AI in Academic Medicine

Good morning!

Today’s VC Notes is the second in a two-parter, which I didn’t share was going to be the case in last week’s column, other than the footnote. I suggested you keep last week’s VC Notes handy because it was going to come back up, and you may have already figured out by the title of today’s column how the two are connected.

I did not write last week’s column and nobody on my team did. It was written by artificial intelligence. Yep, AI gets credit for authoring a VC Notes – which was a one-time thing, I promise.

AI in health care is a hot topic, but not something that I’ve engaged with much, so I wanted to give it a spin. An AI system was used to analyze my writing style by reviewing past VC Notes and produce a 500-word column on the importance of the missions of patient care, health professions education and biomedical research at an academic medical center. What you read in last week’s VC Notes is exactly what AI produced. I couldn’t believe what I read because it was almost how I would have written on that topic. The entire process took less than a minute. The power of that AI system to quickly produce something so accurate to UMMC-related information (with a few exceptions) and like my typical style was both a little startling and exciting. 

Use of artificial intelligence is on the rise, including by academic health science centers, health systems, hospitals and other health care entities worldwide. Technology, when used properly with guardrails, can be useful in many health care-related areas. I wanted to give a brief look into how we are beginning to use AI here at the Medical Center. The content and comments I received from Medical Center personnel who are leading our AI efforts give an encouraging outlook for technology’s future, but also show why it’s prudent we continue to move forward with caution.

A new AI-use policy is under development, led by Sandeep Rustagi, data analytics officer, which will govern the “identification, development and utilization of AI technologies” and create an AI Governance Committee. In the policy, the Medical Center’s definition of artificial intelligence is “… any computer application or technology that can perform tasks which typically require human intelligence. These technologies learn from their experience and aim to perform specific functions autonomously.” The policy will establish standards and guidelines to ensure that any use of AI at UMMC upholds the highest ethical standards and is compliant with regulatory requirements.

Within our patient care enterprise, we’ve incorporated AI systems into several areas. Here are just a few:

• Generative AI, which is AI that uses models and other info to create content, like what we did to produce last week’s VC Notes, has been added to Epic to save a clinician’s time on some routine tasks like replies to patients and getting through the many messages delivered to the InBasket.
• A few departments are using an AI-driven, ambient listening system that significantly cuts down on a clinician’s post-visit follow-up work.
• An AI audio and visual platform is being piloted that could predict a patient fall or a pressure injury. Anything that can help prevent a patient harm instance is a big win.
• A predictive modeling system is under development for the Emergency Department that will be able to predict admissions over the next 24 hours. This would allow ED leadership to more effectively use resources and manage capacity.

AI is also being used within our academic mission area. Here are some examples:

• The biostatistics and data science program in the Bower School of Population Health trains all MS and PhD students to competently use machine learning through AI algorithms and tools to extract information from massive troves of data.
• Clinical training has been enhanced by AI-powered virtual reality systems, which provides realistic hands-on experiences.
• AI tools are being used to optimize development of course curriculums and training materials.
• We are aware of and support students who are using AI to help develop effective study schedules, generate notes from lectures and develop their own test-prep questions.

Below are a few instances of how our researchers are evaluating the effectiveness and merits of implementing AI-backed systems:

• For clinical trials, we have been implementing a software system that is able to mine structured and unstructured data to accelerate patient matching.
• Researchers are examining use of AI-enabled medical devices to enhance accuracy in radiological interpretation.
• We are looking at possibilities of using AI methods in evaluating ophthalmic screening data in hopes of increasing early detection of eye diseases and major chronic diseases like hypertension.

Artificial intelligence is not just the future; it’s already here and has been for a while. Its integration into academic medicine is relatively new, but the rise in its use is steep and shows no sign of leveling off. Our institution, like all others, must consider how use of this technology can benefit us, our patients, our students, our research programs and the communities we serve. As AI gets more accurate and reliable, we’ll apply it in other ways, and it will become another powerful, useful tool that can (virtually) aid in building A Healthier Mississippi.