How Can Medical Schools Educate Students on Artificial Intelligence?

BUILDING THE MD/MS IN AI PROGRAM

The program’s development began roughly four years ago when Rodriguez was giving a report about where the medical school should invest and focus more of its efforts.

“The one area that I felt was particularly pressing was artificial intelligence and the increasing use of these technologies in science and medicine, and how I felt at some point, it was going to become a very important aspect of how we practice medicine,” he explained.

When these suggestions were brought to the school’s dean, Rodriguez and others were tasked with holding a retreat to discuss the current technological landscape and how UT Health San Antonio and UTSA could collaborate on efforts to invest more heavily in analytics, AI, and computer science.

“The university had decided to invest in the development of a school of data science, and they were going to invest heavily in artificial intelligence and hire new faculty specifically in that arena,” Rodriguez noted. “Once they hired several people in that core area, then we got together with them and started to vet how we would actually put together a training program to specifically allow our medical students to capitalize on [AI].”

From there, stakeholders had to meet regularly to determine next action steps, such as what would constitute the core requirements of such a training program, what would be a reasonable amount to teach medical students about these topics, and the logistics of how the program would be carried out across two campuses.

Rodriguez and his colleagues faced multiple obstacles while developing the MD/MS in AI program. One major challenge was how the degrees should be conferred.

“Initially, we were trying to do it as a joint degree program where one degree would be conferred by two institutions simultaneously,” Rodriguez stated. “Once we started that path, we were notified that the complexities of that made it so difficult that it would be better if we had one institution confer the degree and the other form a memorandum of understanding and agreement through which they would allow us to provide students with provisions such as a waiver for the GRE.”

He and his colleagues also had to navigate hurdles related to tuition costs, academic schedules, legal considerations and paperwork, how students would be vetted for participation in the program, and what types of training would be credited toward completion of the dual degree.

Rodriguez noted that overcoming these challenges was a tedious process that took over three years, but that the hard work eventually paid off.

“Now, [the program’s development team is] quite happy with the result,” he said. “Even more importantly, should we ever decide to do other dual degree programs, we have a really good template in place now that would allow this to go much faster.”

HOW THE PROGRAM WILL BENEFIT MEDICAL STUDENTS

When the idea to develop the MD/MS in AI program was first laid out, the prospect received a lukewarm response from many.

“I couldn't get people really excited about it, but I did my due diligence,” Rodriguez explained, describing how he sent out a questionnaire to all the medical students to gauge their interest in the program.

The students were very receptive to the idea, indicating that they thought being educated about AI in medicine was important enough for their careers that they would be interested to learn about the program and potentially consider taking on the extra training required.

“They were much more in tune with this than I think a lot of the older people in the institutions were,” Rodriguez stated. “They knew immediately how important this was going to be, and more than two-thirds said that they would very much like to hear more, and as much as a third would seriously consider taking an extra year.”

He underscored that the students’ willingness to even consider participating in the program was significant, as the extra year of required training would require a significant financial investment in terms of tuition fees and an additional year of potentially not earning income.

For medical students in a position to do so, making that sacrifice has the potential to pay off in terms of residency placements, Rodriguez noted.

“[Medical students] recognize that in this current environment, getting into a high-quality residency is not easy,” he said. “There is a limited number of slots, and there are more students in line for the slots than there are slots.”

The process is extremely competitive, and students who want to get into a good program must be exceedingly strong students. But not everyone can be at the top of the class, even if they perform well academically. This raises the question of how medical students can set themselves apart from their peers in other ways, Rodriguez emphasized.

Now, the program is launching with three students already enrolled at the time of this interview, and more have been in contact with Rodriguez to get more information about the program.

The current program structure involves students completing the first three years of their medical school training as normal. Between their third and fourth years, they take a leave of absence to prepare for their MS in AI training.

The leave of absence helps address the challenges of scheduling classes across two campuses, as UTSA runs on a semester-based schedule while medical schools typically use a block schedule, with each block being anywhere from four to 12 weeks long and based on rotations.

To align the schedules, medical students finish their third year of medical school in July, then start didactic courses at UTSA in August. These courses give them some background on AI, computer science, and data analytics.

When their year at UTSA ends in May, the students prepare to return formally to UT Health Science Center again in July. Currently, one of the courses being developed for students in the program will focus on the use of technologies to interface with the electronic medical record through an application programming interface (API), Rodriguez highlighted.

During their training, students will also be required to complete two additional courses, which are capstone projects to demonstrate their understanding and use of the skills they’ve acquired through the MD/MS in AI program. From there, each student must write a formal thesis and submit it for publication.

Upon completion of these requirements, the MD/MS in AI dual degree can be conferred.

The hope is that the program can prepare the next generation of doctors to be able to use technologies like AI effectively and safely to improve healthcare.

As the first program of its kind in the United States, the program may also provide a blueprint for other academic institutions that want to ensure that their medical students are educated about health AI.

INCORPORATING AI INTO MEDICAL SCHOOL CURRICULA

Rodriguez emphasized two different approaches that medical schools can take to incorporate AI into their curricula.

“One is to use [AI] as a tool to achieve medical education. So, in that case, schools might be utilizing it to help improve the way that students ingest content and get tested on it. And there are some novel ways to do that,” he said.

The other approach involves allowing students to take a much deeper dive into how these technologies are impacting medicine.

To do this, Rodriguez noted that medical schools must make efforts to establish necessary infrastructure, which is part of what the MD/MS in AI tries to do. By investing in and establishing infrastructure, schools can more easily adapt to challenges along the way. However, some medical schools are in a better position to take this on than others.

“If you look across the country, there are multiple medical schools that have access to engineering facilities, computer science departments within a university, or schools of data science and bioinformatics. Most of these have been set up as standalone institutions,” he indicated.

Such standalone institutions are not set up to train medical students to become doctors well-versed in AI. Rather, many of these institutions focus on conducting AI research to create tools for clinicians to use.

Rodriguez noted that this approach could be shortsighted as AI becomes increasingly integrated into healthcare.

“The paradigm shift here is expecting that medical students are going to be more than just end users. They're going to be the innovators. They're going to be the drivers,” he explained. “And it's really important that [clinicians] take that stance because if not, then we will be just the end users, and we will have the technology developers telling us how to take care of patients.”

He likened this to how payers have slowly gained more control over aspects of care like claims denials, impacting how clinicians practice medicine in ways that are not always in the best interests of the patient.

“I feel the same thing could happen with AI technologies if we are not careful,” he underscored. “It is possible that [clinicians] could be the ones being told by AI what to do and not driving the AI to empower us to be the most effective we can be.”

Rodriguez stated that clinicians, rather than technology experts, are likely to be much more in tune with health AI’s potential limitations and the direction in which these technologies should be developed.

“That's part of my drive for wanting to do this. I want the students that we graduate to not only know how to use the tools, but how to create the tools, to understand how we can best apply them, to understand where the best impact can be achieved for their particular area, and to develop them. We should be leading this; we should not be led.”