Chaithra Umesh

"Advancing Patient Stratification and Interpretability in Clinical and Biomedical Data through a Novel Approach"

My research revolves around the dynamic intersection of clinical and biomedical data, primarily focusing on achieving meaningful patient stratification. Delving into high-dimensional clinical tabular data characterized by diverse feature types, I specialize in applying state-of-the-art dimensionality reduction algorithms. My efforts extend to comprehensive benchmarking across various clinical datasets, evaluating the efficacy of a novel approach to patient stratification compared to established algorithms.

In parallel, I am actively investigating the integration of attention mechanisms into the linked tabular data generation domain. This revolutionary approach enhances the interpretability and contextual comprehension of complex relationships within clinical datasets. My idea is to integrate transformer architecture into tabular data so that if a part of the data can not be shared for privacy, this model should be able to learn it.