Courses
Information about current and past courses taught by Dr. Jimeng Sun.
Instructor: Prof. Jimeng Sun
Contact: jimeng@illinois.edu
Platform: Zoom, Piazza, Discord
This graduate-level course explores deep learning (DL) methods and their applications in healthcare, combining theory, hands-on programming, and a research-oriented group project. Students will learn foundational and advanced DL models, apply them to medical data, and work on real-world healthcare problems using the pyhealth package.
By the end of the course, students will be able to:
- • Implement DL models: DNNs, CNNs, RNNs, autoencoders, attention models, GNNs, generative models.
- • Apply DL to clinical tasks: prediction, phenotyping, treatment recommendation, NLP, imaging.
- • Work with real medical data in PyTorch and Jupyter.
- • Use pyhealth for processing EHR data and model development.
- • Reproduce and extend recent DL healthcare research.
Video Lectures & Weekly Readings
Covering both theory and applications.
Labs
Self-guided exercises on DL basics and pyhealth usage.
Homeworks (5)
Programming assignments with auto-grading.
Quizzes
Weekly, non-graded self-checks.
Final Project
A team-based (or solo) reproduction and extension of a published DL for healthcare paper, including:
- • Proposal
- • Draft + peer reviews
- • Final report
- • 5-min video presentation
- • Code repo
Weeks 1–12:
Lectures, labs, homeworks
Weeks 13–15:
Final project execution and submission
Key Dates:
- • HW1–HW5 due between Feb–Apr
- • Group formation: Mar 3
- • Proposal: Mar 30
- • Draft: Apr 20
- • Final: May 7
Late submission penalties:
- • Homework: −10% per day
- • Project: −50% if 1 day late, 0% if ≥2 days late
- • Course Textbook (in weekly PDF chapters)
- • Intro to Deep Learning for Healthcare – Xiao & Sun
- • ML for Drug Discovery – Fu, Xiao, Sun
- • Relevant research papers provided
- • Community: Piazza & Discord
Prerequisites:
- • Python programming (required)
- • Linear algebra, calculus, Linux/cloud familiarity (recommended)
- • No healthcare background needed
🤖 Generative AI Policy:
Use of LLMs (e.g., ChatGPT, Claude) is allowed and encouraged for the final project, but must validate and understand all outputs.