May 15, 2024 — Mostafa Al Arshi (SM '24) is a cardiologist with a passion for computer science. After graduating from his Science in Health Data program at the Harvard School of Public Health, he plans to continue his research using machine learning to improve heart health.
I have always been a science-minded person. I have been interested in chemistry and physics since I was a child, and in college I took biology classes in more depth, which I found very fascinating. At the same time, I started thinking about how I could use my knowledge and skills to do something impactful. I get as excited about theoretical things as the next person, but at the end of the day, I want what I'm learning to make people's lives better. So I decided to pursue a career in medicine.
Cardiology reminded me a lot of physics. It's also very practical in that there are a lot of things you can do to make people feel better. You can talk to the patient, look at the lab tests, the electrocardiogram (EKG), and the imaging studies, and put all that information together, and then go back to the patient and say, “Hey, I’m going to go back to the patient. ”
I have been interested in computers and computer science since I was a child. I don't have a formal engineering background, but I taught myself a lot of programming and coding, and over time I became interested in the intersection of technology and medicine. I remember when the Internet, and then smartphones, started changing the world, and I think it felt like the technology would soon have an impact on healthcare as well.
I am currently a cardiologist. As a research postdoctoral fellow at Massachusetts General Hospital and the Broad Institute, I have spent the last several years conducting research on how deep learning, a type of machine learning, can be used to examine large cardiac datasets and predict heart conditions. We have supported you. Risk of various cardiovascular diseases. Essentially, we're training computers to perform electrocardiograms and imaging tests in the same way that cardiologists do. Computers can often look at images and waveforms of the heart and interpret them better than we can. In theory, a cardiologist could sit down and do all the calculations that a computer does, but it would probably take years to obtain, say, a single electrocardiogram.
One of the projects I led About the risk of high blood pressure, or because high blood pressure is associated with the risk of other cardiovascular diseases, such as heart attack. We trained a deep learning model to assess a patient's risk of high blood pressure using millions of electrocardiograms from the Brigham Mass. General database dating back to the 1980s. They then looked at the health records attached to the electrocardiograms to see which patients had heart attacks after the electrocardiogram. We found that patients predicted by the model to be at high risk for high blood pressure were actually more likely to have a heart attack, and that we could calculate a patient's heart attack risk based on their high blood pressure risk. I also participated in similar studies on atrial fibrillation, coronary artery disease, and heart failure.
Our group is conducting this research By pairing clinical leaders with computer science experts. But in my opinion, if we're going to have a meaningful impact on how clinicians like me use deep learning to improve the health of our patients, we need to look beyond just the medical side of things. , we believe that the computational aspects also need to be truly understood. That's what led me to enroll in Harvard Chan's Health Data Science program. I wanted to gain knowledge, tools, and skills to better understand and improve my research.
One of the best parts of the program is the core curriculum.. The most memorable class for me was the statistical survey class that I had to take in the first semester. That was very basic for me. I, and many clinicians without formal training in public health, use statistics all the time. It is easy to perform regression using a computer. But it's one thing to enter numbers into a computer, it's another to really understand mathematics and understand the problems it can cause for yourself and your research. Taking that class taught me the theoretical foundations of statistics and helped me better understand the strengths and limitations of other people's work and, critically, my own work.
my capstone project had developed a deep learning model that was trained to diagnose mitral valve prolapse by examining echocardiograms (ultrasounds of the heart). I'm glad I was able to practice modeling by myself in a supportive environment. Santiago Romero-Brufau and Tamar Sofer provided great feedback.
Skills that were being developed in real time during the program This allowed them to train a deep learning model to automatically measure the size and function of the right side of the heart. This is important data that is not often captured because it takes a lot of time and effort from the clinician. I really enjoyed finishing the class with new coding skills that allow me to lead parts of research projects that are usually overseen by someone with a more passionate technical background.
After 2 years of study, I don't expect to reach the level of some of my collaborators who have PhDs and have been working on deep learning for years. But I can communicate with them in their native language and really understand their work and don't need them to translate or simplify things for me as a clinician.
After graduation, I will continue my clinical training to further specialize in cardiac electrophysiology, which deals with arrhythmia. Looking further ahead, she envisions a career as a physician-scientist. I would like to someday start my own lab focused on heart health and deep learning.
– Maya Brownstein
Photo: Kent Dayton
