PROJECT: 3D Analysis of AED Provision

Analysing Access to AEDs in NUS UTown

By Terh Shin Huoy, Teo Rui Xiang, Guo Bingkun & Xu Yuting

As part of the course GE3238: GIS Design and Practices, I was required to work in a group on a GIS topic of our choice. Through this project, I learnt about GIS project management, geodatabase management, 3D network building and analysis and 3D modelling. We utilized ArcMap and ArcScene for this project. This page contains part of our final report that we produced at the end of our project.

Abstract

An automated external defibrillator (AED) is a portable device that is used to treat sudden cardiac arrests. The American Heart Association (2001) recommends a response time of 3 minutes from the collapse of a patient to the arrival of AEDs to ensure greater likelihood of victim survivability. With objectives of assessing the adequacy of AED provision and enabling indoor navigation to the nearest AED, this project focuses on the publicly-accessible areas in NUS University Town. A 3D network and model has been created using ArcMap and ArcScene. The network analysis functions of Nearest Facility and Service Area has also been created using ModelBuilder. Using the Nearest Facility tool created, we are able to generate responsive outputs, including visualization of route to the nearest AED, textual directions to the nearest AED, distance covered and time taken) according to user input. Using the Service Area tool created, we conclude that all publicly-accessible areas in UTown within the spatial extent of our project discussion are within 90 seconds of an AED. Hence, we conclude that the current AED provision in UTown is sufficient.

Methodology

1. Data collection

We obtained the plans of UTown from the UTown Management Office, and conducted our own recce to collect the most accurate data.

2. Digitization of basemap

We then digitized the floor plans using ArcMap. The rooms were represented as points, the corridors and paths were represented by lines and the buildings and open spaces were represented by polygons.

3. Topology checks

In order to ensure that the network analysis could be conducted accurately, topology checks were made. Importantly, we checked that the points (rooms) were joined to the end of the lines (corridors) using the 'must be covered by end point of' rule, and that the lines were joined correctly using the 'must not have dangles' rule.

Database Management

The figure below shows the schema for our geodatabase.

Our geodatabase contains the following three feature datasets. Display and Rooms contain polygon features that represent areas occupied by each building and room of interest respectively. Feature classes under Rooms Dataset are arranged by floors (named as RoomFirst, RoomSecond and so on). These two feature datasets are created for better visualisation and display purposes, especially in ArcScene. They are not directly involved in network analysis.

Another feature dataset, Network, contains all feature classes that are directly involved in network analysis.

Network Analysis

ArcScene was used as the visualization platform for our project. However, unlike ArcMap, ArcScene does not have a network analysis toolbar to streamline network analysis functions. Therefore, models were built using the ModelBuilder to replicate the ‘nearest facility’ and ‘service area’ functions in ArcScene. These models basically allow multiple-step processes to be streamlined into a simple tool.

The figure above shows the ‘nearest facility’ model that we built. The first tool is the ‘Make a closest facility layer’ under Network Analysis Tools, which enables the ‘nearest facility’ function. The first ‘Add location’ tool (in Network analyst tools) allows the addition of ‘AEDs’ as the ‘facility’ sub-layer, while the second ‘Add location’ tool allows the input of ‘feature set’ as ‘incidents’ sub-layer. The ‘feature set’ was created as a variable, with its schema imported from a new feature class, UserInput, and was set as a model parameter. Finally, the ‘Solve’ tool prompts the model to solve the network and produce the solution. As an additional function, after the network is solved, a tool named ‘Directions (Network Analyst)’ was added to generate the precise, step-by-step textual directions to the nearest AED from the incident point.

The ‘Service Area’ tool in Network Analyst allows for the evaluation of accessibility. Concentric service areas show how accessibility varies with impedance. In our project, this tool was used to determine the area in UTown that has access to an AED within a certain impedance (travel time).

Data Visualization in ArcScene

After the maps were digitized and the network was created in ArcMap, the point, line and polygon feature classes were imported into ArcScene. ArcScene is well-suited for 3D visualization and also supports 3D network analysis functions. The polygon feature classes of the rooms on different levels are imported into ArcScene and extruded according to their heights. The extrusion method was specified as ‘Adding to a feature’s base height’. Different levels are displayed in different colors, and the transparency was lowered so as to allow visualization through the walls.

We created 3D models of the UTown buildings using Sketchup and inported them into ArcScene. The building models serve visualisation purpose rather than network analysis. In addition, the detailed 3D models of the buildings help in identifying the buildings and establishing the correct orientation. Without the realistic-looking building models, the 3D polygons of the buildings are relatively unidentifiable, and may be confusing to the users.

Results and Discussion

1. Nearest AED from incident point

Our project has built a network that allows the analysis of ‘nearest facility’, more specifically, to determine the nearest AED from any incident point in UTown. For example, if an incident of sudden cardiac arrest occurs at Seminar Room 2 at ERC, the nearest AED solved using our model would be the AED outside the Security Office at ERC, as shown in the route generated (Figure 25). The directions are generated in a separate text file.

This network analysis tool can be applied at the time of the emergency, or for emergency planning.

This analysis is crucial in terms of emergency response, as the knowledge of the nearest AED and the shortest route there would be critical in determining the survivability of the victim. In an emergency, a clear set of instructions to the nearest AED is important as there is likely to be panic and confusion. Many people also do not know where the AEDs are located, and manually searching for an AED would waste precious time. With a clear set of directions to the nearest AED, there will be an increased likelihood of obtaining an AED in time to save the victim.

This model also has applications in non-emergency situations. It could be used in events planning and crisis management planning. The organizers of school events are required to submit crises management plans for different crisis scenarios to the Office of Safety, Health and Environment (OSHE). Most of such crises management plans have fire evacuation procedures and the nearest fire exits, as such information is widely accessible. However, little or no emphasis is placed on where the nearest AEDs are. With our model, event planners can easily determine the nearest AED facilities near their event location, and brief their participants and prepare contingency measures accordingly. This would definitely help in improving the safety management systems and procedures in NUS

2. AED Service Area

The AED service area tool allows us to view the coverage of AED facility in UTown campus. This allows for the determination of whether there is sufficient provision of AEDs in UTown. It also allows for the proposing of the best location to implement a new AED to ensure the most cost-effective coverage of AED provision for the campus.

Our analysis shows the service area with AEDs within 90 seconds. In other words, each point in UTown is a maximum of 90 seconds away from the nearest AED. Given that sources show that a response time of 3 minutes from the collapse of a patient to the arrival of AEDs is recommended, 90 seconds for a travel time from incident point to nearest AED is a suitable estimate for the adequacy of AED provision, to allow for the total to and fro travel time to be a maximum of 180 seconds. Therefore, the results of our analysis show that there is sufficient AED coverage in UTown, and there is no need for additional AEDs to be installed.

Do contact me if you wish to have further details or the full report of our project!