Days
Hours
Minutes
Seconds
AIBio 2025 explores the transformative role of artificial intelligence (AI) in biomedical research, with a focus on medical imaging, multi-omics, clinical data, and digital health. Biomedical data is inherently complex, characterized by heterogeneity, high dimensionality, and scalability challenges, making it difficult to extract meaningful insights. AI provides powerful tools to address these challenges, driving breakthroughs in disease diagnostics, personalized treatment strategies, and healthcare efficiency.
This workshop emphasizes pathology and omics data, where AI has demonstrated immense potential in disease understanding, molecular profiling, and tissue analysis. However, AIBio 2025 also broadens its scope to include translational medicine, digital health, and the role of telecommunications technologies in biomedical AI. This includes advancements in AI-driven telemedicine, edge and cloud computing for biomedical data analysis, 5G/6G applications in digital health, and secure AI models for biomedical data over networks, enabling real-time diagnostics, remote patient monitoring, and scalable healthcare solutions.
AIBio 2025 also emphasizes the ethical considerations and interpretability of AI models in clinical settings, focusing on strategies to mitigate biases related to gender, ethnicity, and age through fair machine learning techniques. Ensuring that AI systems are transparent, interpretable, and equitable is vital for their integration into healthcare practices.
By bringing together experts from diverse fields, AIBio 2025 aims to bridge the gap between AI research, telecommunications, and real-world medical applications, driving practical and impactful advancements in healthcare.
The AIBio 2025 workshop invites researchers, clinicians, data scientists, and industry professionals to submit their latest findings on AI-driven biomedical research. We seek high-quality, original contributions addressing AI applications in biomedical data, including but not limited to:
Authors must submit original, unpublished research contributions in English, formatted according to the Springer Communications in Computer and Information Science (CCIS) series guidelines. Manuscripts should be prepared using the official Springer LaTeX or Microsoft Word templates, available at Springer’s Author Guidelines.
Authors can submit papers in the following categories:
Authors must submit their manuscripts electronically in PDF format via the official submission system: Chairing Tool AIBio2025.
The review process follows a double-blind policy, meaning that:
Each paper will be subject to a rigorous peer-review process by at least three experts in the field. Papers will be evaluated based on the following criteria:
Authors must ensure that their papers do not contain plagiarized content or overlapping submissions to other venues. Papers that do not comply with the formatting, length, or anonymization requirements will be rejected without review.
The AIBio workshop registration policy follows that of the main ECAI conference. Details can be found https://www.ecai2025.eu/registration.
To attend the AIBio workshop, at least one author of each accepted paper is required to register for the ONLY WEEKEND option by the early registration deadline. Authors also have the option to register for the ECAI main conference + WEEKEND. However, please note that registering solely for the ECAI main conference does not grant access to the AIBio workshop.
At least one author of each accepted paper must register for ONLY WEEKEND or ECAI main conference + WEEKEND by early registration deadline, and present their work at the workshop. The presentation is a mandatory requirement for inclusion in the final proceedings.
The conference Proceedings will be published and indexed by the Communications in Computer and Information Science (Springer CCIS) and indexed in major digital libraries, including: Scopus, EI-Compendex, DBLP, Google Scholar, Additional Information. Please note that for a paper to be published, at least one of its authors must register for ONLY WEEKEND or ECAI main conference + WEEKEND by early registration deadline.
All deadlines are at the end of the day specified, Anywhere on Earth (AoE) (UTC-12).
Dr Soumick Chatterjee is a Postdoctoral Researcher at Human Technopole, Milan, and a Lecturer in AI for Medical Imaging at Otto von Guericke University (OvGU), Germany. He earned his PhD in Computer Science (summa cum laude), with a focus on medical physics, from OvGU. His principal research field is the application of deep learning in medical imaging.
The focus of his PhD thesis was on addressing artefacts in MRI, specifically through undersampled reconstruction and retrospective motion correction. In his current postdoctoral research, he is working on learning latent phenotypes from multimodal imaging and identifying their relationships with genotypes. Dr Chatterjee's work also extends to other deep learning projects such as vessel segmentation, anomaly detection, and tumour classification. Acknowledging the "black-box" nature of many deep learning models, a significant part of his research is dedicated to building trust in these systems through enhanced interpretability, explainability, and uncertainty quantification.
Beyond academia, Dr Chatterjee has experience in technology entrepreneurship and is actively involved in the scientific community, serving as the lead organiser for the IEEE SMC's scientific school ISACT since 2021.
The large-scale digitisation of biological archives, encompassing everything from histology to whole-body imaging, now affords us a remarkable opportunity to probe the foundations of human health and disease. Yet, realising the full potential of these data has been persistently constrained by the fundamental bottleneck of manual, expert annotation - a process both laborious and inherently subjective. This talk explores a common strategic principle for overcoming this: using unsupervised AI to learn phenotypes directly from images, without annotation.
This talk will present two powerful, yet distinct, applications of this principle, each tackling a different biological scale. The first one delves into the microscopic world, where a self-supervised Vision Transformer interrogates histology to quantify pathology and predict local gene expression, directly linking tissue morphology to its underlying molecular state. The second project at the macroscopic scale. It uses a diffusion autoencoder on cardiac MRIs to distil novel, heritable phenotypes. These phenotypes are then taken forward to genome-wide association studies, linking whole-organ structure to population genetics and disease risk. Though employing different methods and investigating different scales, both lines of inquiry converge on the same fundamental goal: forging a robust, scalable link between morphology and genomics. Together, they demonstrate a powerful new paradigm for biological discovery.
Federico Cabitza (BSc, MEng, PhD) is an Associate Professor at the University of Milano-Bicocca, where he leads the Modeling Uncertainty, Decisions, and Interactions Laboratory (MUDILab) and teaches courses in human-computer interaction and decision support.
He has collaborated extensively with hospitals in Milan and co-founded the Medical AI Laboratory. His research focuses on designing and evaluating AI systems for healthcare decision-making and understanding their impact on organizations and workflows.
Author of over 150 publications, Prof. Cabitza has co-chaired international workshops, is listed among Stanford’s Top 2% Scientists, and co-authored the book Artificial Intelligence: The Use of the New Machines with Luciano Floridi.
Despite widespread use, AI systems in healthcare are often evaluated solely by accuracy. In this talk, Prof. Cabitza questions this approach, proposing a multidimensional framework for model evaluation.
Drawing from recent lab developments, he introduces new metrics and visualization tools that reflect data reliability, case similarity, and clinical utility. He will present a public platform that applies these insights, helping stakeholders understand AI behavior beyond averages—especially in uncertain and diverse clinical scenarios.
Dr. Francesco Martino holds a PhD in Surgical Pathology, with a focus on developing AI models to enhance the diagnostic quality of Oral Squamous Cell Carcinoma (OSCC).
From 2016 to 2021, he was affiliated with the Surgical Pathology group at the University of Naples Federico II, where his research spanned from the assessment of immunohistochemical (IHC) expression in OSCC specimens to the development of segmentation and classification models. His work culminated in the training of a Generative Adversarial Network (GAN) for the virtual staining of H&E images into IHC representations.
Since 2023, Dr. Martino has been working in Vienna as a Software Engineer, contributing to the development of platforms for managing and visualizing medical imaging data, including both radiology and pathology. His current focus lies in the standardization of imaging formats to improve interoperability across systems and domains.
As digital pathology enters a new era, the promise of AI-assisted diagnostics is often held back not by model performance, but by limited access to structured, interoperable data. In this talk, Dr. Francesco Martino explores how the lack of standardization in data formats remains a key obstacle to translating research models into clinical tools.
Bringing his experience in surgical pathology research and medical imaging software development, Dr. Martino will highlight the practical benefits of adopting the DICOM standard for digital pathology. He will demonstrate how standardization can improve data integration, enhance AI workflows, and support large-scale collaborations—ultimately accelerating the clinical impact of computational pathology.
The following experts form the program committee for AIBio 2025:
For any request you might have, please contact cristian.tommasino@unina.it.