This interactive, asynchronous blog-based learning experience introduces students to the intersection of LLMs and healthcare applications. The 4-part series guide students through LLM fundamentals, ethical considerations, and hands-on projects. The experience combines foundational knowledge with hands-on skill development through downloadable code tutorials. This is a self-paced learning experience which can be done at any time and any place as long as you have a laptop and internet access.

The intended audience of the blog is those who have programming experience in Python and want to learn the technical aspects of training and using LLMs with medical data. The content of the series should introduce students to the basic concepts that are useful for starting research in healthcare natural language processing (NLP).

Learning Objectives #

After engagement with this learning experience, individuals should be able to:

1. Understand the basics of what LLMs are and identify unique challenges in healthcare AI
2. Practice prompt engineering techniques for medical question answering task using the HuggingFace library
3. Execute fine-tuning workflow for developing a medical language model, including data preprocessing, model selection, hyperparameter configuration, and evaluation using HuggingFace library

How to navigate the blog? #

Start from Part 1 and then move through each part in order. Take your time going through the materials. You can find all the blog posts in posts.

1. Foundations of LLMs and Healthcare AI: You will learn LLM basics and explore common healthcare tasks and datasets. The blog explains how healthcare applications differ from general natural language processing tasks. You will complete a hands-on exercise exploring sample healthcare datasets to identify differences in structure, language, and complexity.
2. Ethics, Limitations, and Data Handling: This section covers patient privacy requirements, bias and fairness issues, and AI limitations in healthcare. You will analyze a case study of an AI healthcare failure to identify what went wrong and how it could have been prevented, developing critical thinking skills
3. Prompting LLM: You will learn to prompt LLMs for medical question answering and discover techniques to refine and test prompts for improved performance. Downloadable starter code enables hands-on practice.
4. Fine-tuning LLM: The final part walks you through the complete process of adapting LLMs for medical information extraction tasks, with step-by-step code implementation and practical exercises. This also includes downloadable base code.

For additional resources and links, please check out library to learn more about the topics we cover in each blog post.

Contribute #

You can contribute to this project and the blog by suggesting any new topics to cover or any cool resources to add to the library!

If you have any content suggestions or edits for this blog, please feel free to create an issue on the GitHub repo. Thanks!