Rebecca Peyser Boiarsky

This paper introduces scSet, a diffusion-based autoencoder that learns patient-level representations from single-cell data, outperforming existing methods in clinical prediction with limited labeled data.

An evaluation of scBERT and scGPT on cell type annotation against L1-regularized logistic regression, with ablations of pre-training and tests of fine-tuning stability.

HiDDEN refines case-control labels in single-cell transcriptomics to detect disease-relevant cell populations, improving detection of subtle perturbations in simulated and real datasets.