Meet the Faculty - Yogatheesan Varatharajah
Tell us about your journey to the University of Minnesota.
I am from Sri Lanka and I did my undergraduate degree at the University of Moratuwa in Electrical and Computer Engineering. Since the beginning of my studies, I have been interested in applying the concepts I learned in classes to real-world problems. I did a project looking at how you can put various sensors on people’s bodies to quantify their energy expenditure during exercises. That work inspired me to look into health monitoring broadly and how we can use different sensors on the human body to develop treatments for diseases.
I moved to the United States for my Ph.D. program at the University of Illinois at Urbana-Champaign. During my Ph.D., I did a lot of work in the newly developing area of healthcare machine learning, in collaboration with the Mayo Clinic in Minnesota. As you can see, I was no stranger to Minnesota and because I would spend time in Rochester during the summer time to work on my projects. I felt a certain amount of kinship with the state, so when I got the chance to interview here, I was really excited. Plus I have the chance to continue my work with the Mayo Clinic and work with new collaborators at the University of Minnesota.
I think there is a rich history of medical related research in this University. Even in our Department, there is a significant focus on computing for healthcare applications. That was a big attraction for me. Plus the city is home to a number of medical device companies and there are great hospitals in this area, giving me a lot of opportunities to do computational work in this space. When I came to interview here, I felt like I was well supported and I got the sense that collaboration across the U of M is a priority. So it felt like the perfect opportunity for me.
We would love to hear more about your research!
If you look at the healthcare system these days, there are a lot of things that clinicians have to do that they are overqualified for, like reading scans and writing reports for them. This leads to a lot of burnout which causes errors in clinical decisions, and it takes away the valuable time that could be spent with patients. On top of that, there is a huge shortage of clinicians in the country. So the question I am broadly interested in is, how can we harness the recent advances in AI to improve this problem? I work with clinicians at the U of M medical school and at the Mayo Clinic to develop AI methods to help interpret tests, optimize treatment decisions, and improve outcomes. From a data science and AI perspective, there are also some very interesting methodological challenges. We have to account for the scarcity of curated data, labeling inconsistencies, predictive uncertainty, model robustness, explainability, and data privacy. This is a very rich area for research and we are making progress steadily. The clinical applications we focus on are mostly in the neurological space. The human brain is a very complex system and harnessing the power of modern AI methods is necessary to improve our ability to treat neurological diseases. The diseases we focus on are epilepsy, Alzheimer's, Parkinson’s, and psychiatric illnesses. We work with data that has been historically collected by hospitals as well as some live data with wearable devices.
What do you hope to accomplish with this work? What is the real-world impact for the average person?
The hope is that our work can help automate some of the work that is taking up a large portion of clinician’s time and augment their ability to treat diseases with AI-powered analytics. I believe that our work will enable clinicians to spend more time with patients, improve treatment outcomes, and focus on more complex problems that really require their expertise.
What courses are you teaching next semester? What can students expect to get out of that class?
I will be teaching CSCI 5980/8980 - Machine Learning for Healthcare: Concepts and Applications. I have noticed that the focus of most machine learning courses is on teaching abstract concepts, which is important. But there are a lot of practical aspects students need to learn when they apply ML to specific domains. When it comes to healthcare, there are different data types, different clinical problems, datasets are not generally clean, and we are predicting more rare events. This course will highlight those differences and work on developing methods that are tailored to address these challenges. We will cover some foundational topics in machine learning for healthcare, discuss some real-world examples, and provide hands-on programming exercises for different types of healthcare data sets - genomics, imaging, time series and clinical text data. We will also discuss topics such as trustworthiness and safety, as well as privacy issues and ethics. There is a lot of interest within the healthcare industry for people to be doing this line of work, so I think the course will be very beneficial to land potential jobs if students are interested in getting involved in the healthcare domain.
What do you do outside of the classroom for fun?
I am a bicyclist. I like to bike on the trails around the city. I also like checking out farmers markets. There are so many in the Minneapolis/St. Paul area that I have been exploring. I also like to go on walks in our various parks and look at all the fall colors.
Do you have a favorite spot in the city?
I love the Minnehaha Falls area and walk there frequently. I also bike to work and like stopping at the Franklin Bridge to take pictures of the river and the scenery around it.
Is there anything else you would like to share?
I have a lot of projects in healthcare ML that I would like to start working on, so if there are any students in the data science or computer science programs that are interested in this work, I encourage them to reach out to me! I am open to undergraduate, master’s and Ph.D. students.