By the Numbers

By Sharon Proctor, PhD
Digital illustration Michael Gioffredi

Statistics showing polling results, TV ratings, hockey scores, unemployment rates, batting averages, housing starts, hurricane casualties — we’re inundated with them. Such statistics help us make sense of our world. They help us compare and evaluate things. And today, with modern computers, researchers can create powerful models from statistical data: models that reveal underlying relationships, patterns, and causes that were previously invisible. One of those researchers is Charmaine Dean, a professor in SFU’s Department of Statistics and Actuarial Science. She is working to improve the techniques used to model statistical data.

A “model” is a mathematical representation of a phenomenon. It consists of linked equations that describe the behaviour of the phenomenon and those things that influence it. Dean is especially interested in models that extend over space and time. For example, she and her students have been modelling disease and death rates across Canada, based on data collected over several years. Each model they create is a pioneering effort. No one has done it before.

How it all started

Dean was born and raised in Trinidad, where her father was Deputy Minister of Education. (“He was a workaholic and I’m tending to follow in his path!”) Her homemaker mother cared for eight children, was deeply interested in what her kids learned in school, and eventually became a post mistress. Young Charmaine also had an engineer uncle who lived with them. “Education was important in our family,” she recalls. “And I was always interested in mathematics. For me it was a path of great intrigue. My uncle was an early mentor. In Trinidad at the time, schools were changing to the ‘new math’ teaching method, and Uncle Belix wanted to learn about [it]. So he studied with me every day.”

How did she end up at SFU? “My brother, who studied biology at the University of Guelph, was hired by the Saskatchewan government and worked out of Saskatoon. After I finished high school, he offered me free room and board if I enrolled at the University of Saskatchewan.” There she got an honours degree in mathematics (1980), then returned to Trinidad to teach. Meanwhile, she’d married a young man from Saskatchewan. Two years later they returned to Canada where she earned a graduate degree in statistics at the University of Waterloo (1988).

Dean came to SFU in 1989, where she’s since been designing new and better ways to extract information from statistical data. Her focus is problems relating to recurring illness (e.g., epileptic seizures, cancer, or pine weevil infestations) and the mapping of disease and mortality rates. “We are developing disease surveillance systems that warn when an area’s disease rate is too high and needs special intervention. These models can help health officials more accurately assess the meaning of peaks and valleys in disease rates. Of great current interest is the development of methods to identify susceptibility and resistance to disease.”

Hospitals and patients

A few years ago Dean’s doctoral student, Ying McNab, looked at the survival rates of newborns in Canadian hospitals. She analyzed data on demographics, symptoms, family medical histories, deaths, survivals, and other factors. She wondered why some hospitals experience more infant deaths than others. Are their facilities inferior? Do they just get the more serious cases? Do they get too many cases all at once?

Some intriguing conclusions emerged. “One is that units with ultra-modern equipment attract doctors who tend to push treatment boundaries because of the severity of the cases they face,” says Dean. “They like to test things. So research hospitals often have a higher neonatal survival rate. Then there’s the number of babies being treated at any one time. You’d expect that increasing patient numbers per staff would overwhelm staff and lead to poorer care. Of course at some point this is true, but what is also true is that the more patients a unit handles, the more experience staff gain and the more likely specialized services will be available.”

At Vancouver General Hospital, with doctoral student Jason Nielsen, Dean is tracking women in menopause. Each woman is receiving one or another of two treatments and keeping a daily diary on hot-flush occurrence and intensity. The SFU group is modelling the results over several years. “We want to determine with confidence which treatment is most effective in the long term,” she explains.

Forests and beetles

Dean is also working on problems relating to B.C. forests with doctoral students Farouk Nathoo and Laurie Ainsworth. “We’re studying the pine weevil,” she explains. “On a computer map we located every tree growing on a plot of land, noting whether or not it’s diseased. Now we’re following each tree over time. We’re looking for weevil-infestation trends and trying to identify trees that are resistant. We want to pinpoint if the weevil-free trees are genetically resistant, or are they weevil-free because the diseased trees aren’t close enough to infect them?” She has several years of data on thousands of trees of many varieties.

The results show that highly resistant Trees occur in clusters, with evidence of genetic differences

“So far,” she reports, “the results show that highly resistant trees occur in clusters, with evidence of genetic differences.” She intends to link this work with forest fire analysis and simultaneously study all of B.C. “Forests are an intrinsic part of Canada and its economy. Indeed, the welfare of all Canadians is intimately connected with the health of Canada’s forests, in many ways that are not yet fully understood. My new work will develop tools for the management of Canada’s forests. Of course, such a project would deal with data on a massive scale – a challenge even with our modern techniques.”

In private life, Dean is a wife and mother of two children. She loves the outdoors, yoga, canoeing, bird watching, and Roots Reggae. Professionally she’s part of an old tradition. For more than 2,000 years philosophers and other learned people have observed human beings and nature and tried to interpret and describe what they see. “It’s no different today,” she observes. “We’re just getting better at it.”