Overview

Multimodal Artificial Intelligence (AI) is an emerging field in high-performance computational sciences that integrates multiple data streams —such as text, images, videos, audio, and numerical data— to enhance information extraction, inference accuracy, and bias reduction. By synthesizing diverse data sources, multimodal AI provides a more comprehensive representation of complex physical, medical, and societal processes.

In mission-critical domains like healthcare, multimodal AI has the potential to revolutionize medical analytics, improving disease detection, prediction, diagnosis, risk stratification, referrals, and clinical decision-making. As modern healthcare systems generate vast and diverse datasets—including medical reports, clinical notes, radiology images, physician dictations, patient audio recordings, physiological signals, and genomic data — AI models must effectively process and integrate these different data types, simulating how the human brain synthesizes multiple sensory inputs for decision-making.

Despite its advantages, multimodal AI faces challenges in data alignment across modalities, requiring well-annotated datasets and advanced embedding techniques. Additionally, integrating domain-specific medical knowledge is crucial for clinically meaningful interpretations.

This Workshop will showcase the latest advancements in multimodal AI-driven biomedical research and healthcare applications, addressing key challenges, emerging solutions, and future opportunities.

Submission Guidelines

We welcome submissions of Original Research Articles, Reviews, and Perspectives that align with the specified themes and cover a wide range of theoretical and practical aspects, technologies, and systems.

Submissions should adhere to the Topics List and contribute novel insights, innovative methodologies, or practical applications that advance the fields of health and medicine. We encourage papers that showcase cutting-edge research, comprehensive reviews, and insightful perspectives on these critical areas.

All submissions must be original, unpublished, and not under consideration elsewhere. Papers should be written in English and must not have been published or accepted in any substantially similar form in peer-reviewed venues, including journals, conferences, or workshops.

Each submission will undergo a rigorous peer-review process conducted by the Organizers and the Technical Program Committee to ensure high-quality contributions.

Submitted papers must not exceed 12 pages (including references). Only original contributions that have not been previously published or simultaneously submitted to other venues will be considered for inclusion in the proceedings. All submissions should be made in PDF via the EasyChair system.

We suggest workshop papers are prepared and submitted using this template in LNCS format.

List of Topics

The Workshop invites contributions on all aspects of multimodal AI models for image analysis in healthcare and biomedical applications, with a focus on (but not limited to):

• Data Fusion Techniques. Techniques for integrating diverse data sources — such as genomics, electronic health records (EHRs), and wearable sensors — with image and video data in biomedicine.

• AI in Multimodal Medical Imaging. Advanced techniques for integrating imaging data (e.g., MRI, CT, X-ray) with genomic, phenotypic, and clinical information to enhance diagnostic accuracy and provide deeper clinical insights for personalized medicine.

• Multimodal AI in Remote and Telemedicine Applications. AI-driven integration of data from telemedicine platforms, remote sensors, and patient-reported outcomes for long-distance clinical care.

• Multimodal AI for Assistive Tools. Enabling users to interact with technology through various modalities, such as voice commands and visual gestures, enhancing accessibility and user experience. 

• Natural Language Processing (NLP) and Computer Vision (CV) Integration. Combining NLP and CV to extract insights from clinical notes and medical images, enhancing decision support and clinical analysis.

• Multimodal Large Language Model (LLM) and Conversational AI. AI-driven multimodal approaches that combine textual and visual inputs to answer questions and offer comprehensive insights, supporting medical professionals and caregivers in their decision-making processes.

• Resilient Multimodal Artificial Intelligence. Developing systems that operate effectively in challenging, noisy, incomplete, and uncertain real-world biomedical settings.

• Explainable AI (XAI) in Multimodal Healthcare Systems. Methods for enhancing the transparency and interpretability of multimodal AI models, ensuring that clinicians and patients can trust AI-driven decisions.

• Ethical and Fairness Challenges in Multimodal AI. Addressing bias, data privacy, and the ethical challenges that arise when using multimodal datasets in healthcare applications.

Organizing Committee

• Ciro Mennella, ICAR-CNR, Email: ciro.mennella@icar.cnr.it
• Umberto Maniscalco, ICAR-CNR, Email: umberto.maniscalco@icar.cnr.it
• Aniello Minutolo, ICAR-CNR, Email: aniello.minutolo@icar.cnr.it
• Massimo Esposito, ICAR-CNR, Email: massimo.esposito@icar.cnr.it

Publication

The accepted papers for mAI4Health will be included in a joint Post-Workshop proceeding published by Springer Lecture Notes in Computer Science (LNCS), indexed in Web of Science, Scopus, EI Engineering Index, Google Scholar, DBLP, etc. 

Venue

The conference will be held at the 23rd International Conference on Image Analysis and Processing (ICIAP 2025)

Contact

All questions about submissions should be emailed to ciro.mennella@icar.cnr.it