Home

Medical data spans multiple modals, each providing distinct insights into biological processes. For example, image data (e.g., MRI, CT, PET) captures structural and functional characteristics of tissues and organs, providing critical insights into anatomical variations, physiological processes, and disease progression. Omics data (e.g., connectomics, genomics) captures system-wide interactions within specific biological systems, enabling deeper insights into disease pathways and personalized medicine. Demographic and structured metadata enable population-level modeling and integration of heterogeneous biomedical data for understanding complex biological systems. AI-driven data mining is crucial for uncovering informative patterns and latent structures in multimodal medical data, enabling the integration and analysis of high-dimensional biomedical data to reveal complex associations and underlying mechanisms. However, applying AI to multimodal biomedical research faces significant challenges, including data heterogeneity, interpretability, precision, and the need for trustworthy and generalizable models. These challenges highlight the need for tailored AI solutions that ensure both robust performance and clinical trustworthiness. To address these challenges and explore how to effectively harness the benefits of advanced AI models for medical applications, we propose to organize a workshop that serves as a platform to foster interdisciplinary discussions, share advancements, and explore innovative AI-driven solutions for multimodal biomedical data analysis.

Topics

Overview: The rapid growth of multimodal biomedical data has created unprecedented opportunities for AI-driven data mining. Medical imaging (e.g., MRI, CT, PET), omics data (e.g., connectomics, genomics, proteomics), and structured metadata provide complementary insights into biological processes. AI methods play a critical role in integrating and analyzing these diverse data sources, enabling novel discoveries in disease mechanisms, precision medicine, and population health modeling. However, the complexity and heterogeneity of multimodal biomedical data introduce significant challenges. Issues such as data standardization, model interpretability, precision, and generalizability remain key barriers to effectively leveraging AI in biomedical research. This workshop aims to bring together researchers and practitioners to discuss recent advancements, emerging challenges, and future directions in AI-driven multimodal biomedical data mining.

Topics of Interest: We invite original contributions on topics including, but not limited to:

1. Deep learning for multimodal biomedical data analysis, including but not limited to medical imaging, genetic data, omics, and structured clinical metadata.
2. Large foundation models for healthcare applications leveraging multimodal medical data.
3. Multimodal fusion techniques for disease diagnosis, prognosis, and precision medicine.
4. Techniques for explainability, robustness, trustworthiness, and generalization in multimodal learning systems for medical applications.
5. Discovery of disease- and phenotype-associated biomarkers through multimodal large-scale data-driven approaches.
6. Computational challenges and advanced solutions in large-scale biomedical data mining.

Submissions

Authors are invited to submit original papers that have not been published elsewhere and are not currently under consideration for another journal, conference, or workshop. All submitted papers should follow the IEEE 2-column format.

• Long papers: Maximum 8 pages, including references and appendices.
• Short papers: Maximum 4 pages, including references and appendices.

All submissions will be triple-blind reviewed by the Program Committee based on technical quality, relevance to the scope of the conference, originality, significance, and clarity. Manuscripts must be submitted electronically through the online submission system: Submit your paper here . Email submissions are not accepted. Accepted papers will be included in the ICDM Workshop Proceedings (separate from ICDM Main Conference Proceedings), and each workshop paper requires a full registration. Meanwhile, duplicate submissions of the same paper to more than one ICDM workshop are forbidden.

Presentations

1. Each accepted paper is required to be presented in two formats (both):

• A 10-minute oral presentation (including Q&A)
• A poster presentation during the coffee break, both in the workshop room
.

Note: We are currently assigned to Room Federal B, but this may change. Please double-check the Whova conference app during the event for the latest room information.

2. There is no specific style or template for your presentation slides and poster. However, the recommended size of poster is 36 x 24 inches.

If attending in person: You will deliver the oral talk and present your poster on-site.
If attending remotely: Please record a 10-minute video of your oral presentation (in MP4 format). We will play your video and display your poster at the workshop.

3. For all attendees, please submit your following presentation materials by November 3, 2025 to the Google Drive folder (has sent out to you).

• Your oral presentation slides (.ppt), or a recorded video (.mp4) if remote
• Your poster (.ppt format)

4. To ensure a smooth workshop experience, all oral presentations will be run from our prepared computer. Please ensure your files are submitted on time.

Important Dates

• Paper submission deadline: September 2, 2025
• Notification of acceptance: September 15, 2025
• Camera-ready deadline: October 5, 2025
• Workshop date: 8:00 AM - 12:30 PM (local time), November 12, 2025

Organizers

Program Chairs

Haoteng Tang, PhD Assistant Professor University of Texas Rio Grande Valley

Lu Zhang, PhD Assistant Professor Indiana University Indianapolis

General Chairs

Paul M. Thompson, PhD Professor University of Southern California

Shuo Li, PhD Professor Case Western Reserve University

Danda B. Rawat, PhD Professor Howard University

Pingkun Yan, PhD Associate Professor Rensselaer Polytechnic Institute

Kewei Sha, PhD Associate Professor University of North Texas

Liang Zhan, PhD Associate Professor University of Pittsburgh

Haihua Chen, PhD Assistant Professor University of North Texas

Web Chair

Guodong Liu, PhD Research Scientist Eli Lilly and Company

Local Chairs

Xiaowei Xu, PhD Associate Professor Guangdong Provincial People's Hospital

Di Ming, PhD Assistant Professor Chongqing University of Technology

Program Committee

Keynote Speakers

James Li, Ph.D.

Associate Professor and Director of Research Computing Support

Department of Biostatistics, Bioinformatics and Biomathematics
Georgetown University Medical Center

About the Speaker: James Li, Ph.D., is an Associate Professor and Director of Research Computing Support in the Department of Biostatistics, Bioinformatics and Biomathematics at Georgetown University Medical Center. His research centers on bioinformatics, omics data analysis, and machine learning, with broad collaboration across the Medical Center to support biomedical data integration and reproducible analytics. His recent work focuses on agentic AI for adaptive omics analysis and workflow automation, connecting empirical bioinformatics practice with next generation AI systems. Dr. Li received his Ph.D. in Information Systems from the University of Maryland, Baltimore County.

Title: Agentic AI for Biomedical Discovery: From Adaptive Omics Analysis to Translational Impact

Abstract: The rapid expansion of biomedical omics data has outpaced traditional analytical workflows that depend on static models and manual tuning. This talk explores how agentic AI, an emerging class of adaptive systems capable of reasoning and workflow orchestration that can transform the way we analyze biological data and uncover disease biomarkers. Building on recent empirical studies in omics embedding and adaptive RNA sequencing workflows, the talk highlights how agentic frameworks can enhance reproducibility, interpretability, and generalization across diverse datasets. These approaches connect algorithmic innovation with practical biomedical insight, enabling data driven discovery that integrates both molecular and clinical dimensions. Beyond specific methods, the talk outlines a broader vision of an interconnected agentic biomedical ecosystem with a core of adaptive omics intelligence extending toward clinical trial analytics and AI supported drug discovery. This evolving framework links computational adaptability with translational impact, pointing to a future in which intelligent and self-managing research ecosystems accelerate biomarker identification and therapeutic innovation.

Lifang He, Ph.D.

Associate Professor

Department of Computer Science and Engineering
Lehigh University

About the Speaker: Dr. Lifang He is an Associate Professor of Computer Science and Engineering at Lehigh University. Her research spans AI for health, multimodal neuroimaging, graph learning, and foundation models for clinical decision support. She leads several NIH- and NSF-funded projects on AI-driven biomedical systems and has authored over 200 papers in prestigious journals and conferences. She currently serves as an Associate Editor for ACM Transactions on Computing for Healthcare and the International Journal on Machine Learning and Cybernetics, and as Chair of the IEEE Computer Society at Lehigh Valley Section.

Title: From Brain-Inspired AI to AI-Inspired Brain Science: The Rise of Generalist Models in Neuroscience

Abstract: The long-standing dialogue between artificial intelligence and neuroscience is entering a new era. While early AI drew inspiration from neural principles to design learning architectures, recent advances in foundation and generative models are reshaping how we study the brain itself. This talk explores how generalist AI systems are becoming "computational microscopes" for understanding human cognition and neurodynamics. I will highlight our recent efforts—BiomedGPT, NeuroSTORM, and UniBrain—to show how generalist architectures unify diverse data modalities and multi-task objectives to support scalable clinical analysis and enable personalized modeling and decision support. Building on these examples, I will briefly discuss emerging opportunities and challenges in AI-inspired brain science, where generalist models may accelerate discovery, reveal cross-species neural correspondences, and open new frontiers in ethically grounded, interpretable, and data-efficient research.

Agenda

Time: 8:00AM – 12:20PM, November 12, 2025

Room: Federal B @ Capital Hilton, 1001 16th Street NW Washington, DC 20036, US

Accepted Papers

1. Who Matters More in Radiology Report Generation: Vision Encoders or Language Models?
   Kun Zhao, Yang Du, Rhianna Zhang, Liang Zhan, Dongkuan Xu, Pengfei Gu, and Haoteng Tang


2. Selective Channel-Quality–Guided EEG Denoising for Brain Disorder Prediction
   Yang Du, Siyuan Dai, Marcus Zhan, Guodong Liu, Paul Thompson, Heng Huang, and Haoteng Tang


3. BrainNet-MoE: Brain-Inspired Mixture-of-Experts Learning for Neurological Disease Identification
   Jing Zhang, Xiaowei Yu, Tong Chen, Chao Cao, Minheng Chen, Yan Zhuang, Yanjun Lyu, Lu Zhang, Li Su, Tianming Liu, and Dajiang Zhu


4. Geant4-Based Simulation and Image Reconstruction of Tc-99m for Radiopharmaceutical Imaging
   Ehsan Shakeri, Matthias Gobbert, Vijay Sharma, Lei Ren, Ananta Chalise, Stephen Peterson, and Jerimy Polf


5. Pretraining a Generative AI Model on CBT Sessions to Emulate Human Therapist Strategies: A Study Using the APA PsycTherapy Database
   Fabiha Islam, Kun Zhao, Bart Knijnenburg, Mei-Hsiu Chen, Alex Leow, Purushothaman Muthukanagaraj, Liang Zhan, and Chao Shi


6. Quantitative Phosphocreatine Mapping with a Transformer-based Regression Network
   Qi Huang, Haoteng Tang, Han Tang, Caleb Berberet, Liya Dai, Thomas Schindler, Linda Peterson, Yang Yang, Yan Yan, Pamela Woodard, and Jie Zheng


7. Why Text Prevails: Vision May Undermine Multimodal Medical Decision Making
   Siyuan Dai, Lunxiao Li, Kun Zhao, Eardi Lila, Paul Crane, Heng Huang, Dongkuan Xu, Haoteng Tang, and Liang Zhan


8. SynM-FPP: Few-shot Learning on SynthMorph-based Motion Correction Framework for Myocardial First-Pass Perfusion
   Han Tang, Qi Huang, Haoteng Tang, Liya Dai, Caleb Berbert, Scott Bugenhagen, Thomas Schindler, Linda Peterson, Yang Yang, Yan Yan, Pamela Woodard, and Jie Zheng


9. Agentic AI for Disease-Aware Adaptive Multi-Omics Embedding
   James Li, Andrew Woods, and Ao Yuan

Photos

TBD

Contact

If you have any questions regarding the workshop, feel free to reach out to us:

• Email: haoteng.tang@utrgv.edu (Dr. Haoteng Tang)
• Email: lz50@iu.edu (Dr. Lu Zhang)