Clinical Day

Associate Professor, University of Missouri

Abstract: Cardiovascular disease remains the leading cause of mortality worldwide, and early identification of at-risk individuals is needed to improving outcomes. In this talk, I will present advances in artificial intelligence (AI) applied to multimodal cardiovascular imaging—focusing on ECG, cardiac US, CT, SPECT, and PET—to support early risk assessment. I will discuss the integration of deep learning models with imaging and electronic health record (EHR) data to predict adverse events before clinical manifestation. Emphasis will be placed on recent developments, which enable generalizable, scalable, and interpretable applications across imaging modalities and clinical populations. The presentation will explore findings from studies that demonstrate how multimodal imaging paired with AI may be able to improve prognostic accuracy, and highlight technical challenges such as data harmonization, cross-institutional validation, and model calibration. The goal is to illustrate how AI-enhanced imaging pipelines can move from proof-of-concept to clinical utility, advancing the precision and timeliness of cardiovascular risk prediction.

Biography: Associate Professor of Biomedical Informatics, Biostatistics, Epidemiology, and Cardiology, and Director of Graduate Studies in Health Informatics at the University of Missouri School of Medicine. My clinical and research interests lie at the intersection of cardiovascular medicine and cardiometabolic health, cardiovascular imaging, artificial intelligence, and evidence-based medicine. I use advanced imaging technologies, such as cardiac CT, SPECT, and PET, among others, to assess cardiovascular health and disease. In research, I focus on applying machine learning and artificial intelligence to predictive modeling, particularly in the context of ECG and EHR data, to forecast cardiovascular risks and outcomes. My work is centered on explainable AI to improve clinical decision support and patient care.

Assistant Professor, UTHealth

Abstract: Alzheimer’s Disease (AD) is highly heterogeneous, posing substantial challenges in accurate diagnosis, prognosis, and individualized therapies. While current research often focuses on isolated markers—such as structural MRI, cognitive evaluations, or molecular measures—our work integrates multiple data modalities to uncover robust AD subtypes. In this talk, we present a trilogy of studies aimed at refining computational phenotyping methods that bridge neuroimaging, neuropsychological assessments, and proteomic/connectivity data. First, we show how data-driven clustering on imaging and clinical measures can parse out distinct AD subgroups with shared clinical and demographic features. This multi-modal approach demonstrates that standard “one-size-fits-all” characterizations of AD overlook important subgroup differences. Second, by tracking longitudinal MRI atrophy alongside cognitive decline within a phenotyping framework, we reveal how subtypes diverge in both cortical changes and memory/language deficits—enabling more nuanced progression modeling. Finally, we incorporate age-related proteins and structural connectivity to identify latent molecular-pathology clusters, revealing potential mechanistic underpinnings of certain subtypes. Notably, these protein-network associations suggest new therapeutic targets aligned with the unique biological profiles of each subgroup. Throughout, we leverage advanced computational tools—nonnegative matrix factorization, graph-based network analysis, and outcome-guided clustering—to integrate large, heterogeneous datasets seamlessly. Beyond demonstrating methodology, the talk underscores the utility of these subtypes for more precise patient stratification in clinical trials and personalized interventions. Our findings illustrate how multi-modal phenotyping can disentangle AD’s complexity, advancing AI-driven discovery and clinical translation in neurodegenerative research.

Biography: Dr. Yejin Kim is an Assistant Professor at the McWilliams School of Biomedical Informatics at UTHealth Houston and Associate Director of the Center for Secure Artificial Intelligence for Healthcare. Her research focuses on developing AI algorithms for disease progression modeling and therapy development, with a particular emphasis on neurodegenerative diseases such as Alzheimer’s. She leads multiple data-driven initiatives in causal inference, computational phenotyping, and clinical trial optimization. Dr. Kim’s work is supported by significant NIH funding and has been widely published. She also plays an active role in academic service, education, and community-led AI innovation.

Professor, UTMB

Abstract: Pathology slides contain an extraordinary density of diagnostic information—much of it trapped between layers of glass and historically underutilized. With the growing adoption of digital pathology and the availability of computer vision AI tools across imaging disciplines, we now have the infrastructure to computationally interrogate this data at scale. At the University of Texas Medical Branch (UTMB), we have implemented a fully digital pathology workflow and are incorporating validated deep learning models for cancer detection and grading into routine diagnostic practice. Building on this digital transformation, we have developed a custom GPT-based system using detailed prompt engineering and clinical logic grounded in established, evidence-based guidelines. The model extracts relevant information from uploaded data sources—including clinical notes, laboratory values, radiology findings, and pathology reports—to generate integrative diagnostic reports with risk stratification and personalized management recommendations. It can also generate an optional summary for the patient in clear, non-technical language. These innovations help bridge clinical silos, support shared decision-making, and deliver meaningful value to patients and the health system.

Biography: Dr. Harshwardhan Thaker is a Professor in the Department of Pathology at the University of Texas Medical Branch (UTMB), where he serves as Vice Chair for Digital & Integrative Pathology, Vice Chair for Clinical Outreach, and Director of Anatomic Pathology. His research focuses on the adoption of digital pathology and the application of artificial intelligence in pathology and medicine. Dr. Thaker has a longstanding interest in fetal, perinatal, and placental pathology, as well as molecular genomics. He has collaborated on numerous research projects using model organisms to study the pathology of viral infections.

Associate Professor, MDA

Abstract: Spatial biology aims to understand the organization of complex biological systems including cancer. In cancer research the field aims to better understand the tumor and the tumor microenvironment including: the underlying mechanisms of tumor biology (cell interactions and tumor dynamics); changes of the tumor dynamic over time and under certain treatment (space and time); aid in the discovery of biomarkers; and to identify potential therapeutic targets. Spatial biology relies in the visualization of different types of data (e.g. protein or transcriptomic); the visualization of this data often utilizes histologic slides that are assayed and imaged, analytes are mapped into the tissue which have certain biological, morphological and functional compartments. In this talk, attendees will gain insights in the integration of pathology-based approaches that are used to understand the spatio-temporal features of tumors and the tumor microenvironment. These approaches can help to facilitate the workflows of multi-omic/multi-modal data including spatial transcriptomics and proteomics and provides strategies to improve the quality of data analytics, and interpretation of data.

Biography: Dr. Luisa M. Solis Soto is pathologist with formal training in surgical pathology and extensive experience and interest in translational, molecular, and immune pathology. Currently Dr. Solis Soto is an associate professor in the Department of Translational Pathology (TMP), University of Texas MD Anderson Cancer Center, she directs the Digital ImmuneProfiling and Pathology Laboratory (DIPP Lab), and she is a co-director of the TMP Immunoprofiling Laboratory (TMP-IL) MD Anderson Moonshot platform, which designs, validates and analyzes immune and molecular biomarker on samples obtained in clinical trials using a comprehensive approach. Her research focuses on the integration of pathology and biomarker analysis in tumor tissue for the development of predictive biomarkers, particularly in the realms of immunotherapy and targeted therapy.

Associate Professor, MDA

Abstract: Artificial intelligence (AI) is reshaping the pathology landscape by enabling unprecedented diagnostics, biomarker discovery, and disease modeling capabilities. From enhancing routine histopathological assessments to powering multi-omics integration and spatial biology analytics, AI tools are increasingly embedded in clinical, translational, and research workflows. However, their integration into the pathology ecosystem requires rigorous validation, thoughtful implementation, and continuous interdisciplinary collaborations. In this talk, we will explore the current state of AI in pathology, highlight practical applications and real-world case studies from academic and core laboratory settings, and outline strategic considerations for successful adoption. Attendees will gain insights into how AI is not only enhancing precision and reproducibility in pathology research but also accelerating the pace of discovery and therapeutic development.

Biography: Dr. Maria Gabriela Raso is an Assistant Professor in the Department of Translational Molecular Pathology at The University of Texas MD Anderson Cancer Center, where she also serves as Director of the Research Histology Core Laboratory. Her research focuses on the integration of translational pathology, digital imaging, and spatial biology to advance biomarker discovery and precision medicine. She leads several translational initiatives and multi-institutional collaborations aimed at characterizing tumor evolution, validating therapeutic biomarkers, and enabling high-resolution tissue analytics. With a strong commitment to education and mentorship. Her work is supported by extramural and intramural funding, and she has authored numerous publications at the intersection of pathology and translational science.

Professor, MDA

Abstract: Commercialization of spatial transcriptomics technologies have made it possible to observe the spatial-transcriptional organization in disease tissues from cancer patients of similar diagnosis or going through similar treatment. By comparison with normal tissues, niches (a.k.a, cellular neighborhoods) characteristics of disease histology, i.e., cellular organization and spatially distinct gene expression patterns (such as spatial/metabolic gene expression gradients), can be identified through statistical differential analysis. Functional interpreting novel molecular signatures associated with the niches, however, remains a very challenging task, due to a lack of rich, granular, and context-specific knowledge-base to perform annotation. To address this challenge, we applied large language model (e.g., Llama 2.0) to summarize cell types, genes, pathways, and biological concepts from over 24,000 Pubmed abstracts into a set of immune cell knowledge graphs (ICKG). Through ICKG-based reasoning, we are able to effectively interpret disease-tissue associated niches in immunologists’ language and gain novel insights on tissue physiology and disease etiology. The ICKG tools is available at https://kchen-lab.github.io/immune-knowledgegraph.github.io/.

Biography: Dr. Ken Chen is a Professor in the Department of Bioinformatics and Computational Biology at The University of Texas MD Anderson Cancer Center and Director of Bioinformatics at the Sheikh Khalifa Bin Zayed Al Nahyan Institute for Personalized Cancer Therapy. He also holds an adjunct faculty position in Computer Science at Rice University.

Dr. Chen earned his Ph.D. in Electrical and Computer Engineering from the University of Illinois at Urbana-Champaign, followed by postdoctoral training at UC San Diego. His research focuses on cancer genomics, tumor heterogeneity, and the development of computational tools and AI-based methods for biomarker discovery. He is the co-developer of widely used tools such as BreakDancer, VarScan, and Monovar, and has contributed to major projects including TCGA and the 1000 Genomes Project.

His lab’s work spans single-cell/spatial transcriptomics, multiomics integration, and cancer evolution modeling, with support from NIH, CPRIT, and the Chan Zuckerberg Initiative. Dr. Chen has received several awards recognizing his excellence in research and mentorship.

Professor and Co-Director of IDSO, MDA

Abstract: Artificial Intelligence has great promise for enhancing clinical outcomes and streamlining operations within healthcare institutions. However, the challenge lies in effectively implementing AI to realize its full value and impact. This talk will delve into the critical need to align clinical needs with AI innovations, ensuring that AI solutions are not only cutting-edge but also fit-for-purpose. We will explore the specific contexts in which these technologies are applied, considering the roles of clinical, operational, technology, and data teams. We’ll also examine the importance of data availability, quality, and flow on the performance of these models, all of which depend on robust governance processes and infrastructure. Come join this talk intended for data scientists, clinicians, researchers, and AI enthusiasts as we cover key considerations from model development through to successful implementation – highlighting common pitfalls, challenges, opportunities and considerations to ensure safe and responsible AI.

Biography: Dr. Caroline Chung, M.D., M.Sc., F.R.C.P.C., C.I.P. is Vice President and Chief Data & Analytics Officer at MD Anderson Cancer Center, where she also serves as Co-Director for the Institute for Data Science in Oncology. A clinician-scientist and professor in Radiation Oncology and Diagnostic Imaging, she specializes in central nervous system (CNS) malignancies. Her research focuses on quantitative imaging and computational modeling to improve cancer diagnosis and personalize treatment. Dr. Chung earned her M.D. from the University of British Columbia and completed further training and a research fellowship at Princess Margaret Hospital in Toronto. Nationally and internationally, she plays leadership roles in AI, data science, digital twins and imaging initiatives, including with ASCO, NIH, and ICRU.

Professor, Assoc.Director and Chair of SMAB, HM

Abstract: Spatial intelligence—the ability to model spatial relationships across cells, tissues, organs, and their surrounding environment—is reshaping the future of precision medicine. By integrating spatial biology, medical imaging, multimedia data, and artificial intelligence (AI), we can now unravel complex disease mechanisms and enhance real-time clinical decision-making with greater accuracy and context. This presentation showcases cutting-edge applications of AI-powered spatial analysis in acute stroke triage, tumor microenvironment profiling, and therapeutic discovery for oncology and neurodegenerative diseases. These case studies illustrate how spatial intelligence is driving breakthroughs in diagnostics and targeted interventions, forging a new paradigm that seamlessly bridges scientific discovery and clinical care.

Biography: Dr. Steve Wong is the John S. Dunn Presidential Distinguished Chair in Biomedical Engineering and Chief of Medical Physics at Houston Methodist. He also holds appointments at Weill Cornell Medicine, Baylor College of Medicine, MD Anderson, Rice, and others. Dr. Wong is internationally recognized for his pioneering work in medical imaging, AI in medicine, and systems biology. He has authored over 500 publications, led innovations at institutions including Bell Labs, UCSF, and Harvard, and founded multiple research centers. A Fellow of IEEE, AIMBE, and other prestigious organizations, he continues to advance science at the intersection of engineering and medicine.

Professor, Radiology, Harvard Medical School

Abstract: AI innovations in medical imaging and their clinical applications are beginning to transform how radiologists perform diagnostic image interpretations. Thoracic imaging is one of the leading subspecialties for AI innovations and applications, due to growing clinical demands for faster and more accurate interpretations of a large number of imaging studies day-to-day. As a thoracic radiologist at a tertiary cancer center and Deputy Editor of Radiology, I will introduce recent AI studies from our group and from others published in Radiology, to demonstrate the utility of AI applications in the thoracic imaging topics including chest radiograph interpretations, lung cancer screening, and interstitial lung diseases.

Biography: Dr. Mizuki Nishino is a Professor of Radiology at Harvard Medical School and a staff radiologist at both Brigham and Women’s Hospital and Dana-Farber Cancer Institute. She serves as the Vice Chair of Faculty Development in the Department of Imaging. Dr. Nishino’s research focuses on imaging for precision oncology, particularly in lung cancer, and she has received multiple NIH grants supporting her work. She is an active member of the Fleischner Society and contributes to the Radiology editorial board.

Lawyer, MDA

Lawyer, MDA

Abstract: The rapid advancement of artificial intelligence (AI) technologies has necessitated the development of comprehensive regulatory frameworks to ensure ethical, safe, and effective deployment of AI models. This presentation delves into the critical regulatory guidelines that govern AI model development, spanning the entire lifecycle from initial research to final deployment. We will explore key considerations such as data privacy, bias mitigation, transparency, and accountability, highlighting best practices and compliance strategies. Attendees will gain insights into navigating the complex

Associate Professor, UTSW

Abstract: Artificial intelligence (AI) has immense potential to revolutionize thoracic radiology by improving diagnostic accuracy, enhancing workflow efficiency, and enabling personalized patient care. Despite these promises, translating AI innovations from research to routine clinical practice involves overcoming multiple challenges, including clinical validation, seamless integration into existing workflows, regulatory compliance, and sustainable reimbursement frameworks. In this lecture, we will discuss essential strategies for successfully addressing these barriers, drawn from direct experience implementing AI at our institution. We will highlight examples ranging from in-house developed AI models to externally developed research prototypes and commercially available FDA-cleared solutions. Through specific case studies, we will demonstrate practical approaches to clinical validation, illustrate various implementation models, and provide insights into the effective integration of AI tools within everyday clinical pathways. Our experience provides valuable lessons for radiologists, researchers, and healthcare administrators aiming to unlock the full potential of AI in thoracic imaging

Biography: ​Dr. Fernando Kay is an Associate Professor of Radiology at UT Southwestern Medical Center, where he serves as the Interim Chief of Cardiothoracic Imaging and Medical Director of Cardiothoracic MR at Parkland Health. His research focuses on cardiothoracic imaging, particularly the assessment of lung perfusion using advanced CT and MRI technologies. Dr. Kay has contributed to multiple peer-reviewed publications and is actively involved in integrating artificial intelligence into imaging practices.

MD Anderson Cancer Center, USA

Abstract: Artificial Intelligence is a promising technique for medical imaging applications with many new technologies and software applications on the horizon. However there are immense barriers when translating an idea in artificial intelligence into clinical practice. In this talk, we will explore the challenges encountered in each step of artificial intelligence development starting with (1) data curation (2) model training and development and (3) implementation and deployment. We will focus on a variety of neuroradiology and general radiology applications highlighting real-world examples of challenges and barriers encountered as we journey from pixels to practice.

Biography: Dr. Peter Kamel is an Assistant Professor in the Department of Neuroradiology at MD Anderson Cancer Center, recently transitioning from the University of Maryland School of Medicine. His research interest has focused on artificial intelligence in neuroimaging applications and clinical informatics. Dr. Kamel holds a bachelor’s degree in Computer Science from Rice University and completed his clinical training in Diagnostic Radiology and Neuroradiology at Johns Hopkins University School of Medicine. He is an active software developer, passionate about clinical informatics in medical imaging.