Nan Lin

Multimodal AI for Single cfDNA Profiling and Cancer Screening

The paper develops a Transformer-based model called cfAI to profile and detect cancer-derived cfDNA by integrating methylation, fragmentomics, and histone modifications, achieving high sensitivity and specificity in multi-cancer detection.

scPhoenix: A Contrastive Learning-based Framework with Aux-Core Feature Disentanglement Enhances Sparse Cellular Multi-omics Translation

scPhoenix enhances sparse cellular multi-omics translation using contrastive learning and aux-core feature disentanglement, preserving heterogeneity and improving performance in high sparsity and spatial heterogeneity scenarios.

Policy gradient-guided ensemble learning for enhanced polygenic risk prediction in ultra-high-dimensional genomics

This paper introduces ELAG, an ensemble learning framework for polygenic risk prediction that improves accuracy in high-dimensional genomics data by using policy gradient methods to guide variant selection.