>Design, Computer Science, Human Science

3D Visualization and Simulation of Infectious Disease Spread

Project Summary

This project was designed to look at how large detailed simulations conducted to support federal policy and preparedness could be better communicated and quality checked through the use of Virtual Reality. The goal was to take the underlying population data which is synthesized from myriad datasets and the results of our simulations and create a compelling visualization that allows the dynamics of the simulation to be appreciated in both its detail and scale, while allowing the user to validate the verify the behavior of the simulation embedded in this data. The test case was originally intended to be infectious disease, however, we pivoted to a catastrophic disaster due to a pressing need for the quality assurance part of the project for a follow-on study. Specifically, this disaster was an improvised nuclear attack in downtown Washington DC. More details of this study are reported in the literature (1,2) , in brief we model the behavior of over 700K individual agents in the affected area. They are imbued with full agency, choosing where to go, what to do, and when to do it, based on local information about their health, family status, and others around them. They communicate and move around the DC area while being impacted the condition of their surroundings and are exposed to radioactive fallout. The resulting interface allows the user to see the health state and movement patterns of the individuals and provide a very useful means to verify and appreciate the complexity of this simulation.

Project Outcomes

This project successfully created the intended interface, and is under current revision and iterative improvement. The interface was presented during ICAT day and was demonstrated to dozens of individuals attracting a fair bit of feedback and opportunity for education. Additionally, several other scientists who helped design and implement the simulation being visualized have also tested out the interface. These scientists were greatly impressed and found the visualization very helpful for conceptualizing and testing theories about certain behaviors of the simulation that very challenging to tease out from the raw output and other visualizations developed to date. For instance when particular agents engage in an “Aid and Assist” behavior and move towards ground zero, several of them paused for longer than anticipated while assisting others, this led

3D Visualization and Simulation of Infectious Disease Spread

In the image above, we see the general tools in the foreground for manipulating the timestep of the simulation. The yellow glow on the horizon is ground zero. We also see the map on the floor “darkened” representing that this during non-daylight hours in the simulation, which is one of the cues that can change the agent’s behaviors.

3D Visualization and Simulation of Infectious Disease Spread

In this screen shot, we see many red X’s, representing agents who are deceased, in the vicinity of ground zero. The green triangles represent agents that are on the move, and the hexagons represent agents that are momentarily paused for several timesteps. This glyphs are sized based on the number of agents it represents (captures the scale while not having 300 glyphs all right on top of each other). Additionally, one can note that some of these glyphs are green (full health), orange (injury or degraded health), or red (deceased). Capturing this level of information density while maintaining a “larger” overview was a primary design goal.

3D Visualization and Simulation of Infectious Disease Spread

This screenshot shows the legend which is anchored just north of where all the agents are in Washington DC and thus is accessible to the user at any given time to aid in interpreting what they are viewing. Additionally, the time step, time of day, and time since detonation are also displayed which (with the aid of the time of day lighting) allows the user to stay oriented along the timescale.

3D Visualization and Simulation of Infectious Disease Spread

This screenshot proves a good perspective on the from ground zero to the evacuation points which are place well outside the population of intense analysis. How, when, and who moves to these evacuation routes is a primary outcome of these studies, as we explore what policies and increase survivability and minimize the total time spent in the radioactive disaster zone in the heart of DC.

3D Visualization and Simulation of Infectious Disease Spread

 

 

This screenshot shows an initial prototype for a “controller” linked display. The colored lines shows the counts of individuals in the various states of mobility. The yellow corresponds to the temporarily stopped individuals and one can see a large number of them slowly decline over time as more and more people move and
evacuate. The blue line are those who are moving, the red are the deceased which grows quickly after the blast and then more gradually in the hours following. The green line are those who are permanently stopped and they gradually decline as many of them are grievously injured and eventually succumb to their injuries and die.Overall, this project was a success and has generated a tool that will be quite useful as we continue to conduct experiments using this simulation platform. The current implementation allows reasonably rapid extraction of the key data from the simulation results and packages it into a format readable by the Unity engine and creates the environment as illustrated above.

Submissions for publication: None yet

Presentations:

Exhibitions:

ICAT Day: NPS1 VR Visualization

Student Involvement:

Several students were key members of the team. Van Truong, was a key facilitator and project manager. She organized and wrote large parts of the original grant, interviewed other student workers, organized meetings, helped in the design of the interface, and aided in data organization Vineeth Edam, was hired to provide programming support in Unity for parsing and ingesting the large data files produced by the simulations. He successfully completed his tasks and his code formed and important foundation for subsequent extensions. James Schlitt, joined the project in its later stages and provided the additional programming skills and creative input to properly interpret and use the structure of the data in a compelling visual manner. He will continue to build on this and will use this project as a chapter in his PhD dissertation.

Educational components (K-12):

During ICAT day we interacted with many K-12 students, however, there were no explicit educational goals aimed at this age group.

Supplemental resources used to complete the project:

Computer and GPU to house and drive the VR headset were obtained through other funding mechanisms.

Media coverage:

Coverage surrounding the project, that this project supported:

http://www.sciencemag.org/news/2018/04/what-if-nuke-goes-washington-dc-simulations-artificial-societies-help-
planners-cope

A biocomplexity press release:

https://www.bi.vt.edu/ndssl/news/from-a-bombs-to-zika-guiding-government-planning-for-major-disasters

Anticipated external funding which may result from this project:

A preliminary proposal to the US Dept of Health and Human services, BARDA was submitted and we await notification of an invitation for full proposal. Proposed award: $850K per year, for 3 years.

References:

1. Lewis B, Swarup S, Bisset K, Eubank S, Marathe M, Barrett C. A simulation environment for the
dynamic evaluation of disaster preparedness policies and interventions. J Public Health Manag Pract. 2013
Sep;19 Suppl 5:S42–8. PMCID: PMC3962069
2. Swarup S, Stretz P, Parikh N, Rivers C, et al. Modeling human behavior in the aftermath of a
hypothetical improvised nuclear detonation. The 12th International Conference on Autonomous Agents and
Multiagent Systems (AA- MAS 2013)
2013.

Members:
Bryan Lewis, Biocomplexity Institute, NDSSL
Zach Duer, School of Visual Arts
Van Truong, Biocomplexity Institute, NDSSL

Collaborative Colleges:
Biocomplexity Institute

Areas:
Design
Computer Science
Human Science