Using biological data, such as single-cell and bulk RNA-seq data from patients treated with different onoclogical treatments, to train ML models that can predict response to the therapies.
Clinical data is both temporal and structural. Thus, transformers can be a great way to model this kind of data. We are developing transformers-based architectures to model clinical data and use it to improve risk predictions
We are developing semi-supervised approaches for problems that are usually treated as unsupervised.
