occasionally*

• Visit frequency can often be indicative of a patient’s symptoms severity*

Patients

● Static

• All patient’s bodies react differently to covid-19

• Symptoms are usually debilitating, leaving patients in bed

Evaluation Criteria

• We evaluated the model based on the number of nurses with COVID during the simulation, proportional to the number of nurses in the hospital at the time

• Nurses have a chance to enter infectiously or become infectious from interaction with other nurses and patients.

• Judging effectiveness of pandemic flow by maximum % infected

Results

• Based on the paths of agent movement it is evident that similar paths lead to increased cases

• This suggests that identifying healthcare workers with COVID is incredibly important

• Returning to a specific room to maintain records or don protective equipment also causes exposure risks

• May have had some issues with model resolution causing exposure to be higher than reality

Challenges

Research challenge: COVID has been around for about a year

• Finding specific consistent information about COVID has been difficult especially in the form of studies

• This is most apparent in rates of transmission that was difficult to nail down

• It is an ever-changing situation as well - thus visitor policies and new measures are always being created

NetLogo challenges:

• Pathfinding was challenging to implement, requiring sacrifices in resolution and accuracy

• Issues with floor plan importing, with color and walls

• Differences between NetLogo and traditional programming languages introduced additional difficulty

Back