Xiaoxiang Neurology Rare Disease Academic Salon
Date:
We introduced multimodal machine learning technologies for multi-omics data, and discussed how to discover rare diseases with deep learning methods.
In this talk, we discussed how spatial transcriptomics, as a unique data type containing both gene expression and tissue images, offers new possibilities for multimodal joint learning. We used recent works from our team, including STANDS, as examples to illustrate how data-driven or theory-driven multimodal learning methods can be applied to some specific tasks, including but not limited to discovering anomalous tissue regions, integrating batch effects, and exploring gene co-expression.