Foundations of AI for future physicians: A practical, accessible curriculum.

Abstract

What was the educational challenge?: The integration of machine learning (ML) and large language models (LLMs) into healthcare is transforming diagnostics, patient care, and administrative workflows. However, most clinicians lack the foundational knowledge to critically engage with these tools, creating risks of overreliance and missed oversight. Just as understanding computed tomography (CT) physics became essential for its safe application, clinicians must acquire basic AI literacy. Practical AI education remains absent from most medical curricula.

What was the solution?: We propose a modular curriculum using Colab notebooks to teach foundational AI concepts. Colab's free, cloud-based, and interactive environment makes it accessible and engaging, even for non-data scientists. This hands-on approach emphasizes practical applications, enabling learners to explore datasets, build ML models, and interact with locally run LLMs, fostering critical engagement with AI tools.

How was the solution implemented?: The curriculum consists of five interconnected modules: introduction to data science, exploring datasets, predictive modeling, advanced ML techniques and imaging, and working with LLMs. Designed to integrate into medical school data science threads, the curriculum provides structured, progressive learning tailored to clinical contexts.

What lessons were learned?: Global accessibility, hands-on engagement, and modular design make this approach adaptable across diverse settings. Emphasizing ethical considerations and local relevance enhances its impact.

What are the next steps?: The next step is to integrate the Colab notebook-based curriculum into the authors' medical school data science thread. To support broader adoption, adaptable teaching guides will be developed, enabling implementation at other medical schools, including those in low-resource settings, while leveraging Colab's accessibility for regional customization.