From identifying which crops will grow best in developing countries, to improving our understanding of brain tumours, to changing how we study and understand children’s health, few people have contributed to as many different areas of research as Paul McNicholas.

McNicholas doesn’t specialize in agriculture, oncology, or public health. Instead, he’s an expert in the one thing that connects nearly every field of research: data.

McNicholas, a professor of mathematics and statistics, Tier 1 Canada Research Chair in Computational Statistics and director of McMaster’s MacData Institute, is a leader in developing new, cutting edge methods of analyzing data.

Read: McMaster statistician awarded prestigious Steacie Fellowship

Through his own research program and through his work with a number of  MacData graduate fellows, McNicholas is collaborating with researchers across campus to apply novel data science approaches that are helping to reveal hidden insights from complex data sets, and which are playing an increasingly significant role in accelerating research at McMaster and beyond.

“Here at McMaster, we have lots of faculty who generate wonderful data using cutting-edge equipment and who are asking really important questions,” says McNicholas. “But they may not be getting as much from the data as they could and that’s where we come in – we bring people together to get as much from the data as possible.”

Using statistics and machine learning methodologies, McNicholas is helping researchers from diverse disciplines approach data analysis in new ways that better reflect the complexity of real life and human behaviour – something that’s especially valuable when it comes to children’s health outcomes.

Understanding “turning points” for children with autism

McNicholas has been working closely with McMaster researcher Stelios Georgiades, founder and co-director of the McMaster Autism Research Team(MacART), a research collaborative focused on generating knowledge to improve clinical practices, programs, and policies to support individuals, families, and communities living on the autism spectrum.

Children with autism are currently grouped into “developmental trajectories” based on assessments performed at the time of diagnosis. These trajectories predict how children might develop and can also help determine interventions.

“We assign kids into these rigid and mutually exclusive trajectories, but they often don’t reflect their pathways over time; some kids could change trajectories either because of treatment, natural development, or both” Georgiades explains.

To better understand these changes and their impact, McNicholas and Georgiades went back into the existing data set from the Pathways in Autism Spectrum Disorders study and, using a new data science methodology, were able to identify “turning points” in developmental trajectories – key transitions in a child’s life where they could experience setbacks.

“If we know that some kids tend to deviate from their trajectory at certain transition points – when there’s a change in the family, a change in school – then we can be proactive and make sure the necessary supports are in place for those kids so they can continue to do better rather than continue to have setbacks,” says Georgiades.

Georgiades says McNicholas’ approaches are making it possible to better approximate reality and reflect the complexity of autism. “Kids are changing all the time, but we’re not capturing the change if we stick to the old ways of doing research. With the right tools we can see why kids are changing and we can use that to build new and enhance current interventions.”

Measuring the power of play

Not only are kids constantly changing, they’re always on the go. That’s where McNicholas’ work with Joyce Obeid, an assistant professor with the Child Health & Exercise Medicine Program in McMaster’s Department of Pediatrics, comes in.

Working with Obeid, McNicholas and MacData fellow Peter Tait have been developing data analysis methods to gain new insights into the connection between physical activity and health in children.

Obeid and her team evaluate how children’s everyday movements contribute to their overall health, measuring physical activity with accelerometers – devices that monitor movement throughout the day. The data is then grouped into categories: time spent in light, moderate and vigorous activity, and time spent being sedentary.

While this method captures important data, Obeid points out that it doesn’t reflect how activity is broken up throughout the day – information that she says, could be very significant.

“It’s about more than just the minutes,” she explains. “As an adult, you may go to the gym and exercise for an hour, then you’re done for the day, but kids don’t do that. They’re stop and go, they’re running around on the playground, they sit down for a while, then they’re back up again, and we think there’s something about that pattern of activity that’s important for growth and development.”

McNicholas and Tait are modelling the accelerometer data to better understand how different children move throughout the day and what that means for their health.

“We used to think, OK, if this child is meeting the 24-hour Movement Behaviour Guidelines then technically, they should be healthier,” says Obeid. “But the numbers don’t always show that – not because it’s not true, but perhaps because the way we’ve been looking at accelerometer data is superficial. The technology has outpaced our data science and we’re working with Paul and Peter to change that.”

“The guidelines recommend that kids get 60-minutes of moderate-to-vigorous physical activity every day,” she continues. “We know they’re not going to do this in one 60-minute burst, but we don’t know if there’s a best way to break that activity up. We think that the pattern matters for health, and our research is exploring that hypothesis.”

Obeid, who is working with McNicholas and Tait on multiple projects, says that while her use of these methodologies is still in the early stages, it’s changing the way she looks at data analysis.

“Paul is helping us to see things we wouldn’t otherwise have been able to see in the data and to represent it in a way that’s meaningful,” she says. “Meaningful to a scientist, but also meaningful to a clinician who’s talking to their patient about physical activity.”

Unpacking the “immigrant paradox”

McNicholas’ work is also providing new insights into research aimed at helping children with a different kind of health challenge.

McMaster researcher Kathy Georgiades, the David R. Offord Chair in Child Studies and an associate professor of Psychiatry and Behavioural Neurosciences, has been working with McNicholas and MacDATA Fellow Michael Gallaugher to better understand the mental health outcomes of immigrant and refugee children in Hamilton.

Georgiades explains that numerous past studies – including her own research – have all pointed to the same conclusion: Immigrants, despite often experiencing social adversity like poverty and other stressors, tend to have better mental health outcomes than those born in Canada, a result that researchers call the “immigrant paradox.”

But local teachers were telling Georgiades a very different story. “There were two groups of children they were most concerned about – ESL and refugee children,” she explains. “Teachers were reporting elevated risk in these sub-groups and they weren’t seeing the same pattern of resilience.”

To investigate further, Georgiades and her team went back into their data set, separating the ESL and refugee students from the broader foreign-born population, and soon started to see elements of risk.

“Some children were doing well in terms of mental health outcomes and some weren’t doing as well – particularly those who were struggling with English,” she says. “So, we were actually starting to see a diversity of mental health functioning within this population that we had previously grouped broadly as foreign-born.’”

To better understand this diversity, Georgiades turned to McNicholas and Gallaugher who are now analyzing Georgiades’ data to identify the factors that best predict mental health challenges in immigrant and refugee children.

Instead of grouping all immigrants into the category of “foreign born,” McNicholas and Gallaugher have identified sub-populations and are factoring in a host of variables including pre- and post-migratory  stressors, such as exposure to war and trauma, and significant  economic hardship, or neighbourhood disadvantage, to find out which have the strongest  correlation to mental health functioning within each sub-group.

“These data science methods are making it possible to dig deeper and go beyond broad categorizations to understand the impact of actual life experiences,” Georgiades says.

“By using these approaches, we’re going to identify robust associations in the data. This will help us better understand how the experiences of these children and their families shape mental health outcomes and will allow us totarget mental health investments and supports for these children more effectively.”

The MacData Institute was created to identify synergies to foster collaboration among McMaster University’s institutes, centres, and researchers whose work involves the many facets of data. 

The MacData Institute will offer its first summer school May 27-31. Learn more about the MacData Institute and the summer school.