Industry Day

NVIDIA

Wednesday | April 8, 2026 | 8:30 – 9:15

Abstract: Clinical datasets increasingly contain diverse modalities, including medical imaging, genomics, spatial transcriptomics, clinical notes, and electronic health records (EHR). Integrating these heterogeneous data sources has the potential to improve clinical diagnosis and optimize healthcare workflows. In this talk, I will present recent research and foundation models developed at NVIDIA to integrate biomedical data across molecular, cellular, and patient levels. The approach leverages vision–language models and intermediate modality fusion strategies to align imaging, text, and structured clinical information within a shared representation space. By enabling cross-modal reasoning across imaging, molecular data, and clinical records, the framework aims to build more transparent and interpretable AI systems that help clinicians understand complex patient information. Such capabilities may support clinical decision-making, facilitate biomarker discovery, and advance data-driven clinical development.

Biography: Jonny is a Senior Solutions Architect in the Healthcare and Life Sciences team at Nvidia. Originally hired as a Deep Learning specialist, his role is about enabling scientists, clinicians and developers to accelerate their work using the latest tools and techniques. During his eight years at Nvidia he has covered many aspects of the HCLS landscape, from genomics to radiology. Computational histopathology has been a constant theme in his work and he was instrumental in the formation of the MONAI pathology working group. With a software and algorithms past, Jonny has worked on the optimisation of many workloads but has a keen interest in research – still managing to get an occasional paper co-authorship.

GE Healthcare

Wednesday | April 8, 2026 | 9:16 – 10:00

Abstract: The rapid expansion of AI in medical imaging requires close collaboration among research institutions, industry teams, and healthcare providers to ensure the safe and effective translation of algorithms into clinical practice. In this talk, Dr. Hongxu Yang will provide an overview of the GE HealthCare AI team and its role within the broader European ecosystem, highlighting cross-sector collaborations and contributions to EU-level innovation initiatives. Building on research-driven proof-of-concept studies, the second part of the talk will present several synthetic data–based methodologies developed within ongoing projects. These examples will serve as a foundation for discussing practical considerations related to the use of synthetic medical data in industrial applications. Through these case studies, the talk will outline key challenges and emerging opportunities for accelerating AI model development in industry, with particular emphasis on improving model robustness, enhancing development efficiency, and reducing overall cost.

Biography: Hongxu Yang, Ph.D., is an AI Scientist at GE HealthCare specializing in medical imaging and clinically aligned artificial intelligence for healthcare applications. His work focuses on developing reliable imaging algorithms that enhance diagnostic workflows, increase research throughput, and promote the safe and ethical use of synthetic medical data. Dr. Yang received his Ph.D. in Electrical Engineering from Eindhoven University of Technology. He collaborates closely with cross-functional clinical, scientific, and engineering teams to translate advanced algorithms from research prototypes into product-oriented solutions. He has been a key contributor to several AI-based products, an inventor on more than twenty patent applications, and an author of multiple publications in leading venues on medical imaging with deep learning.

Newton’s Tree

Wednesday | April 8, 2026 | 10:30 – 11:15

Abstract: The evidentiary paradigm for clinical artificial intelligence (AI) remains heavily weighted toward retrospective performance validation, often using single-site datasets and accuracy-based endpoints. However, once deployed into live clinical pathways, AI systems operate within complex sociotechnical environments where distributional shift, workflow interaction, automation bias, and data quality variability can materially affect safety and effectiveness.

Drawing on experience designing national AI trials and leading real-world multi-site evaluations, this presentation will examine methodological approaches for generating robust post-deployment evidence for clinical AI systems. Topics will include:

• Prospective and stepped-wedge cluster trial designs for pathway-level evaluation
• Continuous performance monitoring and drift detection
• Alignment with clinical risk management standards (e.g., DCB0129/0160, ISO 13485 lifecycle principles)

Using examples from imaging AI and large language model–based clinical documentation systems, I will argue for a transition from pre-market performance validation toward continuous, deployment-centered evidence generation. Establishing scalable real-world evaluation frameworks will be essential to ensuring the safety, effectiveness, and health economic value of AI systems integrated into routine care.

Biography: Haris Shuaib is founder & CEO of Newton’s Tree, and AI deployment and governance platform hospitals. Additionally, he is a consultant clinical scientist at Guy’s & St Thomas’ Hospitals where he used to lead the clinical scientific computing team. He also sits on the MHRA’s Commission for the Regulation of AI in healthcare.

Microsoft Research

Wednesday | April 8, 2026 | 11:15 – 12:00

Abstract: Foundation models have shown strong potential for medical image understanding, yet most systems remain specialised interpretation. In this talk, I will present COLIPRI, a method to build vision–language foundation models that integrate 3D images and clinical reports for accurate clinical interpretation.

Biography: Fernando is a Senior Researcher at Microsoft Research Health Futures. His work focuses on vision–language foundation models for healthcare and their translation to clinical practice.

Prior to joining Microsoft, he was at the Centre for Neuroimaging Research at the Paris Brain Institute, building histological and MRI brain atlases for deep brain stimulation. He then moved on to UCL and King’s College London for his PhD in Medical Imaging, where he investigated the potential of AI to improve the treatment of epilepsy, developing open-source software tools such as TorchIO in the meantime.

Siemens Healthineers

Wednesday | April 8, 2026 | 13:30 – 14:15

Abstract: Artificial intelligence in medical imaging continues to advance rapidly, with impressive technical performance demonstrated across a wide range of research studies. However, translating these innovations into routine clinical practice remains challenging. This talk will explore the journey from imaging AI research to real‑world deployment, drawing on industry experience across scientific collaboration, validation, regulatory readiness, and clinical adoption. Key themes will include aligning technical development with clinical workflows, navigating evidence generation and trust, and understanding where industry, academia, and healthcare systems must collaborate more effectively to deliver meaningful patient impact. The session will highlight lessons learned from current imaging AI implementations and outline practical considerations for researchers aiming to see their work adopted at scale.

Biography: Dr Craig Buckley is a senior leader at Siemens Healthineers GB&I, with over 13 years’ experience driving innovation across engineering, service management, and scientific leadership roles. He holds responsibility for enabling high‑impact clinical and research collaborations across Great Britain and Ireland, working closely with healthcare providers, academia, and industry partners to translate advanced imaging technologies into real‑world outcomes.
Craig has led multiple multi‑million‑pound Innovate UK–funded programmes, supported the development of commercially relevant PhD partnerships, and advises on national and international imaging initiatives through roles on several strategic advisory boards. He is a long‑standing reviewer for translational funding programmes and brings a strong track record of aligning research, technology, and commercial strategy. Previously, Craig held academic and industrial roles focused on the development and integration of next‑generation imaging technologies.

Novo Nordisk

Wednesday | April 8, 2026 | 14:16 – 15:00

Abstract: Recent progress in vision foundation models is creating new opportunities for cellular imaging, with the potential to move beyond narrowly trained models toward reusable systems that support a broad range of research and drug discovery tasks.

Biography: Dr. Navid Alemi is a Principal Computer Vision Scientist at Novo Nordisk working at the intersection of AI, cellular imaging, and drug discovery. With over eight years of experience in healthcare AI, he has led scientific teams and delivered projects from early proof of concept to production-scale solutions. Previously, he worked in computational pathology, developing models for cancer diagnosis and prognosis, and his current interests include foundation models and deployable models for biomedical imaging.

University College London, UK

Wednesday | April 8, 2026 | 15:30 – 16:15

Abstract: Medical foundation models have achieved substantial progress in ophthalmic image analysis. However, most existing models remain task-specific and typically demonstrate optimal performance only under idealized and controlled conditions. In real clinical environments, heterogeneous data sources, diverse imaging devices, and complex multimodal information pose significant challenges, and the generalization capability and reliability of current systems remain limited.

This talk focuses on extending beyond the boundaries of specialized models to develop more generalizable ophthalmic intelligence systems. It outlines the evolutionary pathway from expert models to versatile systems, and presents research and practical implementations of highly generalizable frameworks in ophthalmology. Large-scale AI–expert interaction studies and real-world multi-center validation cases are showcased to demonstrate system robustness and clinical applicability. Beyond assisting in ophthalmic diagnosis, the system also lays a foundation for future oculomics research, offering new technical support for uncovering intrinsic links between retinal biomarkers and systemic health. Using Airdoc as a representative example, the talk further illustrates how algorithms are translated into deployable products, serving millions of individuals in China through retinal disease screening and systemic disease risk management.

Biography: Dr. Lie Ju is a Postdoctoral Research Fellow at University College London under the supervision of Prof. Pearse Keane and serves as an Honorary Research Fellow at Moorfields Eye Hospital. He obtained his PhD from Monash University in 2024, supervised by A/Prof. Zongyuan Ge and Prof. Paul Bonnington.

His research focuses on open-world learning, long-tail learning, and efficient learning under label noise in both medical and natural image domains. He also has extensive practical experience in deploying multimodal understanding systems and agent-based frameworks in ophthalmology. As first and corresponding author, he has published in leading venues including IEEE Transactions on Medical Imaging, IEEE Journal of Biomedical and Health Informatics, MICCAI, and AAAI. He serves as a reviewer for several top-tier journals and conferences, including Nature Medicine, TMI, Medical Image Analysis, MICCAI, and CVPR.

Dr. Ju also brings substantial industry experience in product development, with a strong emphasis on translating cutting-edge research into real clinical and industrial applications. He holds more than 20 granted invention patents and software copyrights.