Multimodal Data Analysis Session (MDAS)
Biomedical research often brings together measurements that describe different aspects of the same person. Multimodal data analysis asks how these complementary sources can be combined to address a research question, and how to evaluate what the combination adds.
Organised by the AI Research Engineering team at the Centre for Machine Intelligence, with support from the EPSRC-funded UK Open Multimodal AI Network, MDAS introduces practical examples from the PyKale biomedical research ecosystem. The Cancer tutorial combines clinical variables and pathology patch features for classification and survival analysis. The Cardiac tutorial explores chest X-ray and ECG representation learning with CardioVAE, and complementary ECG leads with LS-EMVAE.
Work through the examples with us at the session or independently afterwards. The materials explain the inputs, modelling choices and evaluation, starting with synthetic data so no patient dataset is needed.
Explore the tutorials
Track |
What you will explore |
|---|---|
Combine clinical variables and pathology features to predict a category or analyse survival. |
|
Learn from chest X-rays and ECGs with CardioVAE, or from complementary ECG leads with LS-EMVAE. |
Open either tutorial overview and choose an example. Use Run in Google Colab on its notebook page, or Run locally using Quick Start.
For preparation and privacy guidance, see the Participation Guide. To start an example immediately, use Quick Start.
Session details
Date |
Attendee programme |
Venue |
|---|---|---|
Wednesday, 18 November 2026 |
13:00–17:00 |
Ada Lovelace Room (108), Level 1, Regent Court, University of Sheffield |
Synthetic demonstrations teach the software workflow; they do not establish clinical performance. Keep patient data and confidential research off public execution services.
Acknowledgement
We gratefully acknowledge the collaborative contributions and insightful discussions provided by Bilal Tahir, Daniele Tartarini, Zhiyang Chen and Daniel Griffin for the KaleCancer project.