Anton Nikouline: Incorporating machine learning into emergency medicine
FRCP Emergency Medicine Resident Anton Nikouline will present his research project on December 7 as part of the T-CAIREM Trainee Rounds.
T-CAIREM will feature presentations from emerging U of T researchers on December 7. Emergency Medicine Resident Anton Nikouline will discuss his research into using machine learning to improve emergency medicine. We caught up with him to learn more about his innovative research.
What attracted you to Emergency Medicine?
I picked Emergency Medicine because of its breadth and ability to impact patients quickly and will soon be completing a fellowship in Critical Care. I took a particularly keen interest in Trauma after seeing how dynamic the specific patient population is. There is something called the “Golden Hour” in trauma, where patients have a higher likelihood of surviving if they receive immediate interventions within the first hour. This takes a lot of coordination from paramedics, hospitals, and admitting services to get patients quickly assessed and given the intervention they need to survive. This coordination of teams, high-stakes decision-making, and huge impact has made me really passionate about improving trauma care.
What inspired your research?
Currently, best evidence shows that providing early blood products for patients that need them gives them the best chance of helping them survive. Unfortunately, blood products are a limited resource and are not a benign treatment when given quickly. As a result, it becomes very important to decide who needs blood products immediately with certainty. A few prediction scores have been developed with decent results, but require data points that are either unavailable quickly or difficult to get. Luckily, we have a very large database of trauma patients. I believe that machine learning could be a great tool in predicting the need for blood by using early markers found in this dataset.
What outcomes do you hope your research will eventually lead to?
The hope is to have a model that can accurately predict which patients need blood even before they get to the hospital. Ideally, this model could calculate predictions based on data points collected by paramedics and the blood could be ready as soon as the patient rolls through the hospital doors. This would be a physician-support tool throughout the trauma resuscitation as well, helping doctors make informed decisions along with their judgement.
What are you professional goals after you complete your studies at the U of T?
I have been accepted to a critical care fellowship at Western University and will be continuing my work there. I hope to continue my passion for creating both trauma and critical care prediction tools to deliver optimal patient care.
What advice would you give to new students who are interested in following in your footsteps?
Be kind to yourself. It’s important to make plans of what you want, but know that those plans may not work out. Roll with the punches and know that how you handle defeat is just as important as your accomplishments.
What’s the best part of doing the type of research that you do?
I think the multidisciplinary team that I’ve been lucky to work with has been the best part. Trauma is a dynamic process with multiple healthcare specialists including RT, nursing, Anaesthesia, Orthopedics and General Surgery. My research also has me working with specialists in computer sciences, clinical epidemiology, and others outside of medicine. I believe this diversity of backgrounds allows for a great combination of ideas and suggestions that have really pushed me to be a better researcher.
