I like that Insights kind of takes the best things of all of these different parts, from dashboards to widgets, and allows you to have additional capabilities that you just don't get anywhere else. I think the need for Insights was at the right time for us. It did exactly what we needed it to do.
Division of US Army Forecasts COVID-19 Pandemic Peaks to Distribute Resources with Analysis Software
The United States Army Corps of Engineers (USACE) is the permanent engineer branch of the United States Army that has approximately 37,000 dedicated civilians and soldiers delivering engineering services to customers in more than 130 countries. With environmental sustainability as a guiding principle, USACE districts around the US aim to "build strong" and strengthen the nation by constructing and managing infrastructure, restoring the environment, and maintaining waterways.
The USACE Galveston District in Galveston, Texas, executes the mission of USACE along the Texas coast by providing vital public engineering services to energize the economy and assist in emergency management. To protect Texas residents during the devastating COVID-19 pandemic, the district was presented a unique challenge in 2020: collect and analyze data to help forecast peaks in the pandemic to efficiently distribute resources and establish alternate care facilities if hospitals were filled to capacity.
A small team was formed at the USACE Galveston District to produce data models and provide key information to senior leadership to help inform decisions. Initially, the dashboard created was challenging to understand with the inclusion of many different data models, and it also didn't provide details at a glance. The model and analysis team then developed a new solution using software that enables more iterative and exploratory data analysis, which has yielded an array of benefits, including improved analysis, more efficient workflows, and better usability.
The USACE Galveston District established the COVID-19 model and analysis team to examine nation- and statewide COVID-19 data and analyze trends to inform decision-making. The team, composed of four members, gathered data from several prominent data models, including COVID-19 Hospital Impact Model for Epidemics (CHIME), Institute for Health Metrics and Evaluation (IHME), Columbia University Model, and the Center for Army Analysis, and created data models to evaluate specific patterns in data. For each model, the team analyzed data for each state and county in the US to determine where USACE resources and personnel were needed at present or in the future.
Leadership was interested in when the peak dates for the COVID-19 pandemic were going to happen to better assess the availability of statewide resources and when resources will be needed, so the team specifically examined hospital bed utilization, ventilators, and intensive care unit (ICU) beds in Texas. This data was also important in determining the need for alternate care facilities (ACFs), which are sites where patients with COVID-19 can receive medical care. Predicting any peaks would help the team know ahead of time when hospitals may be at capacity.
The existing method involved the use of a digital dashboard that displayed all five national data models separately, making it difficult for senior leadership to decipher the data and also caused inefficiencies in analysis and reporting. The dashboard would simply show an average of all peak dates. Jason Jordan, a senior GIS analyst with the Galveston District, explains that having so many outputs to look at was difficult for leadership.
"[The digital dashboard] was effective but so clunky when actually loading the data daily. Every day it was just a whole bunch of summary statistics and joins over and over again. Then some queries to get out all the peaks. The workflow was down, [and] it just took forever to do each day," Jordan says.
Members of the model and analysis team, led by Galveston District Geospatial Branch Chief Rick Vera, wanted a solution that would allow them to combine all the data into one dashboard-type tool for better readability and understanding. They also wanted to geo-enable the static graphs and spreadsheets they were presenting to gain usable location intelligence.
"The models that senior leaders and commanders were used to seeing were all bar graphs or some type of line chart and spreadsheets with numbers and percentages. They didn't tie it back to geography. We needed [a better way] to visualize what was actually happening around the nation," says Vera.
To find the best software to help the Galveston District create a new, streamlined model to display data, Vera says he began looking at the tools available from Esri. A previous USACE employee had done some work with ArcGIS Insights and showed Vera how it worked, and he thought it could be beneficial for this project. ArcGIS Insights is a data analytics workbench that allows users to perform iterative and exploratory analysis with data from a variety of sources, including spreadsheets, and then create interactive reports.
"I'm just glad we have access to just about every product Esri has to offer and that we can actually look at different tools and pick the right one," says Vera. "There were a lot of new workflows that we could take advantage of with Insights that, of course, [our dashboard tool] doesn't have. So that's why it made the more ideal platform to put the five models together."
In addition to the workflows, Vera thought Insights could help them automate the process of gathering and updating information. According to Martha Bass, an Esri technical consultant, "They didn't have hours a day to spend updating this manually. They had their day jobs to do. So, the idea was to script it out so it would be easier for them to do updates."
Vera and the team went to Leslie Weiner-Leandro, a defense account manager at Esri, and Debbie Murphy, their advisor through the Esri Advantage Program, for assistance. They began working to transfer their existing dashboard data into Insights, and they also enlisted the help of Bass to develop and configure the tool. The team wanted the data displayed in a single feature layer.
Bass says her role as a technical consultant was to figure out how to display [their metrics] in a meaningful way and make sure we're showing them the details they want to see. "They are a good example of people who are really pushing the boundaries of trying to display a lot of complex data all at one time and create meaningful visualizations,” she says.
Vera explains that Bass set up Insights for them and provided an ArcGIS Notebook in ArcGIS Pro to help update data automatically. "All we have to do every day is download our datasets from our different sources, dump them in a folder, run the notebook, and it would basically do the entire workflow for us," he says.
The end product is an Insights dashboard with several pages that show relevant data, including the peaks on the curves for all the different models for inpatient beds, ICU beds, and ventilators at the state and county levels. It also has capacity predictions at the state and county levels, and each model's projections can be visualized separately or together.
Vera and his team have been using the new model since April 2020, and the convenience of having national-, state-, and county-level data in one view in Insights has yielded positive results for the Galveston District, including the availability of more real-time data at a glance.
"[The data] is 10 hours younger than the dashboard would be. As soon as the datasets come in, it goes online in 10 minutes rather than later that day," Jordan says.
The streamlined view has also helped senior leadership better understand the data. The previous method did not easily allow leadership to engage with data as the display was basically a serial chart with only one place to see the average of the models' peaks. Now, Jordan says Insights has all the views in one and all three types of visualization.
"Insights still has the tabular part on there, and they can switch between that, the timeline of graphs, or the time-enabled maps, depending on what provides the information they need," says Jordan. "Everything is filterable if they want to just take a look at one model or one area."
Vera adds, "Colonel[David] Hibner [commander of the Army Geospatial Center] kind of took the new Insights dashboard and ran with it. And I think it put [the data] in perspective for most senior leaders. In the end, it's a simple visualization."
According to Vera, the use of Insights has also made their workflows of gathering and analyzing complex data more efficient, saving the team valuable time. He says the modeling team worked on COVID-19 related tasks 12 hours a day, 7 days a week from April through September 2020, and the costs associated with staff doing tasks like gathering data were considerable.
"I would say it probably saved a good month of time at least . . . over that 6-month period. The [time this saved] allowed us to focus on other things. It allowed us to go fight another fight. That's valuable time to us that we could be spending looking at additional datasets and bringing other things online,” says Vera.
Jordan adds, "[With] all that time that was spent just updating products every day, we could take that and focus it on producing new products or exploring other types of analyses to perform or break out."
The automation aspect of the new Insights dashboard has also increased the efficiency of the team's workflows. With the ArcGIS Notebook, the on-the-fly processes would update, and data would be up-to-date in about 10 minutes as opposed to several hours with the previous digital dashboard, according to Vera.
"I stress automation. I think that if we can free up time to go do the next thing, I'm all for it because being the geospatial branch chief, [I take into consideration the] time, money, and effort of my staff,” says Vera. "The on-the-fly joins were super helpful for us just because we could go in and swap those things out every single day and have it and run it, and it'd be up-to-date."
Jordan echoes this and says, "I'm a huge fan of anything that slices and dices data, visualizes well, and is very hands off. I'm not interested in typing in numbers or plugging a spreadsheet into something every day. So, Insights is definitely right up that alley.” He says the team simply runs a Python script and all the graphs, charts, maps, and tables are instantly updated.
Jordan says the new Insights display better enables a deep dive into datasets. He says that instead of datasets just being broken up at the surface and put into a single feature layer only, team members could still have all the original data that they could dive into as well instead of just the summarized version.
"Putting the data in Insights and deep diving into it, being able to click on an individual line, a peak line, for any of the data models [made it easy for leadership to understand],” he says. "There really wasn't an authoritative model that was giving us a peak that we can understand and make decisions by. And I think that's what Insights brought to the game because the [models] were so spread out."
From working with account managers to technical consulting, the relationship with Esri was a significant part of this undertaking by the Galveston District, and they look forward to continuing this collaboration.
"The most exciting work that I get to do as a technical consultant is working with customers who are at the cutting edge and really pushing the boundaries of what the software can do and thinking creatively," says Bass.
"It's a marvelous partnership. From my perspective, it's probably my most favorite thing that I get to do as an account manager is that sort of creative, intellectual collaboration with customers who are willing to consider the art of the possible," Weiner-Leandro adds.