A Trip to the Island

The “island” in the title is the Public Health Preparedness island that is part of the virtual world Second Life. The island is developed by the Center for the Advancement of Distance Education (CADE) at the School of Public Health at the University of Illinois at Chicago.

Our collaborators at the Montgomery County Public Health Services organized a meeting this morning for some of their partners to see a demo of the Public Health Preparedness island. CADE has built on the island a hospital, a school, an urban neighborhood, and some other facilities for users to setup, visualize, and interact with others in mass vaccination and dispensing clinics (aka PODs). (This effort follows other projects by CADE to develop games for training public health professionals.)

A group of about 12 of us were in a conference room, watching a screen while one of the Montgomery County staff entered the island with her avatar. There were five other avatars on the island with her; controlling each one was someone from CADE. Each user could speak to the others, and we could hear anyone whose avatar was close to ours. The avatars moved through the island by walking, flying, or teleporting.

In some ways, it was like a normal tour: walk around the facility, see the place, and stand around asking questions and listening to the host’s answers. But of course we were seated in a conference room in Maryland, our hosts were in Illinois, and nothing on the the whole island was real.

Using a virtual POD means that one doesn’t have to go through the logistical challenge of setting up a real one to do an exercise, and there is a lot one can do in the virtual one. The island has been used for designing PODs, since one can setup every detail, including tables and signs, and then documenting the three-dimensional layout by creating snapshots or renderings for documentation. One can create and save different designs and switch them on or off as desired. It allows interactions and so one can do role-playing to train staff on how to deal with situations that may occur.

The limitations include the learning curve to become good at navigating and interacting with the virtual world and the limited number of avatars (30) that can be on the island at any point in time, though CADE is working on programming automated players (or “bots”) that can answer questions from a user. One can use immersive interfaces such as head-mounted displays or a CAVE, though that is not common according to the folks at CADE.

Mass Casualty Assessment

The latest addition to our list of public health preparedness models is a mass casualty assessment model called EMCAPS, created by researchers at the PACER center at Johns Hopkins University. According to the developers, “it is intended to allow users to model disaster scenarios for drill planning and to use as an education resource. The EMCAPS Model allows you to estimate casualties arising from biological (Anthrax, Plague, Food Contamination), chemical (blister, nerve and toxic agents) radiological (dirty bomb) or explosive (IED) attacks. These scenarios are based on the Department of Homeland Security Planning Scenarios (April 2005).” The software also includes the CDC FluSurge Model.

To use it, download the 32 megabyte ZIP file from the project web site. (There is no cost or login required.) It takes about five minutes to download, unzip the files, and run the setup program. Running the program gives you the opportunity to select a scenario and enter important values. For instance, in the anthrax scenario, you need values for quantity of release agent, line of release distance, population density, and dissemination efficiency. Some guidance is available for understanding the variables and selecting values (for example, it includes the population density of major U.S. cities). After you click on “Compute,” the software then quickly generates a one-page report showing how many casualties to expect on each day of the scenario. You can then change the values and ask it to recalculate.

Model inputs

A user in New York asked for clarification of some of the inputs to the Clinic Planning Model Generator. Here is my response for everyone’s use.

Patient arrival batch size: Do the patients arrive in batches (because of transportation like buses)? If not, then just enter 1.

Batch size variance:What is the variance of these arrival batches? If they are always the same size, then enter 0.

Interarrival time variance: What is the variance of the interarrival times? If you do not have any data, you can enter 0 as an assumption.

Estimating cycle time

A Clinic Planning Model can estimate the cycle time of a POD (clinic) – the average time that patients spend in the POD (from the time they arrive until the time they leave). This estimate is based on a steady-state approximation of the POD; that is, it assumes that the arrival rate (in patients/hour) remains constant. The arrival rate is determined by the number of people, the number of days and hours, and the number of PODs.

What about when the arrival rate is not constant? Some periods may have large arrival rates, others smaller. The cycle time will be larger when the arrival rate is larger. One approach to handle this is to change the model inputs to model each period and determine the cycle time for each period.

For example, suppose that one expects one period of time when patients will arrive to a POD at a rate of 100 per hour and another period of time when the arrival rate will be 500 per hour. For the first period, let the “Size of the population to be treated” equal 100, with 1 day for treatment, 1 hour of operation, and 1 clinic site. The outputs will provide the cycle time for this period. For the second period, change the “Size of the population to be treated” to be 500. The outputs will provide the cycle time for this period. If you plan to change the staffing for different periods, then change these as well.

Using Excel 97

A planner from California wrote about some problems using the Clinic Planning Model Generator with Excel 97. We investigated and found that the software used a function that was not available in Excel 97. We created a specialized version with a substitute function that performs the same calculation but is compatible with Excel 97. If you need it, please let me know at jwh2@umd.edu.

Open Source and the Web

After one of my talks in Seattle, a colleague suggested that we should consider open source software instead of relying on Microsoft Excel for the Clinic Planning Model Generator. His comment was motivated by Massachusetts’ intention to use open source software. I have not heard of any other states moving in that direction.

Perhaps the point will become moot, however, for we are currently developing a web-based clinic planning model to avoid problems with Excel, especially different versions and operating systems.