For years, Louisville, Ky., has struggled with poor air quality and a preponderance of asthma cases among city residents. A year ago, the Louisville Metro Government tried once and for all to figure things out—and make a significant difference in the lives of many.
Their solution for this problem: To turn over what they knew to consultants from IBM as part of the company’s Smarter Cities Challenge grant program. The consultants would come in, get a sense of the health and other data sources available and make recommendations on how the city could better utilize that data and create new resources for the city to reduce asthma rates.
The challenges facing Louisville may have a specific public sector focus, but they illustrate some themes common to big data analytics projects: Get all data sources working together. Design a governance model for the data and processes for people to follow. Establish communications channels to use insights from the data. Set up a way to learn from experience so that project managers can apply lessons to improve the system.
In the case of this project, there was another condition for the consultants: Assess the situation and deliver recommendations in about three weeks.
Going Beyond Public Surveys
When the project began early last year, according to Ted Smith, the city’s chief of economic growth and innovation, the state of the city’s asthma data at that time didn’t amount to very much at all; just some facts and figures from surveys. In 2009, for example, the city reported that asthma was fourth among the top 10 reasons citizens under 20 years old were hospitalized. Fifteen percent of the city’s adults reported they had asthma (above the national average of 13.5 percent).
Such statistics did not yield the kind of insight local officials sought. “Every two years, we got some survey data that said asthma gets worse on certain hot days,” remembered Smith. “The fact that we actually had this information was great, but the information itself was embarrassing. It was almost as if the data quite frankly was a distraction.”
And so Smith and his colleagues applied for and won the grant. IBM’s Smarter Cities initiative is a $50 million philanthropic project that eventually will serve 100 cities around the world over three years. Local authorities get services, IBM gets attention for work it often does in private, and the company’s employees compete vigorously to ply their trades in a philanthropic endeavor.
Investigating Data Sources
The IBM team arrived in July 2012 and spent three weeks working on the project. Sondra Renly, an IBM research scientist whose expertise is in health care systems interoperability, was one of five in the bunch.
“Ours was the first Smarter Cities project to deal with a health care question, so immediately we were up against new challenges,” Renly said. “Health care uses so many specific sources of information and formats and expectations. It’s a struggle just to keep everything straight.”
Because data sources were so disparate and difficult to collect, the team spent its first few days interviewing Mayor Mike Fischer and other city officials about what sorts of data were where. The IBM team ruled out PDFs and Word documents, and opted instead for raw data. Through this initial investigation, they identified silos of data in five key areas:
- Propeller Health, a Madison, Wis.-based company that makes a system to monitor asthma patients. At the time the IBM team arrived, this company was operating a Federal Drug Administration-approved and GPS-enabled device that sat atop inhalers and recorded every time asthma patients used the devices for relief.
- The Jefferson County government, which was monitoring about a dozen air quality sensors and keeping data on them.
- Health care providers, including hospitals, private physicians, nurse services and more.
- Private sources, which were monitoring and compiling air pollen and mold counts.
- The Environmental Protection Agency.
Renly and her colleagues didn’t stop here, notching nearly 70 additional interviews in the weeks that followed. An interview with the regulations department turned up information about controlled burns in rural areas, so the IBMers sought data from local fire departments in the region. They also sought permit data from the city government; for example, a week-long event tied to the annual Kentucky Derby horse race featured a huge fireworks display.
Finally, the IBM team requested data from the Jefferson County School District about absenteeism rates throughout the school year, and for absentee data from the government and health care providers, seeing if they could establish a link between sick kids and the parents who stay home to care for them.
“We know that when children are out there is a caretaker issue and we wanted to link these two together,” Renly said.
Making Sense of the Data
After nearly four weeks of research, the IBM team presented the data it collected to the 70 stakeholders whom they had interviewed. The team then returned to their respective offices and collaborated on a list of 13 recommendations specific to the situation in Louisville.
Though Louisville Metro hasn’t made the report public, Renly said these recommendations spanned the gamut from keeping citizens engaged in the fight to preserve cleaner air to sharing data about air quality with constituents and deploying more sensors to collect air quality data. Another big one: Collaborating with the school district. Specifically, one of the recommendations was to mine public schools for asthma information, work more closely with school districts, and share the information with constituents as a whole. None of this was data Louisville had used to inform public health research in the past.
For example, the city never thought to link school attendance with asthma, Smith said, adding that Louisville officials recognized the benefit of addressing this concern. “If we can get attendance up by controlling asthma, everyone wins,” he said.
Using Data to Inform Policy
Louisville plans to heed at least some of the recommendations and use data to drive policy changes across the board.
First, Smith said officials planned to rethink the way the school district accesses data it already has, and noted that the district already is looking into making more data available for download by the general public.
Second, Louisville leaders were planning to embrace the mash-up, a process by which they planned to overlap asthma data from a number of sources to single out the most egregious surface traffic and stationary pollution issues. This process was supposed to give city leaders a broader and more well-informed perspective.
“If one data set indicates that one particular freeway is causing pollution, the city could close the road,” said Smith, noting that the City of Atlanta did something similar during the Olympic Games there in 2000.
Finally, the Louisville leadership was in the process of planning an exhaustive search for more data. To accomplish this, the group was soliciting donations from local companies to defray the cost of another 500 air quality sensors—about $150,000 in all.
Down the road, Smith can envision 3,000 sensors up all over the Louisville Metro Area. Until then, however, he and his colleagues will have to settle for a new approach to data-driven decision making, and the satisfaction of a new process that involves the entire community.
Home page photo of Louisville skyline seen from the Ohio River by Anindya Chakraborty via Wikipedia.