R.Ready or not, healthcare is undergoing a massive transformation driven by artificial intelligence. But medical school has barely started teaching him about AI and machine learning. flawed algorithm When biased Decision support system.
Erkin Ötleş, a machine learning researcher who is about to complete his medical degree and PhD, said: .D. at the University of Michigan. “If you don’t have a set of grounding knowledge about how these things work, you’re at a disadvantage.”
In a recent commentary published in cell report medicine, Ötleş, and a group of doctors and educators at the University of Michigan called on medical educators to make AI a core concept in medical training at their universities, rather than an afterthought. They emphasize the spiral his curriculum idea that students first learn the key points of AI in medicine and come back again and again as they acquire more specialized skills.
advertisement
But doing so won’t be easy, said co-author Jim Wooliscroft, former dean of the Michigan Medical College. are hindered, and faculty themselves may not yet have the expertise to teach a new generation of physicians. In an interview with STAT, this student and educator elaborated on how the medical educator began the process of revamping his AI training.
What is the current state of medical education in artificial intelligence?
advertisement
Jim Woollycroft: Medical education programs have not evolved at all in nature. There is a little tweaking, but no seismic changes that need to happen.
Elkin Etresh: I am totally unprofessional when it comes to artificial intelligence and machine learning. I often see people who are interested take the time to do a master’s degree, or a combined MD/Ph.D like I do. Otherwise people may be exposed through studies that can be done as electives. As a student, you still have to go and guide your own research and provide yourself with information.
Is that enough?Or should it be AI Will it be incorporated into the general medical curriculum?
Etreche: Artificial intelligence and machine learning are so pervasive in our daily work that everyone should have at least some basic level of understanding in order to evaluate the tools we use. You don’t have to be an expert, you don’t have to develop something like this, but you need to be able to say “I don’t think this will work” and call the developer and say. , “I think there’s a problem.” Now that we’re behind the 8-ball, we need to start teaching people immediately.
Woolly Croft: Medical students don’t know these things, but they need to understand something as basic as pharmacology and physiology. Already, machine learning algorithms and more generally AI are essentially everywhere.
One of the real problems is that our faculty are unaware of this. Pre-Corona, I gave a lecture on machine learning, and people were saying, “Why is this important?” They didn’t even know that the university hospital at the time had eight programs running continuously in the background to monitor patients’ physiological variables. Teachers do not have the expertise to teach it.
AI teeth What strategies are used if they are taught in medical schools?
Etreche: The emphasis is on a particular technology or tool, but it shouldn’t be. All of these techniques make extensive use of Python programming and should be taught to medical students. And, as you know, I love Python programming, but I don’t think all my medical school colleagues should know how to program in Python.
Woolly Croft: Some of the examples that seem to work are really contextual, like radiology. That’s good. But what we need is to get the earlier students to understand some basic questions to ask. What is the database and what care was taken to ensure that the data used to build the algorithm was clean? What were the gold standards utilized? All of these are broadly applicable I have a question.
So what would be a more successful framework for AI medical education?
Etreche: Due to time constraints, teaching the fundamental concepts of AI and machine learning should be prioritized. You have to understand what is most important and use that as a basis. So once you have that foundation, you can continue to refer to it over time and grow it or relate it to other concepts as needed.
Woolly Croft: One thing that hasn’t been done, as far as I know, is this spiral curriculum concept. As students move into the clinical realm they come back again and again. So when you’re seeing a radiologist, you can ask: So what was this interpretation of the mammogram based on? Was it? Oh, I see. Well, here in Michigan there are a lot of people from the Middle East. So does this apply to this collective? Once you get down to all these different things you have a foundation that you can plug in these specific examples and fill in the flesh of the bones that have been laid .
W.What are the biggest hurdles in introducing this kind of change to the curriculum?
Woolly Croft: Essentially, most medical colleges have not changed their faculty structures to reflect changes in the underlying science of medical practice. We have this structure, this legacy that leads to enormous inertia. This is mainly because all sorts of things are involved, such as budgets and personnel.
Another real problem is making decisions about the curriculum. Students don’t see it as valuable. It needs to be integrated, and that requires really changing many fundamental things that faculty have been doing.
W.There are some first steps to addressing these barriers, but who should take the lead?
Etreche: It should be led by a doctor. It’s probably the medical center of a university, where I have colleagues in engineering, computer science, informatics, and learning health systems. You can get resources that can be combined quickly. That’s what we need and we need that speed now.
And what about medical schools that don’t have such resources?
Etreche: We try to push this as a conversation. We believe that having a foothold of focusing on this basic knowledge and then continually coming back is an important way to lay these out. People may think we’re completely off base, but I hope they agree that it’s important that we move, and that it’s important that we move quickly in this area. As part of that, I would like everyone to think about how we can share resources. When you build your curriculum you share it, when you build your tools you share it. That way, you can get to teaching and learning instead of wasting time recreating.
Woolly Croft: it happens. I’m just worried that it will happen sooner. This is the same as other technological innovations over decades and even centuries. When a new technology is introduced, it is often introduced from outside biology, applied to biological problems, a field is born, a department is created. I think it’s important to move forward. This cannot happen because the patient will die.
This article is part of a series investigating the use of artificial intelligence in healthcare and the practice of patient data exchange and analysis. supported by funds from Gordon and Betty Moore Foundation.
