Northeastern’s Provost and Senior Vice President for Academic Affairs is a world-renowned statistician with an academic focus on health care. We spoke to him about COVID-19, statistical modeling, public health, and the role universities can play in using big data to make big decisions. Here are 5 takeaways.
COVID-19 will change epidemic modeling forever.
“Within the world of academic statistics, epidemic modeling [used to be] a bit of a backwater. Now all of a sudden, these [researchers] are the most important people in the world. Never before have we focused so much energy and intellectual firepower on a specific modeling problem. It remains to be seen what effect this has on our use of big data and AI, but we have learned a lot.”
Universities present a huge opportunity to improve the accuracy of COVID-19 data.
When statistics are reported in nations and municipalities, “you’ll see numbers for current infections, but it’s quite misleading, because it’s amongst those who are tested. What’s missing is an unknown number of asymptomatic people who are infectious, but don’t know it. What’s going on in colleges right now is actually [shedding light on] this, because complete populations are being tested, symptomatic or not. It gives you much-better-quality data.”
For the best COVID updates, dig deep into the stats.
“I find the New York Times coverage a bit hysterical. There’s a website called Worldometers that’s terrific in terms of basic statistics on deaths and infections all over the world. Every day I look at the University of Texas COVID-19 Modeling Consortium, which shows forecasts for COVID mortality.”
Epidemiological modeling has deep historical roots.
“It really originated with a guy called John Snow, in 1854. There were major cholera outbreaks in London, and there were all kinds of theories about evil forces spreading the disease — crazy stuff. So Snow drew a map where he put a dot on the city streets for every case of cholera, and he discovered that cases were tightly clustered around a particular water pump, which turned out to be the source of the disease. This type of plotting is still the bread and butter of clinical epidemiology. [Later on], Florence Nightingale was a major pioneer in statistical visualization. She came up with extraordinary graphics to explain data that are breathtaking to this day.”
Statistician and Provost are surprisingly compatible jobs.
“A lot of my colleagues have been in and out of dean and provost positions. The provost at Georgetown is a statistician, and at Tulane. I think there are two reasons for that. One, [statistics] is a collaborative, interdisciplinary field, and that lends itself well to academic administration. Second, statistical training, at its core, is all about modeling and making decisions under uncertainty, which is central to what we do in running a university.”