AI Talk: American Medical Informatics Association – Annual Symposium 2021 (part one)

November 12, 2021 / By V. “Juggy” Jagannathan, PhD

This week and next, I will be summarizing my observations from virtually attending the American Medical Informatics Association (AMIA) Annual Symposium 2021.

 

I personally have a long history with AMIA. I attended my first symposium in 1995, where I presented as well. Back then it was called “Symposium on Computer Applications in Medical Care.” That old title still captures the essence of what this symposium is about. To me this event is a leading indicator, like the stock market, as to what applications are coming down the pike for real use. The conference also marks a personal milestone: I started blogging exactly five years ago—my first post summarized the events of the 2016 AMIA symposium!

Attending a conference virtually has its drawbacks. You don’t get to swap stories with your fellow researchers or interact directly with them for insights. However, being virtual does allow you to skim through recorded sessions to see what to pay attention to.

Opening and closing keynotes at the AMIA Annual Symposium

Dr. Eric Topol of Scripps Health provided the opening keynote. He is a visionary and wrote a best seller Deep Medicine detailing the impact of deep learning in medicine a few years ago. He ran through a hundred slides with very rich content in 45 minutes. A few highlights from his presentation:

  • Precision medicine should not be our goal. Precision, when applied to targets, simply means a grouping that is clustered closely together. In medicine, you not only want precision, but accuracy as well. All diagnoses need to be on the mark—accurate and precise.
  • Deep neural nets are forging solutions no humans had the ability to perceive. For instance, it is now possible to diagnose a range of conditions from retinal images—which was not possible a few years ago. The implications? A simple smartphone can be an early warning indicator of a variety of conditions. Kidney disease, diabetes, heart conditions, Alzheimer’s and more—can all be diagnosed using retinal images!
  • Deep learning is going to transform the practice of medicine—radiologists, pathologists, dermatologists, cardiologists, psychiatrists, gastroenterologists, oncologists, geneticists, palliative care and every physician out there.
  • Keyboards will be history, implying most interactions will be through voice recognition technology.
  • Ambient artificial intelligence (AI) and intelligent assistants will be everywhere. AI will augment everything a physician does but will not replace physicians themselves.
  • Regulations need to keep up with the AI advice being provided by researchers and ensure such advice is up to date and relevant. Proper and continuous evaluations of AI solutions need to be prioritized.
  • He mentioned the development of “digital twins” (which I covered in an earlier blog) for charting the course of cancer in patients.
  • “It takes a planet” – Dr. Topol emphasized the importance of collaborating with China and other countries on the development of new solutions.

If Dr. Topol’s presentation was a window into the future, the closing keynote Dr. Irene Dankwa-Mullan’s focus was on the here and now: health equity and racial justice. She is currently with IBM Watson, but traced her roots back to a small village in Ghana. Earlier this year she edited a book titled The Science of Health Disparities Research while at National Institute of Health (NIH). Her main message? We must infuse AI applications with values aimed at eliminating disparities, make sure data representations are robust and representative, avoid algorithmic bias and promote and sustain a social mission. You can hardly ignore or quibble with her advice!

I am always looking for feedback and if you would like me to cover a story, please let me know! Leave me a comment below or ask a question on my blogger profile page.

V. “Juggy” Jagannathan, PhD, is Director of Research for 3M M*Modal and is an AI Evangelist with four decades of experience in AI and Computer Science research.