HQP Spotlight: Omar Boursalie, McMaster University


Omar Boursalie, McMaster University

Omar Boursalie is a second-year PhD student in the School of Biomedical Engineering at McMaster University. His research focus is on overcoming challenges in health care through big data and machine learning. His supervisors are Dr. Thomas Doyle and Dr. Reza Samavi at McMaster University. The industry partner for his research project is Real Time Medical.  Boursalie won first prize at SOSCIP’s Impact Conference in 2017. SOSCIP reached out to Boursalie to find out what he’s been up to.

What is your SOSCIP R&D project?

My project is “Pan-Canadian Data Collection and Analysis Platform for Patient Radiation Protection and Safety”. Dr. David Koff, Dr. Thomas Doyle, and Dr. Reza Samavi (McMaster University) are the principal investigators. Medical imaging like CT scans and x-rays are powerful clinical tools to diagnose patients without invasive surgery. However, there are ionizing radiation risks associated with medical imaging that needs to be considered. Our SOSCIP project is to develop a decision support system for evaluating a patient’s cancer risk due to low-dose radiation exposure from medical imaging.

What has it been like to collaborate with Real Time Medical?

It has been a great opportunity to collaborate with our industry partner Real Time Medical. They have provided valuable insights on how to connect our research with industry, by helping us understand the market expectations for our project.

How did you get involved?

I was invited by Dr. David Koff (McMaster University Medical Centre) to join this project due to our expertise in health informatics, health data privacy management and mobile predictive health analytics. For my MASc thesis, I developed a remote patient monitoring system for heart disease that analyzed health data using machine learning algorithms directly on a mobile device.

Why is this project important to Canadians?

Canadians are concerned about the potential risk of low-dose radiation when they go for a medical scan. The SOSCIP project gives me the opportunity to help solve this challenging problem in our health care system. Developing a system that can potentially give Canadians some peace of mind is a great way to spend my Ph.D. Furthermore, the project has given me the opportunity to work with hospitals, medical vendors, health care professionals and the SOSCIP platform. I had no previous experience with high-performance computing – I did my master’s degree on a laptop! Having access to a supercomputer is great for my research as I can now run programs in a single morning that used to take me the entire day!

Why should everyday people care about this project?

Studies have shown that about 50 per cent of our radiation exposure in our life comes from medical imaging. All Canadians are affected by medical imaging, either directly by having a medical scan, or indirectly through a friend or family member. For example, my father is concerned about the potential risk from low-dose radiation exposure because he has had multiple medical scans over the last few years. Being able to give my dad piece of mind the next time he has a scan motivates me while conducting my research. Our proof-of-concept system will allow health professionals to monitor the risks of low-dose radiation emitted by medical imaging tests for improved radiation benefit-risk dialogue. Long term, our research will benefit the health of Canadians by contributing to our understanding of the long-term effects of low dose radiation.

Why is this innovative or unique?

There is currently no system in Canada to track a patient’s lifetime cumulative radiation dose from medical sources. In addition, existing models for low dose radiation are controversial because they are based on the radiation exposure from Japanese atomic bomb survivors. Interestingly, current CT and x-rays scanners do record the amount of low-dose radiation that is emitted. However, the emission data has not been leveraged to develop a risk model. Our contribution will be using radiation exposure from medical devices as well as data from a patient’s health record to construct a low-dose radiation risk model.

What do you hope to achieve with this technology?

Our project has three goals: 1) develop a model for medical risk assessment using medical records and imaging data; 2) deploy the model as part of a decision support system; and, 3) develop architectures and frameworks for decision support systems that can be used to advance research in other fields.  The goal is to make our project accessible to everyone.

Your prize for winning the competition was IBM mentorship. How has this opened new opportunities for you?

The IBM mentoring prize was an invaluable opportunity to develop my professional skills such as leadership, networking, career planning and time management. It was a fantastic experience meeting and learning from senior executives at IBM. I am very grateful to Mr. Stephen Timms (Business and Operations Executive at IBM Canada) who was my principal mentor at IBM.

What are you up to now and what are your goals for the future?

You know the saying, “you can’t teach an old dog new tricks”? Well, this saying describes what happens to a machine learning model once it is deployed into the workflow. Currently, to update the machine learning model the hospital would need to hire a machine learning expert to rebuild the entire system. In our case, we would need to go back to the SOSCIP platform and rerun everything. I am investigating ways to assist machine learning algorithms to relearn while minimizing human oversight. In effect, we are trying to help the old dog (machine learning models) learn new tricks!

My future goal is to complete a postdoctoral fellowship at IBM Canada to gain industrial research experience before pursuing a longer-term career in academia or industry in Canada.