A Cardiology Fellow’s Experience Studying Artificial Intelligence

For this blog, I am excited to have my co-fellow and husband, Tony Prisco, discuss how he became interested in studying artificial intelligence in cardiovascular diseases and some tips for others who are potentially interested in this field!

My name is Tony Prisco and I am a PGY-5 at the University of Minnesota in the Physician-Scientist Training Program. I am pursuing my fellowship training in cardiovascular diseases. I finished my PhD in 2014 focusing on mechanisms of angiogenesis induced by adult stem cells. Since then my scientific interests have transformed to mathematical studies, including fluid mechanics and artificial intelligence. The majority of my work since my PhD has focused on blood flow mechanisms in patients with mechanical circulatory support (ventricular assist devices and VA-ECMO).  I had been hearing about various forms of artificial intelligence since at least 2014 but did not realize the potential clinical applications until I started my residency in 2016.

When I started my dedicated research time back in 2019, I initially had planned on continuing studies that used computational fluid dynamics to better understand the mechanisms of cardiovascular diseases and the device-human interface. A month into my research, due to the growth of the field of Artificial Intelligence (AI), I watched several YouTube videos describing the math behind “deep-learning,” which is a form of AI. Interestingly, most of these videos were targeted towards the tinkering hobbyist working to do interesting home projects. They focused on getting started with a very low budget (i.e. free) and minimal computing background. Tutorials were primarily using high-level programming languages such as Python or MATLAB. After a day of struggling, I was then able to successfully recreate an example to train a neural network (type of deep-learning technology) that could identify the numbers 0 – 9 within an image.

I discussed this with my clinic mentor at the time, and after a few weeks we came up with a useful clinical project that I have spent the last 18 months working on. In doing this, I have learned 3 important lessons regarding using artificial intelligence in clinical cardiology:

  1. Artificial intelligence is a great technology to answer a simple question, for example—is a study normal or abnormal? Complex questions, i.e.— “Should I list this patient for a heart transplant?” not so much.
  2. Artificial intelligence will most likely improve patient care by helping physicians to interpret clinical data faster—especially areas where a significant amount of data exists (i.e. imaging and ECGs). Most likely, deep learning algorithms will help to give a “preliminary” interpretation of a piece of clinical data that will ultimately be analyzed by a cardiologist.
  3. Artificial intelligence ultimately is an analysis tool. It does not make up for having a good experimental design and/or following the scientific method.

The underlying mathematical principles of artificial intelligence have been around for at least 50 years. It is only recently that we have had the computational power to apply those principles to clinical data sets. I do not have a background in computer science—but I believe with the resources available online including free courses, tutorials, and software—the barrier to entry is low enough that anyone with enough intellectual curiosity can be up and running within a few weeks. Resources I would recommend looking at for those interested in starting out in this field are:

Software—Python (https://www.python.org/) is free for all. The majority of code available online and tutorials are done in Python. MATLAB is another option and for those at most major universities, this will be free as well. I use MATLAB primarily because it works better with the supercomputer we have at the University of Minnesota, but in most circumstances, this will not be necessary.

Tutorials—there are many available online, including full courses. I went through the lectures of Stanford’s CS221 on YouTube (primarily taught in Python). MATLAB has a few dozen examples on their website as well.


“The views, opinions and positions expressed within this blog are those of the author(s) alone and do not represent those of the American Heart Association. The accuracy, completeness and validity of any statements made within this article are not guaranteed. We accept no liability for any errors, omissions or representations. The copyright of this content belongs to the author and any liability with regards to infringement of intellectual property rights remains with them. The Early Career Voice blog is not intended to provide medical advice or treatment. Only your healthcare provider can provide that. The American Heart Association recommends that you consult your healthcare provider regarding your personal health matters. If you think you are having a heart attack, stroke or another emergency, please call 911 immediately.”