Instructor: David Schuff, Section 003

AI Cardiologist Aces Its First Medical Exam

Rima Arnaout, an assistant professor and practicing cardiologist at UC San Francisco, created a neural network that outperformed human cardiologists in a task involving heart scans. She does not think the AI she created is ready to replace human cardiologists yet, but it was easily able to complete the first step in what a cardiologist does when evaluating an echocardiogram. The AI only performed this first step in the analysis of a heart image and the making of a diagnosis; however, Arnaout is now working to improve this technology so it can take the next steps to identify different diseases and heart problems. Experimental artificial intelligence systems are making rapid progress in the medical realm, so there could be major changes in the medical field in the near future. Despite the advances being made with AI in medicine, most people still want to see a human doctor when they have a health condition. Do you think we will ever get to a point where the majority of the population is comfortable being diagnosed by a machine and not a doctor? Will AI decrease job opportunities in the medical field in the near future?


3 Responses to AI Cardiologist Aces Its First Medical Exam

  • Gabrielle,

    Interesting topic relating healthcare to AI. I think that despite the advances that have been made in medical technology, most people including myself will still want to be diagnosed by doctors. I think this technology is great for helping doctors analyze their patients, however, the decisions should ultimately be made by licensed practitioners. I think that this technology could potentially decrease jobs in the medical field, however I do think that this technology will open up different types of opportunities for doctors, like enabling them to learn and utilize new skills.

  • Hi Gabrielle – I think the healthcare space is a really interesting and exciting opportunity for AI applications. As for patients being comfortable being diagnosed by a machine, I think that is going to take a while for the feeling of security to come about because current AI applications have a hard time explaining how they attain the solution they’ve determined. When given the right data, AI is able to give a diagnosis of the patient issue, but in some cases, like with IBM Watson, they cannot explain how they attained that diagnosis. Baba Guda, a researcher at University of Nebraska, sums it up pretty well when he states “There is no way to validate what we’re getting from IBM is accurate unless we test the real patients in an experiment.”. In all, I think patient confidence is a long way off. What needs to come first is doctor confidence.


  • Hi Gabrielle! This is such an interesting use of AI and a really exciting outcome. I agree with Noah and Tessa concerning the likelihood that the general population will be opposed to receiving a diagnosis from machines and I don’t think we’re anywhere near having them completely replace physicians. However, this experiment does provide some insight into the application of AI in healthcare in the long term. The sheer amount of information and speed with which it’s processed by a machine is especially useful in this field and in providing diagnosis. However, as the experiment shows, the technology is not quite there. In addition, there will be ethical and regulatory obstacles that will prevent widespread use of this technology as a primary source for diagnosis in the near future.

Leave a Reply

Your email address will not be published. Required fields are marked *

Where and when do we meet?
Alter Hall 603
12:30-1:50 Tuesdays and Thursdays
Office Hours
Tuesdays and Thursdays, 2:00-3:00 PM