AI is changing healthcare and Harvard Medical School is following suit

The Harvard Medical School started offering a month-long introductory course on AI in healthcare for HST students—the first of its kind offered at a medical school.

By Veronica H. Paulus and Akshaya Ravi 

n her first course at Harvard Medical School this fall, Kayton E. “Katie” Rotenberg, an HST MD student, wasn’t taught anatomy or biochemistry.

Instead, she learned how to be a physician-scientist in a world that is beginning to depend on artificial intelligence for research and medicine.

The month-long introductory course on AI in healthcare is required of all students in the Harvard-MIT Program in Health Sciences and Technology track (HST)—an inter-institutional MD and PhD program between Harvard and MIT that focuses on translational medicine and engineering.

But the class—the first of its kind offered at a medical school—is just one of the ways HMS is adapting to the changing medical landscape and AI’s increasing role in healthcare.

HMS Dean for Medical Education Bernard S. Chang is one of several senior administrators pushing to integrate AI into the training curriculum at the school.

AI is being introduced in three broad categories, Chang said.

The first is as an educational tool. Chang pointed to HMS’ flipped classroom model—in which students learn content on their own, practice cases during class, and take quizzes on the material afterwards—as one case already benefiting from AI.

“AI can take all of these however many hundreds of responses each day. And the pilot tool allows faculty to determine the level of granularity that they want to get the feedback,” Chang said.

Students and faculty across the university have already incorporated AI into the classroom.

In 2023, Computer Science 50: “Introduction to Computer Science” integrated artificial intelligence via a “CS50 bot” that finds bugs and offers feedback on projects. Earlier this year, Economics 50A: “Using Big Data to Solve Economic and Social Problems” introduced a similar chatbot, called “ec50.ai,” which answers students’ questions.

In August, the university also began providing Harvard College students with access to OpenAI’s ChatGPT Edu as part of a broader set of initiatives to probe the use of AI in teaching and research.

Chang said HMS is also trying to develop an AI “tutor bot,” aiming to roll out the software by 2025.

The second two categories, Chang said, prepare MD and MD/PhD students for how AI may impact their future clinics and laboratories.

“As the world of healthcare changes around us, we also need to evolve to respond to that, to prepare our students the best way,” Chang said.

Chang highlighted the ability of AI to automate routine tasks for clinicians, allowing medical schools to focus on teaching the intangible skills of being a physician and spend more time on “how to apply the more difficult and higher level analyses to information.”

In response to the growing presence of AI in healthcare and education, HMS has adjusted its course offerings.

The school piloted HT 16: “Artificial Intelligence in Healthcare I” in August and September, HMS’s Fall 1 term, and its counterpart, HT 18: “Artificial Intelligence in Healthcare II,” will run in January.

The course—now mandatory for the 30 HST students entering each year’s medical class of about 160—is taught by Collin M. Stultz, director of the HST program on the MIT side.

“It covered AI, machine learning, LLMs, particularly with a focus on ways in which AI is already going to emerge into healthcare,” Chang said.

Chang also emphasized that the class strived to be an interactive introduction to AI, engaging students through a project-based model.

“We’re trying to teach not just knowledge but skills,” Chang said. “Part of it is, yes, teaching about how AI works and how it is present in healthcare, and how it might revolutionize healthcare. And part of it is, here’s some problem sets. Do these coding examples. Write a paper with your group, present it, get feedback.”

Chang praised HMS Dean George Q. Daley, HST MD '91, for the push towards incorporating AI, praising his “investment” in making HMS a “worldwide leader in AI, in our educational programs, in our research enterprise.

To Rotenberg, the AI course provides students with two main takeaways.

The first is understanding new tools and methods at the frontier of medicine: “To see a new study, a new paper that comes across your desk, a new diagnostic tool when you are a future doctor,” Rotenberg said, and to “ask questions about if it really is a reliable piece of technology.”

Natalie E. Fulton—a first-year medical student in the HST program—said the course taught her to appreciate how AI tools can make better predictions than physicians, and how they can improve the risk stratification of specific patients.

“Navigating their use in clinical practice is the biggest thing I got from the class,” Fulton said.

The second goal of the course, according to Rotenberg, is to arm students with the knowledge needed to conduct research involving AI.

“You don’t have to be the person doing all of the AI coding, but I think the goal of this class is to get you up to speed so that you can have a strong understanding of what types of AI tools you would want to be using to ask those questions,” Rotenberg said.

“You can then collaborate with another researcher that has more of a computational background to solve those challenges,” she added.

Though the course is only currently mandatory for a fraction of the HMS class, Chang said he hopes more students—including those from the more traditional pathways program—will engage with the course.

“I think the rest of the medical students are definitely going to benefit from some formal introduction to AI,” Chang added.

The introduction of AI in medicine courses marks one change in a broader cultural and programmatic shift at the Medical School.

HMS offers students “other opportunities to go into greater depth,” according to Chan.

The Department of Biomedical Informatics has started offering an Artificial Intelligence in Medicine PhD program, which encompasses coursework in medical AI, collaborations with AI researchers, and clinical rotations at HMS-affiliated hospitals.

The Office of External Education also offers an intensive continuing education course, dubbed “AI in Clinical Medicine,” that seeks to offer additional training in emerging medical technologies for already-licensed physicians from other institutions.

Last year, Daley also established the Dean Innovation Awards for the Use of Artificial Intelligence in Education, Research, and Administration, which provide selected projects with grants of up to $100,000 for education, further diversity, healthy aging, and more.

“The grantees are the educational leaders who teach our medical students, who are trying out various projects,” Chang said. “So we have really taken the intentional lead on this, which I’m very happy about.”

Chang said the integration of AI into healthcare is following the same arc as the arrival of Google.

“I remember when patients used to say, ‘Ah, I hope you’re not one of those doctors that uses Google.’ And now, can you imagine a patient saying that?” Chang said.

“In a couple of years, most patients would very much want to see physicians who were using the latest technology to help them in diagnosis and treatment—and that will be AI,” he added.

—Staff writer Veronica H. Paulus can be reached at veronica.paulus [at] thecrimson.com (veronica[dot]paulus[at]thecrimson[dot]com). Follow her on X @VeronicaHPaulus.

—Staff writer Akshaya Ravi can be reached at akshaya.ravi [at] thecrimson.com (akshaya[dot]ravi[at]thecrimson[dot]com). Follow her on X @akshayaravi22.

*Originally published in the Harvard Crimson.