Texas Christian University
I am an Assistant Professor of Computer Science at Texas Christian University (TCU). Previously, I worked as a researcher in the Department of Artificial Intelligence & Informatics at Mayo Clinic and an assistant professor of computer science at Bowling Green State University. Before that, I received postdoctoral training at the School of Biomedical Informatics, UTHealth, and I earned my Ph.D. in Computer Science from Embry-Riddle Aeronautical University.
My research areas include Transfer Learning (TL), Graph Representation Learning (GRL), and Biomedical Informatics (BMI). My current research focuses on multimodal electronic health record (EHR) modeling for precision healthcare, reliable AI-driven drug repurposing and discovery, and fundamental learning algorithms for heterogeneous biomedical data. In particular, I am interested in transfer learning for multi-center EHR modeling, context-conditioned graph learning for biomedical knowledge discovery, and integrating knowledge graphs with multi-omics data to generate reproducible and interpretable therapeutic hypotheses. My long-term goal is to develop reliable and generalizable computational methods that accelerate drug discovery and enable precision healthcare. For more details and collaborations, please check here.
Recent News
08/2026: On behalf of the SPACESCANS team, with collaborators from TCU, Mayo Clinic, Harvard, and Indiana University, we hosted a successful workshop at ISEE 2026 in Munich, Germany, demonstrating our environmental data standardization and linkage tool.
08/2026: Two works on LLM4EHR from collaborations between TCU and Mayo were accepted to IEEE ICHI 2026 and IEEE HealthCom 2026. Codes and full papers are coming soon.
"LLMs for Real-World EHR Modeling: When to Prompt, Fine-Tune, and Rely on Classical ML" proposes a LLM-based method for HER modeling, and discusses the feasibility of LLM-usage in EHR modeling.
"Enhancing ICD Code Assignment Accuracy Using a Fine-Tuned Llama-3.1 Model with Chain-of-Thought Reasoning" provides a LLM-powered automatic ICD accuracy validation method.
08/2026: One collaborative work between Mayo and KSU was accepted to IEEE ICHI 2026.
08/2026: One work led by Mayo Clinic was accepted to NPJ Health Systems. This work offers a highly automated tool for ontology enrichment, a critical step for grounded knowledge update and heterogeneous data harmonization.
OntoCodex: a multi-agent biomedical ontology enrichment framework. - Jingna Feng & Yue Yu
06/2026: I will join the Department of Computer Science at the Texas Christian University in Fall 2026. I am looking for self-motivated PhD students to work on AI4Healthcare.
06/2026: The 1st Young Scholar Symposium was a great success at ICHI 2026! We will run it again next year.
03/2026: Congratulations to Ryan Jeon, Ph.D., on beginning his career as a Technical Manager at Papa John’s! Wishing him great success as he works on developing Agentic AI and driving innovation in the field.
03/2026: I started a new service role as the Newsletter Editor of ACM EIGTRUST (Emerging Interest Group on Trustworthy and Responsible Systems). The first newsletter will be delivered in May, 2026. Join EIGTRUST at no cost for now: Join Today!
03/2026: I co-presented on the Environmental Health Language Collaborative (EHLC) web series and workshops hosted by NIEHS. Check out the slides.
02/2026: I am co-chairing a special Symposium for High School and Undergraduate Students on ICHI 2026: check it out here.
01/2026: One work led by Mayo Clinic and UTHealth was accepted to NPJ Health Systems.
11/2025: One collaborated paper was accepted to WACV 2026.
Ryan and Lijing will attend their Ph.D. graduation ceremony at BGSU, 08/07/2025 at 5 PM. Feel free to stop by and say hi to them if you would be in town!
08/2025: One of our recent works led by Mayo Clinic and BGSU was accepted to CIKM 2025.
06/2025: Lijing Zhu and Dong Hyun Jeon have defended their dissertations! Congrats!
05/2025: Lijing Zhu and Dong Hyun Jeon will defend their dissertations at 06/12 and 06/13 respectively.
05/2025: One of our recent works led by BGSU & Mayo was accepted to ECML-PKDD 2025.
04/2025: Lijing Zhu, Ph.D. candidate in Data Science, received a computer science tenure-track assistant professor offer from the University of Houston at Clear Lake. Congrats to Lijing!
01/2025: Lijing Zhu and Dong Hyun Jeon passed their dissertation topic approval defenses! Congrats!
12/2024: I will join Mayo Clinic in February 2025, as a researcher in the Department of AI and Informatics.
12/2024: I will become an adjunct Assistant Professor at BGSU.
10/2024: Two papers from my Ph.D. students (Lijing Zhu and Dong Hyun Jeon) were accepted to IEEE BigData 2024.
10/2024: One collaborated paper was published at 2024 IEEE 36th International Conference on Tools with Artificial Intelligence (ICTAI).
10/2024: I gave a talk at INFORMS 2024.
07/2024: One collaborated paper was accepted to International Journal of Molecular Sciences.
02/2024: One collaborated paper was accepted to Journal of American Heart Association.
Featured Works
A decade survey of transfer learning (2010–2020) - IEEE Transactions on Artificial Intelligence 2021
Shuteng Niu, Yongxin Liu, Jian Wang, Houbing Song
Abstract: Transfer learning (TL) has been successfully applied to many real-world problems that traditional machine learning (ML) cannot handle, such as image processing, speech recognition, and natural language processing (NLP). Commonly, TL tends to address three main problems of traditional machine learning: (1) insufficient labeled data, (2) incompatible computation power, and (3) distribution mismatch. In general, TL can be organized into four categories: transductive learning, inductive learning, unsupervised learning, and negative learning. Furthermore, each category can be organized into four learning types: learning on instances, learning on features, learning on parameters, and learning on relations. This article presents a comprehensive survey on TL. In addition, this article presents the state of the art, current trends, applications, and open challenges.
Bridging the computational-experimental gap: leveraging large language model to prioritize Alzheimer’s therapeutics based on comparison of learning models - NPJ Health System 2026
Manqi Li, Shuteng Niu, Yifeng Xu, Jianfu Li, Xinyue Hu, Duan Liu, Merve Atik, Xiaolei Xu, Liewei Wang, Nilufer Ertekin-Taner, Cui Tao
Abstract: Alzheimer’s Disease (AD) necessitates accelerated treatment discovery, positioning drug repurposing as a vital strategy. While computational approaches such as knowledge graph reasoning and transcriptomics show promise, they often yield divergent results, complicating the selection of candidates for experimental follow-up. To bridge the gap between computational prediction and in vivo validation, we propose an advanced framework leveraging large language models (LLMs). We systematically evaluated three state-of-the-art computational methods (TxGNN, CompGCN, and regularized logistic regression (RLR)) to generate a unified list of 90 candidates. An LLM-based agent was then used to automate evidence synthesis from biomedical literature, mimicking expert curation to efficiently refine the list using transparent selection criteria. Validated against real-world AD patient data, clinical trial registries, and pharmacological reviews, our framework demonstrated high robustness and clinical relevance. By integrating computational predictions with scalable evidence synthesis, this approach enhances the efficiency and consistency of candidate prioritization. Ultimately, this versatile framework offers a scalable pathway to accelerate the translation of repurposed drugs for AD and other complex diseases.
ETT-CKGE: Efficient Task-driven Tokens for Continual Knowledge Graph Embedding - 2024 ECML-PKDD
Lijing Zhu, Wenbo Sun, Li Yang, Yixin Xie, Tiehang Duan, Cui Tao, ..., Shuteng Niu
Abstract: Continual Knowledge Graph Embedding (CKGE) aims to incorporate new knowledge while retaining prior information, but existing methods face challenges in efficiency and scalability. These stem from suboptimal knowledge preservation using manually designed importance scores and costly graph traversal procedures. We propose ETT-CKGE (Efficient, Task-driven Tokens for Continual Knowledge Graph Embedding), a task-guided method that uses learnable tokens to capture task-relevant signals without explicit node scoring or traversal. These reusable tokens enable efficient, token-masked embedding alignment via simple matrix operations. Experiments on six benchmarks show that ETT-CKGE delivers competitive performance with significantly improved training efficiency and scalability.
Leveraging Vulnerabilities in Temporal Graph Neural Networks via Strategic High-Impact Assaults - CIKM 2025
Dong Hyun Jeon, Lijing Zhu, Haifang Li, Pengze Li, Jingna Feng, Tiehang Duan, Houbing Herbert Song, Cui Tao, Shuteng Niu
Abstract: Temporal Graph Neural Networks (TGNNs) have become indispensable for analyzing dynamic graphs in critical applications such as social networks, communication systems, and financial networks. However, the robustness of TGNNs against adversarial attacks, particularly sophisticated attacks that exploit the temporal dimension, remains a significant challenge. Existing attack methods for Spatio-Temporal Dynamic Graphs (STDGs) often rely on simplistic, easily detectable perturbations (e.g., random edge additions/deletions) and fail to strategically target the most influential nodes and edges for maximum impact. We introduce the High Impact Attack (HIA), a novel restricted black-box attack framework specifically designed to overcome these limitations and expose critical vulnerabilities in TGNNs. HIA leverages a data-driven surrogate model to identify structurally important nodes (central to network connectivity) and dynamically important nodes (critical for the graph's temporal evolution). It then employs a hybrid perturbation strategy, combining strategic edge injection (to create misleading connections) and targeted edge deletion (to disrupt essential pathways), maximizing TGNN performance degradation. Importantly, HIA minimizes the number of perturbations to enhance stealth, making it more challenging to detect. Comprehensive experiments on five real-world datasets and four representative TGNN architectures (TGN, JODIE, DySAT, and TGAT) demonstrate that HIA significantly reduces TGNN accuracy on the link prediction task, achieving up to a 35.55% decrease in Mean Reciprocal Rank (MRR) - a substantial improvement over state-of-the-art baselines. These results highlight fundamental vulnerabilities in current STDG models and underscore the urgent need for robust defenses that account for both structural and temporal dynamics. Code and Data are available at https://github.com/ryandhjeon/hia.
Main Collaborators
TAO Lab @ Mayo Clinic
Dr. Yixin Xie @ Kennesaw State University
Kai Zhang @ Lehigh University
Yushang Jiang @ University of Connecticut
SONG Lab @ UMBC