Projects

A machine learning classifier trained on 172 expert-annotated clinical flow cytometry samples, achieving accurate cell typing, overcoming biases, and scaling to over 1100 samples, demonstrating high accuracy and significant time savings.

Employment of advanced deep learning methods and traditional approaches to identify key factors in predicting clinical outcomes for severe pneumonia patients using diverse patient data, including single-cell RNAseq and electronic health records from large-scale NIAID-funded study.

Exploring single cell spatial transcriptomics, the study compared 10x Xenium and Nanostring CosMx, using Python and R tools to assess their applicability in the laboratory based on factors such as transcript coverage and specificity as well as ability to preserve biological variation.