Dr. Laleh Seyyed-Kalantari, associate scientist at the Lunenfeld Tanenbaum Research Institute, reveals her fearless work ethic, her upbringing in Iran and how she invented an earthquake alarm that proved nearly fatal for her father!
Dr. Laleh Seyyed-Kalantari is an associate scientist at the Lunenfeld Tanenbaum Research Institute, Sinai Health System.
She was a postdoctoral fellow at the Computer Science Department at the University of Toronto and the Vector Institute for Artificial Intelligence, focusing on developing AI-based medical image diagnostic methods. She received a Ph.D. degree in electrical engineering from McMaster University in 2017. Her research interests span from machine learning in healthcare, AI in medical imaging, to optimization, and numerical modeling. She has received several highly competitive national, provincial, and institutional awards including the NSERC Postdoctoral Fellowship (2018).
What initially sparked your interest in medical AI?
My background is in electrical engineering but my interest in healthcare AI was sparked when I fell ill while completing my PhD. I remember consulting several doctors and each of them had varied opinions and proposed different treatment. Interestingly, my PhD research focused on optimization in the context of electrical engineering. That got me wondering if I could optimize care so that patients get diagnosed and treated more rapidly. I recovered eventually and began working on the application of AI in healthcare, which I believe would have real impacts on people’s lives.
Is engineering something you always wanted to do?
I’m from Iran. My father was an engineer so I guessed he somehow must have influenced my career choice. Yet, growing up, I would tell people that I wanted to be a scientist rather than an engineer. I remember reading books and stories on famous inventors and how things we used in everyday life were made. Gradually, I started thinking about inventing things that I could call my own.
One of them was the earthquake alarm bell. Earthquakes were common where we live, and many people would be killed whenever one struck. When I was eight or nine, I told my dad I wanted to invent something to alert people. I dug a hole in our yard and made it bigger every day so that it was big enough to insert a pole that had a bell tied at the end. My thought was when an earthquake strikes, the pole would shake, and the bell would ring to notify people to run. Unfortunately, my father accidently fell into the hole and decided it was so dangerous it had to be filled in!
My next target was to invent an artificial heart. I was 11 at the time and had some preliminary ideas of doing literature review and systematic documentation. Since there was no computer or internet, I looked for books related to human anatomy, record down the different parts of a human heart and materials that I could use to reconstruct it. Of course, I never succeed in building anything, but they were the formative years of my becoming who I am today.
In what areas do you see the next big advances in medical AI?
I believe we are still in the early stages, but I hope to see more deployment of good AI model in the clinical setting. Indeed, what we are doing right now would help physicians and patients but much of it remains in research papers. It’d be nice to witness how some of these new advances are impacting people in meaningful ways. At the same time, I am a little worried about health equity. I think we need to consider very seriously the consequences of deploying AI in clinical practice and whether they might bring harm to minorities or underrepresented populations.
People seem unsure how to continuously assess AI to ensure it’s safe for all patients regardless of their sex, age, race or socioeconomic status. We are building algorithms for the public, so we need datasets from as many healthcare systems as possible to ensure the end-product can be generalized and deployed to different patient populations. This is challenging. As long as healthcare systems keep data to themselves rather than sharing with researchers, it will delay AI deployment.
Does gender or racial inequality in medical AI frustrate you? What can be done to better address those challenges?
It does frustrate me greatly. Last year, I was leading a project which leverages AI to examine the fairness of medical image classifiers. We found that if the patient was young, he or she is at a higher risk of being misdiagnosed. Likewise, if a patient is an ethnic minority, of lower social-economic status, and on some sort of government benefits, they are also likely to be misdiagnosed at a higher rate compared to general poulation. I am glad that the algorithms were able to identify disparities that exist in practice even though they only read chest x-rays of patients and didn’t reveal any other details. So I believe as much as we are concerned that AI may aggravate health inequities, it also shows the potential of solving some of them.
I am glad that many researchers have realized this and are trying to adjust the algorithms and decisions made by them to include more criteria so that AI models can work better and be fairer. But it’s important to remember that AI is not perfect and suffers from unfairness. So, the process of making AI workable is ongoing and non-stopping.
Professionally, what do you consider your biggest achievement and failure?
I don’t believe in biggest achievements and failure. When I want something professionally, I try my best to attain it. But I know once I reach it, there will be something higher and out of reach that I want to attain. So I regard achievement as a continuous process, like climbing up a mountain. There is no such thing as the “biggest achievement”.
As for failure, I remember when I was in my 20s, I tended to imagine the worst-case scenario. Like my career would be destroyed if certain things didn’t turn out to be the way I expected. Over time, as I became more mature, I realized that nothing is permanent. Things would still go to plan even if I didn’t achieve something or something went wrong along the way. As long as I keep moving and do my best, failure isn’t a failure. Like achievement, it’s just part of a process.
What’s the greatest challenge you’ve overcome?
Again, I am not sure if it can be accounted for “greatest”. As mentioned, I became interested in healthcare AI while completing my PhD and before that, I was mostly working on developing optimization algorithms for engineering application. I had strong mathematical and optimization background in engineering but I don’t have much understanding and experience in healthcare.
When I began my postdoctoral training under the supervision of Dr. Marzyeh Ghassemi, I thought medical AI is a fast-paced domain. If you have a research idea, you have to build a model and test it out within months or someone else will be doing it. So, the research culture is vastly different from engineering. It’s logical. Electrical engineering, computer science, AI and medicine are all different domains, so the research cultures ought to be different, coupled with the different tools and topics. There are so many things to learn for those who are moving from another domain.
I remember my supervisor had assigned me a task; I must submit a paper to a conference in three months. As the paper’s first author, I needed to build all the models required for the project. That was very challenging for me since I am not experienced in the field, but I managed to complete it before the conference deadline. If it wasn’t for my supervisor, I wouldn’t have pushed myself. My supervisor regarded me as someone professional enough to accomplish this and I became that person. I believe this is one of the biggest challenges I have overcome in last two years.
Who do you admire most?
My father. Others say that we tend to see fathers as a big, strong figure in our lives when we were young. But for me, it’s more than that. My father is hardworking and a person with a strong will. He never gives up. Once he sets himself a goal, he will stick to it, even when everyone else has given up. Witnessing that attitude in him since I was young has shaped who I am today. As a researcher, I know I mustn’t give up too easily.
What’s been the best piece of advice you ever received?
I took the Iranian University Entrance Exam at the age of 18. I remember a teacher told us if you are looking for the best time or environment to study, that perfect situation will never happen. So try to use what you have right now. Use all the time and resources you have to do what you want to do. Don’t wait for anything. This piece of advice stayed with me through different stages of my life and it’s precious. So I don’t wait for things to happen. I just delve straight in and make things happen.
What advice would you give to someone starting a career in AI?
Dig into some problems. Don’t be afraid, even if you only have some basic knowledge, you are capable of undertaking a project and building your knowledge in practice. Again, don’t wait for the perfect time because it will never come.
How do you relax?
I love playing with my three-year-old daughter. I also love chocolates. But I try to control myself by not keeping any chocolate around!
What would you tell your younger self?
I would say enjoy all the moments you have and don’t be afraid of the future. I remember when I was young, I had this feeling that if I didn’t reach this specific goal, my world was going to end or reaching to my future professional goals would be destroyed.
For example, if I wanted to attain this goal, I’d tell myself, you have to go from step A to step B and then step C. I had a fixated mindset of what I could do and what I had to do. But as I became more mature, I realized there are many ways to achieve the same outcome. Even if I didn’t get to step B, it doesn’t mean I won’t attain my goal.
So, I’d like to tell my younger self, as long as you keep moving, nothing is permanent, including failure. You are likely to pick up from where you failed and start all over again. Just enjoy time and make the best of what you have and don’t be afraid. It’s an important lesson for life.