Q1 opens with 8DM60 Multimodal AI in Biomedicine, 5CTA0 Statistical Signal Processing and DCB200 Aesthetics of Interaction. September is calm except for two hand-ins; October 14–29 drops seven things in a row, including both exams. Everything worth knowing, course by course.
| Date | What's due |
|---|---|
| Thu 3 Sep, 15:30 | 5CTA0 — introduction lecture, AUD 1 week 1 |
| Fri 5 Sep | 8DM60 — confirm project group or you're dropped from the course blocking |
| Mon 8 Sep, 13:30 | DCB200 — Studio I: critique of analysis + experience prototype formative |
| Fri 11 Sep, 23:59 | 8DM60 — project topic proposal (max 400 words) hand-in |
| Tue 15 Sep, 13:30 | DCB200 — individual A3 poster (10% of grade) hand-in |
| Wed 14 Oct, 23:59 | 8DM60 — project: 1-page poster + reproducible code (ZIP) hand-in |
| Sun 18 Oct, 23:59 | 5CTA0 — lab report, via Ans Delft (30% of grade) hand-in |
| Tue 20 Oct, 13:30 | DCB200 — final prototype + video ≤90 s + pitch (60%) hand-in |
| Thu 22 Oct | 5CTA0 — lab presentations (15-min blocks) hand-in |
| Mon 26 Oct, 13:30 | 5CTA0 — written exam exam · 70% |
| Wed 28 Oct, 13:30–16:30 | 8DM60 — written exam exam · 85% |
| Thu 29 Oct, 17:00 | DCB200 — individual pictorial with reflection (20%). Note: the schedule page says Oct 30, 23:59, but the Canvas assignment closes Oct 30 at 12:00. Assume the earlier one. hand-in |
| resits | 5CTA0 — exam in January 2027 (date TBA) · 8DM60 — 20 Jan 2027, 18:00–21:00 |
Multimodal AI applied to biomedicine: language models, vision models, omics, hospital data, clinical workflow integration and agentic AI. Flipped classroom from lecture 2 on: recent papers before class, the session itself is critical discussion. All material lives on github.com/tueimage/8dm60-multimodal-ai; Canvas is only for announcements and submissions.
| Project (poster + code) | 15% |
|---|---|
| Exam multiple-choice questions | 20% |
| Exam questions on YOUR project | 15% |
| Open / problem-based questions | 50% |
The project: groups of 5, chest X-rays paired with reports (Indiana University / OpenI dataset), starter notebook using MedGemma and QLoRA. You must go beyond the starter: own research question, baseline, quantitative evaluation, failure cases. The guided self-study sessions practice exactly the style of the open exam questions — 4–5 questions per session.
Difficulty: heavy weekly reading of recent papers; assumes neural networks and training are already known. Compute: GPU via one representative per group on the TU/e HPC-server, or free Colab.
The math course of the term: probability and random processes, estimation theory (MLE, CRLB, least squares, Bayes), spectral estimation and detection theory. Flipped classroom: videos and reader on Canvas, Thursday sessions are review and Q&A. There's a Discord server for TA questions.
| Group lab of 4 (report + peer review + presentation) | 30% |
|---|---|
| Written exam (MC + open exercises) | 70% |
To pass you need ≥5.0 on the exam and ≥5.5 averaged with the lab. No opt-out from the lab: skip it and it counts as 0. The exam allows a calculator, one self-made A4 sheet and the official formula sheet.
The lab (repo: tue-bmd/5cta0-project-student) starts now: form a group of 4 on Canvas before you can start. Canvas hosts old exams 2019–2022 with solutions plus the tutorial sets (SLT) with solutions — the material to work through for the exam. The exam repeats the same 4-block structure every year: probability (~1.5 pts), estimation (~2.75), spectral (~3.0), detection (~2.75).
Interaction design: aesthetics of interaction, experience prototyping, embodied and tangible interaction. You work in groups of ≤5 designing an aesthetic experience around an everyday ritual, through iterations: from lo-fi prototype to high-fidelity demonstrator, with critique sessions (studios) every two weeks. The week 1, 3 and 7 lectures don't fit everyone in Helix 0.01 — there's a livestream.
| Individual A3 poster (analysis + redesign) | 10% |
|---|---|
| Final prototype + 60 s pitch | 40% |
| Video ≤90 s in context (sound design + credits) | 20% |
| Individual pictorial (DIS format) with theory reflection | 20% |
| Attitude (detail, critique, proactivity) | 10% |
To pass: average above 5.5. The Sep 15 poster is individual: analyze an everyday ritual through Affordances and the Interaction Frogger Framework, redesign it using one of four papers, and bring an experience prototype to Studio I on Sep 8. Note: the poster goes through a plagiarism check (formative this time, no Examination Committee); the final pictorial doesn't get that leniency.
Difficulty: not hard by content but by cadence: manual iteration, public critique, design readings. Very different from the other two; the risk is letting it slide and having Oct 20 and 29 land on you at once.