I work on extending deep learning methods in personalised medicine and federated learning. My project ideas relate to integration of multi-omics pathology data (like metabollomics, proteomics) with radiology images (MRI/PET scans), to get a wholistic understanding of the disease. I aim to develop methods for earlier disease diagnosis and treatment effects on an individual (for example, the effect on Immune checkpoint inhibitor therapy on cancer patients). I also work on enabling medical data sharing without privacy leaks to promote further research into this domain.
In the past, I have worked on knowledge bases & information completeness, reinforcement learning & optimal control, and unsupervised learning methods.