WELCOME & ROADMAP
What participants should expect and how today's materials will support other hands-on sessions during the summer school.

































Two community workshops the weekend before, then a full academic week. Summer School into Symposium into Datathon. Pick the pieces that matter to you, or do the whole run.
SAT · JUL 25
Hands-on with clinical LLMs. Fine-tuning, evals, safety.
SUN · JUL 26
AI & STEM program for high schoolers. This year's edition is a robotics day - build, code, and team up with our research lab.
MON 27 → WED 29 · + SUN PRIMER
Three days of lectures and labs on model evaluation, drift detection, and post-deployment monitoring. Optional Sunday primer for newcomers - taught by MinJae Woo.
THU · JUL 30
Keynotes, panels, lab showcase, and an after-party reception to close out the day.
FRI JUL 31 → AUG 2
Three days of building. We pair you with a balanced team - clinical experts, data scientists, and researchers - and you ship an evaluation worth publishing.
SUN · JUL 26
What participants should expect and how today's materials will support other hands-on sessions during the summer school.
A practical framework for evaluating predictive models using XGBoost as the running example. Learn how to assess discrimination, calibration, subgroup performance, explainability, and clinical utility while avoiding common evaluation mistakes.
How do evaluation principles evolve beyond traditional machine learning? Compare evaluation strategies for tabular prediction, computer vision, and vision-language models, highlighting which concepts remain universal and which metrics are task-specific.
Work through real-world evaluation scenarios drawn from published studies and peer review. Participants will critique models, identify common mistakes, interpret SHAP values, assess subgroup performance, and develop an evaluation plan for a supervised learning model using a structured checklist.
Discussion of evaluation strategies, common challenges, and how today's concepts will be applied throughout the Datathon.
A few of the organizations fueling the 2026 Health AI Datathon. Research labs, clinical partners, and platforms writing the next chapter of AI in medicine.
Become a sponsor



You don't get a score. You get a trajectory. Track how a deployed model degrades across sites, scanners, and populations, then build the monitor that catches it first.
dedicated server per team. Balanced teams curated by us.
AI mentors on the floor
Don't bring your own. Every team is 5–7 people, hand-balanced across clinical experts, data scientists, and researchers. Different lenses, sharper outcomes.
breakfast + lunch every day, all week
Symposium afterparty. Meet the speakers off-stage.
De-identified medical imaging from real clinical practice - the kind of messy, varied distribution you only see after a model has been deployed. Not a Kaggle leaderboard.
“When different disciplines share a table, innovation follows. That's the datathon.”
Saptarshi P.
Associate Professor · Indiana University
Full conference-fee coverage for students and early-career researchers. Rolling review, apply with your CV.
Four tracks, priced so you can join the week that matters to you, or do the whole run. Scholarships cover conference fees for students and early-career researchers.
See what's includedAI & STEM program for high schoolers - robotics, clinical AI, and a peek at real medical data work.
Keynotes + a hands-on weekend primer.
Every event from Monday through Sunday.
HITI ASPIRE is a separate track (high-school students only).
Pair the ideas with the build.
Breakfast + lunch included · Prices in USD · Taxes apply at checkout