Modern Technology Enables Objective Data Mining

Would you say you live an active lifestyle? How many steps do you think you take each day? How long do you spend being physically active? How intense is this activity? See if you can answer these questions without looking at your phone or your smart watch! It's hard to put a number on physical activity, especially when we're asked to remember what we did several days, months or even years ago. Yet decades worth of research has relied on people's answers to self-assessment surveys asking precisely these kinds of questions. These studies have established a link between dementia and physical inactivity, which is estimated to account for some 3% of the overall risk of developing the disease. Physical inactivity is what's known as a modifiable risk factor because it's linked to our behavior. Non-modifiable risk factors, such as genetic factors, meanwhile, account for around 65% of the overall risk.

The established risk associated with physical inactivity is undoubtedly small. "But the true figure could be much, much higher," says Charlyne Bürki, who's just completed her Master's thesis focused on comparing the robustness of data from subjective studies and from objective, measurement-based research. "With modern technology, we have the tools to mine objective data," she adds. "We should use this capability to extract new insights."

Bürki got her first taste of data mining during a semester project at EPFL's Digital Epidemiology Lab, which is led by Prof. Marcel Salathé. For her research, she studied the link between diet, physical activity and blood sugar levels. "Marcel Salathé's approach really clicked for me," says Bürki. "It uses the latest technologies - such as smart watches and a real-time blood sugar monitors - to continually and non-invasively track participants' physical activity and physiological parameters. For me, it's a much more objective way to collect data than asking someone to remember how active they were after the fact."

Personal motives

For her Master's thesis, Bürki focused on quantifying the link between dementia and physical inactivity. It's a subject that's close to her heart. "Alzheimer's disease runs in my family," she explains. "My grandmother is living with the condition right now, and I've seen firsthand how it affects the caregivers. We also know that women have a higher risk of developing dementia - and I have two sisters."

Bürki, who holds dual Swiss and U.S. citizenship, conducted her research as part of a six-month fellowship at Harvard Medical School. "Retroactive studies have shown that people who live sedentary lifestyles could be at greater risk of developing dementia and other cognitive disorders," she says. She tested this hypothesis against objective data mined from the UK Biobank, a large-scale database containing health and genetic information from half a million UK participants recruited between 2007 and 2010. "Some 100,000 of these participants, aged between 45 and 79 years, wore an accelerometer for a week five years after they were recruited," adds Bürki. "That provided the robust data I needed for my research."

The first surprising finding was the exceptionally low prevalence of dementia among the cohort, irrespective of physical activity. Could that be attributed to selection bias? "Possibly, since the participants were generally well-educated," says Bürki. "But all the factors are interlinked and it's very hard to isolate the effect of the modifiable factors" - i.e., the non-genetic ones. "For instance, it's impossible to tell whether having a higher-education degree or working in a mentally challenging job prevents someone from developing the condition."

Another unexpected discovery came from the ex-post survey, in which respondents over-reported levels of physical activity: "The accelerometer data showed that, on average, participants spent two-and-a-half hours more each day being sedentary than they claimed." Although the participants couldn't accurately recall how long they'd spent moving, the data showed that they were nevertheless active. But what constitutes "physical activity" remains unclear, especially when it comes to its ability to protect against dementia. At what age does physical activity make a difference? How often should people exercise, and at what level of intensity? "It's hard to isolate this factor, not least because we know that high blood pressure, diabetes and obesity increase the risk of developing dementia, and they're all closely linked to physical inactivity," explains Bürki.

Striking a healthy work-life balance

"My thesis research challenges the validity of subjective studies and highlights the importance of relying on objective, measurement-based data," adds Bürki. We now have the tools to objectively calculate the levels of modifiable risk factor truly achieved by participants. This in turn will be critical to assess an unbiased correlation between the former and dementia. Prevention will prove critical to keeping health-care costs down in today's aging society."

The task facing Bürki now is to thoroughly edit her thesis before it's published. Beyond that, she intends to begin studying for a PhD at ETH Zurich, where she hopes to specialize in medical data mining and embedded technologies. So is she turning her back on EPFL? "I've been at the School since 2015," she says. "It's time to move on." What did she make of her time in the United States? "I enjoyed the experience, but I don't think it's the right place for me to do a PhD," she replies. "I'm a strong believer in the need to strike a healthy work-life balance. I'm more likely to achieve that here in Switzerland than in the United States. But I may go back there one day - who knows?"