ƬƵ

Building a Tool Around the Power of Data

Dr. Madathil's dissertation wins a prestigious award, and lays the ground for more innovative work to come.

When Sreenath Madathil was a practicing dentist in India, he frequently saw patients that had oral health issues related to their betel quid chewing habit – a product containing smokeless tobacco, betel leaf and betel nut that is popular in South East Asian Countries, and India in particular, and poses a huge risk for oral cancers. He could see the ill effects on the oral health of betel quid users—both in his family, and his patients—but lacked the tools to effectively communicate the health risks these habits pose.

Now, using the research established by his dissertation project, which was awarded the Gordon A. Maclachlan prize for distinguished dissertation in any discipline of the health sciences, Sreenath is building that very tool. By putting the power of health data in the hands of clinicians and patients the world over, he hopes to help them better communicate and understand the health risks such habits pose.

From Calicut to Montreal

Sreenath met Dr. Belinda Nicolau, Faculty of Dentistry professor and researcher, when she and her team were starting the HeNCe Life Study in the same school he was completing his residency. In the HeNCe Life Study, ƬƵ researchers take a life-course approach to studying the etiology of head and neck cancers, conducting hospital-based control studies in multiple urban health centers, including Montreal (Canada), Sao Paulo (Brazil), and Calicut (India). Sreenath already had a profound interest in research, and so he joined the study as a research assistant.

During the two years that followed, Sreenath collected biological samples and conducted life course interviews with over 350 subjects of the study. As his research assistantship in the HeNCe Life Study was ending, Sreenath decided to join Dr. Nicolau at ƬƵ to complete his Masters in Dental Science, where he further analyzed the data he helped gather over the past two years. More specifically, he was interested in the intergenerational transmission of betel quid chewing habits among Indian families and the dose-response relationship between betel quid chewing and oral cancer. In short, he found that the risk of oral cancer grows as tobacco dosages increase, but there is no minimum dose of betel quid which is devoid of any risk for oral cancer. “Put otherwise,” Sreenath notes, “there is no risk-free amount one can use, and most people often underestimate that risk.”

Yet Sreenath identified an important gap in the literature: how the risk behaviours in a person’s entire life affect their risk of developing cancers. His research asked, are there periods in a person’s life that are highly sensitive to risk behaviours like smoking, and put individuals at higher risk of developing cancer later in life? “In the study,” Sreenath continued, “we could already see that smoking earlier in life is more harmful than later in life – but to do this empirically is very hard, and there are no statistical models to accomplish this.”

Fortunately, working with the HeNCe Study put Sreenath in the exact right place to answer this very question: because he had access to participant data gathered from their entire lives, he could then develop a methodology to predict the risk these habits pose. In fact, the research Sreenath conducted on the betel quid chewing habit proved his hypothesis: chewing one betel quid per day before the age of 20 carries the same health risk as chewing 3 per day between the ages of 20-40, or 11 per day after 40. “The earlier the exposure, each of the betel quid has more weight.”

Delivering a New Message

As of now, the common public health message for tobacco use is, ‘Don’t start,’ or, ‘if you’re already doing it, stop.’ But Sreenath identified the need for a more accurate public health message that accounts for when and at what dosage that habit develops.

“We have found that there are periods in life that are very critical for developing these habits, and this has profound public health implications…So more public health resources should be directed towards reducing early initiation, and the period before 20 years of age….this will help prevent early risk to accumulate.”

But in order to change the public health message to a more accurate, nuanced, and personalized communication, you need the proper tools to help support that assessment. And that’s exactly what Sreenath is building next.

Building the Tool

Partnering with Dr. Marc Coates, professor in the ƬƵ University Department of Electrical & Computer Engineering, Sreenath is using machine learning techniques to create an accurate and personalized risk prediction tools for use in clinical settings. Based on the data accumulated in the HeNCe Life study and the steady flow of incoming user health data as the tool enters testing, this algorithm will allow clinicians to assess the health risks and probabilities of developing cancers due to these specific risk behaviours. “We are building a decision support tool for clinicians,” Sreenath claims, “a tool clinicians can use to help them make health recommendations to their patients. But it is also a powerful communication tool for patients to understand the health risks they are taking.”

Although it is currently being engineered as a risk prediction tool specific to these cancers, and these risk behaviors, Sreenath believes that in future iterations, the tool can be reconfigured for other types of cancer or risk behaviors.

From the methodological innovation that allows researchers to understand how certain periods in our lives are more sensitive to risk behaviours, to building a machine learning tool to support clinicians in their everyday challenges, Sreenath’s success as a researcher, innovator, and leader in his field are indebted to the vibrant culture of research in the Faculty of Dentistry, its rich network of interdisciplinary collaborators, and the support of his peers, colleagues, and mentors. The Gordon A. Maclachlan prize may only signal the beginning of a longer story yet to unfold.

Back to top