WPI Research is the research magazine of Worcester Polytechnic Institute. It contains news and features about graduate research in the arts and sciences, business, and engineering, along with notes about new grants, books, and faculty achievements.
Issue link: http://wpiresearch.epubxp.com/i/90696
VIDEO EXTRA Professor Radzicki explores predictive analytics in trading. " Fraud is a huge and growing problem in healthcare. By helping reduce fraud, the system can potentially play an important role in lowering costs in the healthcare space." — Michael Radzicki disciplines will analyze large data sets using The Diver Solution, Dimensional Insight's business intelligence soft- ware. Students will learn to find meaning in data and to place the insights gained into a larger context. "By housing the lab in the humanities department, we hope to address the human dynamics issues, the narrative, and not just the technical issues," he says. Faber learned the importance of using the right language to frame data-driven initiatives while he was director of analytics and new project development at Canton-Potsdam Hospital in Potsdam, N.Y. He found that if he wanted to persuade the staff to act, he needed to use the terms they were accustomed to thinking in. "While a lot of the language used at the hospital was quantitative — blood pressures, pulse rates, and so on — the staff didn't have a vocabulary for talking quantitatively about hospital operations," he recalls. "Quantifying everyday things like the infection rate and the readmissions rate seemed to motivate the staff to work toward process improvement." Faber hopes the Analytics Lab will ultimately be a nexus for undergraduate projects, graduate research, and corpo- rate-sponsored research. As a step in that direction, he and the other lab founders are working on a large-scale analysis of data related to medical visits by patients with congestive heart failure. They hope to learn the impact of preventative care on outcomes, explore which treatments seem to work best, and to even quantify how much money those measures can save. Faber is excited to bring these sorts of complex issues into the new lab. "Through this new initiative we are building the capac- ity to examine all sorts of medical and health-related data," he says. "Ideally, we hope to eventually be able to cross-link disease, utilization, and finance data to get a more detailed description of medical practice." Toward a Smarter Future In addition to shaping behavior and driving change in the present, big data can be used make predictions about the future — for example, to investigate questions such as who is most likely to contract a particular disease, where the Euro will trade a minute or a day from now, or who is planning to commit healthcare fraud. Michael Radzicki, associate professor of economics, has long been immersed in this area, called predictive analytics. He recently built it into a new WPI program in currency trading system development and is using it in an innovative new project to reduce Massachu- setts healthcare costs. Radzicki's interest in predictive analytics dates back to the early 1990s, when he and a group of math and computer science professors developed a set of trading strategies based on analyzing years of trading footage from the Chicago Board of Trade. While their strategies performed well in computerized tests, they ultimately didn't hold up because they couldn't factor in trading lag time. Trades were con- ducted through a chain of hand signals, meaning that by the time the trade was finally executed, the market had usually changed. "It wasn't until the Web and online trading that these barriers were eliminated," says Radzicki, who also started the system dynamics program at WPI. "Commissions were much lower, and though there were still latencies, they were nowhere near what they had been." [26]