Publications | Zhi Lab

Selection emphasizes influential methods, reusable software, and representative papers from the lab's main research directions. Citation counts change over time; please see Google Scholar for the live profile.

Clinical AI

EHR Foundation Models

Med-BERT: pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction

L. Rasmy, Y. Xiang, Z. Xie, C. Tao, D. Zhi. npj Digital Medicine, 2021.

Representative clinical foundation-model work for structured EHR diagnosis sequences.

View paper

A study of generalizability of recurrent neural network-based predictive models for heart failure onset risk

L. Rasmy, Y. Wu, N. Wang, X. Geng, W. J. Zheng, F. Wang, H. Wu, H. Xu, D. Zhi. Journal of Biomedical Informatics, 2018.

Early large-scale EHR deep learning work on portability and generalizability across heterogeneous clinical data.

View paper

Genome Informatics

IBD, Relatedness, and Pangenomics

RaPID: ultra-fast, powerful, and accurate detection of segments identical by descent in biobank-scale cohorts

A. Naseri, X. Liu, K. Tang, S. Zhang, D. Zhi. Genome Biology, 2019.

Biobank-scale IBD detection method and reusable software for population genetics informatics.

View paper

RAFFI: accurate and fast familial relationship inference in biobank-scale cohorts using IBD segments

A. Naseri, D. Zhi and collaborators. PLoS Genetics, 2021.

Representative work on family relationship inference from IBD structure in large cohorts.

Journal site

FiMAP: detecting disease-associated IBD segments in biobank-scale cohorts

Zhi Lab and collaborators. PLoS Genetics, 2023.

Representative IBD mapping work connecting shared haplotypes with trait association signals.

Journal site

Imaging Genetics

AI-Derived Phenotypes and GWAS

UDIP-Brain: unsupervised deep imaging phenotypes for brain imaging genetics

Zhi Lab and collaborators. Communications Biology, 2024.

Representative work using unsupervised deep imaging phenotypes for neuroimaging GWAS discovery.

Journal site

UDIP-FA: unsupervised deep imaging phenotypes for white matter diffusion MRI genetics

Zhi Lab and collaborators. Nature Communications, accepted 2026.

Extends UDIP to diffusion MRI white matter FA maps and large-scale imaging genetics.

Accepted

Bioinformatics

Representation Learning

Gene2vec: distributed representation of genes based on co-expression

J. Du, P. Jia, Y. Dai, C. Tao, Z. Zhao, D. Zhi. BMC Genomics, 2019.

Highly visible gene representation learning work connecting co-expression data with embedding methods.

View paper