Core curriculum · Module 1 of 3
Fundamentals of AI
Build the basic mental model first: what AI is, how models learn from data, how performance is measured, and why a model can fail outside the conditions in which it was developed.
- Core question: What is AI, and how does it work clinically?
- 5 lessons
- 15-20 min each
- No AI background needed
Lessons
Five short lessons on what AI is and how it works, starting from the basics. One made-up tool, PTX-Detect, a chest X-ray pneumothorax detector, runs through all five.
- 1.1
AI Around You
AvailableWhat AI is, how a learned model differs from a hand-written rule, the family tree of AI, machine learning, deep learning and generative AI, and where AI already shows up in medicine.
- 1.2
How Models Learn
AvailableThe training loop (guess, compare, adjust, repeat), supervised, unsupervised and reinforcement learning, neural networks, and overfitting.
- 1.3
Data & Labels
AvailableThe data radiology AI learns from, where labels and “ground truth” come from, training/validation/test splits, and data leakage.
- 1.4
Measuring Performance
AvailableThe 2×2 table, sensitivity and specificity, the accuracy trap, positive predictive value, and how the threshold trades misses for false alarms.
- 1.5
Limits & Failures
AvailableDataset shift, shortcut learning, hidden subgroups, and confident wrong answers, plus three habits for using any AI tool.
Where this module fits
Three modules, then the assessment. Research Lab and Leadership are optional tracks you can open at any time.