Two Georgia Tech students created an easy-to-use Web application that lets health agencies in Africa monitor and protect blood supplies. It’s one of several projects from the schools’ Computing for Good course, and part of an emerging trend to give technology a human face.
It wasn't your typical computer science problem that caught the attention of two Ph.D. candidates at Georgia Tech's College of Computing. It was more of a life-or-death problem: monitoring the safety of blood supplies in African nations ravaged by HIV and AIDS.
And the fact that it caught anyone's attention is due in large part to the college's recently launched Computing for Good project, known as C4G. It's a course that encourages Georgia Tech students and faculty to look at how computer technology can be applied concretely to improve — and even save — lives.
The C4G course was sparked by a faculty presentation in the fall of 2007 by its lead advocate, Santosh Vempala, distinguished professor of computing at the college's School of Computer Science. In spring 2008, 17 Georgia Tech students signed up for the first C4G course, breaking up into teams to work on seven projects as close as the school's home in downtown Atlanta and as far as Africa. It was taught by Vempala, assistant professor Michael Best, of the School of International Affairs, School of Computer Science Chair and Prof. Ellen Zegura.
The Georgia Tech course is just one example of a growing interest in what's often called computing for a cause or socially relevant computing. As one paper, by researchers from State University of New York in Buffalo, Rice University and Microsoft Research, put it: "It presents computer science as a cutting-edge technological discipline that empowers [students] to solve problems of personal interest…as well as problems that are important to society at large…."
Keeping blood safe
The Web tool for blood safety monitoring is one of the latter. In a few weeks, it will go live in 14 African countries. The idea was to create an easy-to-use Web application that could be used by public health staff in Africa to monitor the safety of national blood supplies from collection through distribution. The problem is especially critical in nations where HIV and AIDS infections are epidemic.
The need for such an application came to Vempala's attention last year, when he talked with John Pitman, a public health advisor in with the Global AIDS Program at the Centers for Disease Control and Prevention (CDC) in Atlanta, who had created an Excel-based spreadsheet that public health staff could use to report quarterly on about 80 different data points that were indicators of blood safety.
But there were obvious problems. Staff needed to know Excel and have it loaded on a local computer. With separate applications running on PCs, aggregating and sifting data from different locations was awkward, inconsistent and time-consuming. As a result, getting local staff to use the spreadsheet consistently, or even at all, was a struggle. Pitman and Vempala thought a Web-based application could change all this.
The problem and the project appealed to two Ph.D. candidates in the C4G course, Adebola ("Bola") Osuntogun, originally from Nigeria, and Stephen Thomas, both of whom had heard Pitman's presentation. Putting their skills to practical work and seeing lasting results that would benefit someone other than researchers was a big part of the attraction. "It seemed a way to do something that was not purely academic," Thomas says. "We'd be improving the quality of life."
Vempala was the third member of the team, acting as faculty advisor. Osuntogun and Thomas quickly sketched out a design, based on the data in Pitman's original Excel spreadsheet, for a Web-hosted application, accessed via a browser over low-bandwidth connections. Initially, they assumed local computing resources would be minimal. The back-end application was based on the PHP scripting language and used MySQL as the database, with conventional database tables implementing the CDC data requirements.
On-site: eyes opened, brains re-engaged
But when the trio arrived in Zambia for a two-week field evaluation, a lot of the original ideas were scrapped or transformed. As soon as they saw the initial Web interface in action, staff with the Zambia National Blood Transfusion Service realized they could use it for much more than quarterly data reports to CDC. "They realized instantly it would be good for actually managing the blood supply through real-time data collection," Vempala says.
Regional and national NBTS directors typically start fielding phone calls late in the day from hospitals and remote blood centers that are asking for blood. To make those decisions, they have relied on a purely mental picture of the available blood supplies nationwide, of demand trends and historical data, and transport availability. The Web-monitoring tool would give them current, accurate, consistent data on blood inventories. They started asking the Georgia Tech team for more data fields, and for a flock of summary and analytical reports, including trend analysis and regional data comparisons.
That led to a new database design to make for much more flexible data analysis and reporting.
But network connections were highly problematic: some sites had none, others were dependent on low bandwidth dialup or satellite links. "One blood center had all the computing equipment but no Internet access," Osuntogun says. "Data was transferred via a flash drive handed to a visiting driver." That "connection" took a week. There was no fiber connection in Zambia to the outside world; the data was carried on satellites. "In some cases, the [network] roundtrip for a request was 5 seconds," says Thomas. "That could make downloading a Web page very painful."
But the team also found that computing facilities at all levels were fairly advanced. The most common browser was a version of Microsoft Internet Explorer as the browser. The Web application was redesigned with extensive use of advanced Ajax function calls. That meant the initial page could load quickly, and users could begin working, while the function calls pulled down additional elements in the background, to be available when needed.
The redesign: keeping it flexible
The redesign also added the ability for each office to customize selected parts of the application to meet local requirements for data collection and reporting. Some sites collected weekly data on blood collections but monthly data on distribution, for example. The new design also let users delegate different roles and responsibilities to different users in the reporting chain.
"It was an 'ah-ha' moment," recalls Thomas. "We'd been working with the CDC in Atlanta, with specific goals in mind. But the CDC weren't the people actually using the application. In Zambia, we met people who would use it day in and day out. That changed our perspective."
Osuntogun and Thomas are hoping to improve the application by adding forecasting capabilities that can factor in a range of variables, including the increased difficulty and time for blood transport during the rainy season.
On Jan. 1, ministries of health in 14 African nations, from Bostswana to Zambia, will start using the Web-based tracking application. Osuntogun and Thomas have met with officials of the United Nation's World Health Organization, after some WHO staff saw their project presentation at the American Association of Blood Banks conference in Montreal. WHO is weighing the use of the application for worldwide reporting on blood safety.