AI in medical school: the impact on tomorrow’s doctors

Introduction

In 2014 I was preparing for my medical school interview. I’d gone over the common questions countless times. What do you know about our curriculum? What personal qualities make you well-suited to medicine? Etc etc 

One question that hadn’t crossed my mind was “How do you think AI will impact your education or how you practice as a doctor?”

And nor should it. This was the year before OpenAI was founded and 8 years before the ‘ChatGPT’ revolution, no layperson had even heard of DeepMind. In 2014, if you asked what machine learning was, you would have gotten an answer along the lines of “something in a sci-fi movie”. Today, medical students find themselves in a sea of new technology, new terminology and uncertainty around how AI will impact their future clinical practice. 

Stats, stats, stats 

The statistics for how quickly AI has been adopted are mind-blowing. For example, consider how long it took the internet and leading social media companies to reach 1 million users. The World Wide Web took 7 years, Instagram 2.5 years, TikTok 9 months – it took ChatGPT a mere 2 months. 

This widespread popularity and adoption is mirrored in medical schools across the UK. In a study of almost 500 students across 19 medical schools, 88% believe that AI will play an important role in healthcare. Additionally, 78% agreed that all medical students should receive training in AI as part of their medical degree. 

Source: Techopedia

Future UK medical curriculum

Traditionally, medical education has operated on a backwards working model, whereby students learn established practices and methodologies. The trickling down of new information can be slow. There might be the odd slide on “novel drug x” in lectures, but this is never formally tested therefore, the information is not reinforced. If this were to remain the case we would have a lag of AI literate clinicians.

A forwards working model 

Medical students need to be exposed to new information and technologies as they happen. As the clinical safety and effectiveness of AI is proven, these teachings need to be quickly incorporated  within a new curriculum.  This approach should reduce (if not eliminate) the lag time of AI literate clinicians. Not only should tomorrow’s doctor have a robust working knowledge of AI, they should also be poised to leverage these innovations to improve patient care.

One example of AI integration within undergraduate medicine is a pilot study from the University of British Columbia. The university ran workshops over a five-week period – focusing on large data analysis, clinical implications of AI and how to collaborate with engineers in developing AI tools. Results included improved knowledge around AI (addressing prior concerns of loss of roles), programming proficiency and understanding how future clinical roles will be adapted. However, they also acknowledged a large variation in prior knowledge and issues with information retention.

What would a forward AI curriculum look like?

Student selected components in clinical AI

  • These have always been a good avenue to explore interests outside of the ‘rigid’ curriculum. Mine were centered around pharmacology and anatomy which had a huge impact on my future career (I’ve since intercalated in clinical pharmacology and I’m a core surgical trainee).

Ethics lectures on AI 

  • As AI becomes integrated into our daily lives, ethical considerations of how and when to use AI should also be at the forefront of decision-making. 
  • How should AI be regulated? Who should be regulating it? 
  • Debates such as: “If AI made your diagnosis of patient X more accurate, do you have a moral obligation to use it?”

Expansion of the multidisciplinary team (MDT) 

  • A shift away from the standard case studies and interacting with nursing and pharmacy students
  • Case studies to be focused around developing core AI competencies and integration with students from computer sciences backgrounds

Intercalated degrees

  • If you’re fortunate enough to be able to intercalate this is something I would highly recommend. 
  • There’s no specific ‘intercalated degree’ for AI however there are modules within existing courses e.g. biomedical engineering. Another option would be to find an AI-focused supervisor within your intercalated degree. 
  • The best way would be a masters and there are multiple courses available (links at the bottom).

Exams 

  • Part A) Anatomy of the abdomen, Part B) Anatomy of the large language model
  • I also imagine that using AI technology will be eventually integrated into the OSCE examinations, in the same way that using a sat nav is part of the driving test.

A TORTUS vision for the future

At TORTUS our mission is to eliminate human error in medicine through clinicians co-working with AI. We believe that this mission will take us to the very beginning of a clinician’s journey. We envisage our technology being used within medical education: an aide-mémoire for study, helping students to improve their OSCE performance, teaching correct documentation practices – AI will be central to the training of the next wave of clinicians. This is just the beginning, there’s a lot more to come.    

Are you a medical student and don’t know where to start? Follow us on LinkedIn

  • Our last event was centred around AI in primary care. We have more coming up. Watch out for these and for our posts on projects medical students can help us out with. 

Other (free) resources for medical students: 

Google’s Introduction to Generative AI Learning Path

AI E-learning for clinical research delivered by the NIHR 

Widely popular: AI For Everyone on coursera (by Andrew Ng)

Dev & Doc podcast 

European union act on regulating AI

Medical AI related degrees