Yuda Bi

This paper introduces a residual-ratio framework to audit transcriptomic gene signatures by quantifying their independence from background expression structure, showing it discriminates biological signatures from random gene sets across multiple cancer types.

This paper introduces SpecRNA-QA, a reference-free RNA 3D quality assessment method that outperforms local descriptors and statistical potentials by capturing global topological coherence.

This paper introduces a multimodal vision transformer (MultiViT) to classify schizophrenia using structural and functional MRI data, achieving higher accuracy than unimodal and existing multimodal methods.

This paper introduces a generative deep learning framework to predict functional network connectivity biomarkers from brain structure using a modified vision transformer GAN.