From mathematics and engineering school to cardiac anesthesiology and epidemiology, Dr. Louise Sun’s journey has been remarkable. The University of Ottawa Heart Institute’s cardiac anesthesiologist talks big data and health care delivery with Hazel Tang.

Two years ago, a study published in the Canadian Medical Association Journal hit the headlines. In summary, it revealed that even though men are more likely to have heart failure, women tend to be the ones dying from it. The study speculated men and women might be suffering from different forms of heart failure as they did not share the same symptoms. Often, the type of heart failure detected in women was not as readily identifiable and came with fewer effective therapies. In a follow-up study looking at the impacts of gender and ethnicity in heart failure patients, again, women who underwent coronary bypass surgeries were found to have worse outcomes than men.

The principal investigator behind this renowned series of women heart health studies is Dr. Louise Sun, MD, SM, FRCPC, FAHA Cardiac Anesthesiologist & Director of Cardiocore Big Data and Health Informatics Research at the University of Ottawa Heart Institute; Assistant Professor of Anesthesiology and Epidemiology at the University of Ottawa; and an Adjunct Scientist at the Institute for Clinical Evaluative Sciences (ICES) in Ontario, Canada. Sitting down with Dr. Sun, she tells me the findings suggested possible gaps in the delivery of care in female patients. Generally, less invasive testing was done in women even when they were diagnosed with coronary disease and they were referred for specialist care or interventions sometimes too late in their disease spectrum. But Dr. Sun believes some of these disparities, can perhaps be bridged if data is leveraged constructively. “We can integrate detailed clinical and administrative data to design personalized care maps and inform policymakers and healthcare leaders of the probable biases and loopholes in the system,” she says passionately. “We should also bring patients onboard to make sure the care is patient-centered. That’s what our team is working on.” Indeed, Dr. Sun and her research team are using a variety of machine learning techniques to model disease progression and predict outcomes mainly after cardiac procedures. They are predicting the types of resources needed by patients after surgery or after being diagnosed with chronic cardiovascular disease.

Interestingly, data has not only been an important part of Dr. Sun’s work but also her life as well. Coming from a family of mathematicians and engineers, she has always been attuned to mathematics, physical sciences and computer programming from an early age. But she always yearned to be in a field where she could exercise her numerical wit in addition to being able to directly make a difference in others’ lives.

“I pursued my undergraduate study in biomedical engineering, thinking it’s a specialty that incorporates a lot of modern sciences and medical practices, so it’s a place where I could exercise my talents and passion,” she says. Dr. Sun enjoyed and did well in engineering school but it was not until she received a national engineering and science scholarship that enabled her to work on cutting-edge biotech projects at the Canadian National Research Council, that she believed she could do even more.

“I spent a summer in Winnipeg figuring out ways to refine the 3D reconstruction of brain tumor images, which would help to improve the precision of Gamma Knife therapy. I was really inspired by my peers who were doing all sorts of insightful experiments. Several of these kids were planning to pursue a career in medicine and they did a good job persuading me to do the same. Somewhere along the program, I decided to give med school a try. That fall, I applied, not fully appreciating this would open my future to a new domain with so many unsolved challenges.

Anyway, I think that no matter where I am, I will always be a mathematician at heart. I have always been into the physical types of physics. I love mechanics; especially biomechanics and fluid mechanics. I knew I had to settle on a specialty which combined human physiology, technical skills and practical aspects of mechanics, and cardiac anesthesiology was the one. It consists a lot of realtime fluid dynamics, circuitry, and human physiology. One would have to rapidly synthesize these together to care for the sickest patients in the operating room and in the Intensive Care Unit (ICU). It just seemed to be a perfect fit.”

Dr. Sun began to code with clinical databases at the time she was enrolled in a Master in Clinical Epidemiology at Harvard as a resident physician. Given her engineering and programming background, she chose to focus on the biostatistics and data mining. She maneuvered comfortably between different statistical and software packages. It was at the time of her clinical and research fellowship training that she realized there are bigger datasets out there; those that were collected beyond institutional level. She realized some of these datasets had important clinical and policy related implications and had a real influence in guiding the decisions made by governments and ministries.

Experimenting with bigger datasets also exposed Dr. Sun to the many challenges relating to data. As she got more experienced, she began to appreciate that there are nuances to every dataset. These need to be well understood by the data scientist in order to make algorithms efficient and effective. For instance, knowing the clinical context of the data collection, the pattern of missing or misclassified values, and even the idiosyncrasies of individual data abstractors could help to produce more reliable and reproducible results. She also believes that the quality of data is not something one can take for granted.

“You know, there are big databases and there are big databases that are well done,” she says. “Most of the time, knowing where the data comes from, and knowing whether these data sources are suitable for accumulating specific types of information, can really help you in manipulating and interpreting that data.” Another key challenge is that the sense of propriety over the ownership of data is keeping people from sharing and working together. To tackle that, Dr. Sun felt one has to “work with the right groups of people who believe in science as a collaborative effort”.

“I would say, we need to work with other clinicians, researchers, as well as policymakers and administrators to get a well-rounded understanding of the key issues in healthcare. We also need to partner with patients to understand what treatment goals are most important from their perspective. We need to let them know the differences everyone can make when there’s synergy.”

That said, Dr. Sun points out that sometimes strict privacy and data protection rules make data sharing a lengthy process. “For instance, in one of the research projects, we tried to pull data together from several provinces but we ended up each analyzing our own dataset. We had to synthesize our results using meta-analysis methods because, administratively, it would have taken too long to pull all the data together”. Nevertheless, Dr. Sun thought that should not be a road block to further venture into the world of data, particularly in face of a global health crisis such as the COVID-19 pandemic.

“At ICES, data are usually updated once every three months or so, and sometimes longer depending on the data source. But because of COVID-19, we are actually getting daily updates on various COVID related test results just so that researchers can have real-time feed to make accurate and up-to-date pandemic-related predictions. This type of support is very important for pandemic response and planning”.

Dr. Sun believes the impact of COVID-19 may have been altered if there were more early reports from the source of its origination and if these reports received enough international attention earlier on. As such, being equipped to pick up on data from local outbreaks and having the capacity to model regional and global spread may help to avert widespread crisis. But before that, as Dr. Sun noted, the definition of big data should also be expanded.

She believes the term “big data” should not be restricted to what’s being collected from administrative and electronic health records but also those from the field where care is delivered. This would also include bio-banked materials and genomic data. Consequently, as the type and quantity of data collected, as well as the practice of medicine continue to evolve over time, so are the models that are built upon them. This adaptive modeling process will lead us into the future of medicine and healthcare system planning. This is a future where preventive care can be personalized, end stage complications avoided, and pressure on the healthcare system reduced.

“I think big data’s potential will be limitless, not just in terms of making personalized risk predictions but also in supporting micro-simulations at the population level,” she says. “For instance, what does the future look like for a particular group of people with a certain set of risk factors? How should we plan the delivery of care to arrive at the best possible outcomes? How do we prevent bottlenecks and save on healthcare cost while still delivering the best care possible? There’s a lot of work to be done but this means there are also a lot of exciting breakthroughs we can look forward to.”