Workshop about Robust AI for Healthcare (RAH)

About IJCNN

IJCNN is the premier international conference in the area of neural networks theory, analysis and applications. Since its inception, IJCNN has been playing a leading role in promoting and facilitating interaction among researchers and practitioners, and dissemination of knowledge in neural networks and related facets of machine learning. And Rome with its history and geographical position will further contribute to grow and maintain the role of the IJCNN as a prominent platform for exchange of knowledge in neural networks and artificial intelligence.

IJCNN 2025 will be held in the year and the city of the Jubilee, making it an even more special occasion. Hosted at the Pontifical Gregorian University, just a coin's throw from the iconic Trevi Fountain, attendees will have the unique opportunity to experience Rome's rich history and culture. The Eternal City will serve as an open-air museum, providing an inspiring backdrop for our discussions and networking.

About the Workshop

The workshop “Robust AI in Healthcare” aligns with IJCNN’s mission to advance neural network research with meaningful practical applications. Focusing on robustness, this workshop explores strategies to ensure safe, reliable, and efficient AI models in healthcare. The workshop will help to bridge the gap between theoretical advances and practical implementation significantly. The workshop will also feature discussions among clinicians, data scientists, and AI researchers to address real-world implementation challenges. Your contributions are critical to improving AI systems in healthcare. We look forward to welcoming you to Rome.

Workshop Organizers

Prof. Despina Kontos: Professor of Radiological Sciences (in Radiology) and of Biomedical Informatics, Columbia University Irving Medical Center, USA.

Associate Prof. Seyedmehdi Payabvash: Department of Radiology at Columbia University Irving Medical Center, USA.

Assistant Prof. Dr. Junhao (Hao) Wen: Assistant Professor of Radiological Sciences, Columbia University Irving Medical Center, USA.

Associate Research Scientist Dr. Anh Tuan Tran: Department of Radiology at Columbia University Irving Medical Center, USA.

Teams

1. Dr. Phi Le, Data Scientist, University of California San Francisco, USA
2. Dr. Nam Bui, Assistant Prof., University of Colorado Denver, USA
3. Dr. Le Van Vinh, Head of Information Technology Faculty, HCMC University of Technology and Education, VN
4. Dr. Gia An Vien, Dongguk University (South Korea), Camera AI Research Engineering (Finland)
5. Dr. Huy Quang Ung, KDDI Research, Inc., Japan
6. Dr. Van-Dung Hoang, Associate Prof., HCMC University of Technology and Education, VN
7. Dr. Tran Quang, Head of AI Department, HCMC University of Technology and Education, VN
8. Dr. Tran A Tuan, University of Science, Vietnam National University

Registration https://2025.ijcnn.org/registration

Accepted papers

1. Fulcrum Rebalancing and Hybrid Classification for Multi-class Multi-labeled ECG
Ayan Mukherjee (Tata Consultancy Services) Varsha Sharma (Tata Consultancy Services) Anirban Dutta Choudhury (Tata Consultancy Services) Chirayata Bhattacharyya (Indian Institute of Science) Aniruddha Sinha (Tata Consultancy Services)
2. Quality and Robustness of Generative Watermarking on Biomedical Data
Liyue Fan (UNC Charlotte) Luca Bonomi (Vanderbilt University)
3. Beyond Accuracy: The Role of Calibration in Computational Pathology
João Nunes (INESC TEC) Diana Montezuma (IMP Diagnostics) Domingos Oliveira (IMP Diagnostics) Tania Pereira (INESC TEC) Inti Zlobec (University of Bern) Jaime Cardoso (INESC TEC)

No presentation
4. A Method for Explainable Medical Abstracts Classification through Transformer Models
Luca Petrillo (IMT School for Advanced Studies Lucca, IIT-CNR) Anna Giacomello (University of Molise) Fabio Martinelli (ICAR-CNR) Antonella Santone (University of Molise) Mario Cesarelli (University of Sannio) Francesco Mercaldo (University of Molise, IIT-CNR)

Preliminary Schedule (half-day session)

Room: Montale

June 30th, from 09:30 AM to 1:00 PM

09:30 - 09:45: Welcome and Introduction
09:45 - 10:15: Invited Speakers 1: Associate Professor Seyedmehdi Payabvash
10:15 - 10:30: Invited Speakers 2: Pursuing the SOTA of your AI models? Lessons from Alzheimer’s disease classification (Assistant Professor Junhao Wen)
10:30 - 11:00: Invited Speakers 3: Rad4XCNN: Bridging Deep Learning and Radiomics for Explainable and Accurate AI in Medical Image Analysis (Assistant Professor Francesco Prinzi)
11:00 - 11:20: Oral Presentations 1 (will be updated)
11:20 - 11:40: Oral Presentations 2
11:40 - 12:00: Oral Presentations 3
12:00 - 12:30: Invited Speakers 4: Towards Resilient AI in healthcare (Prof. Paolo Soda, Università Campus Bio-Medico di Roma)
12:30 - 13:00: Discussion

Invited Speakers, Abstract

Title: Rad4XCNN: Bridging Deep Learning and Radiomics for Explainable and Accurate AI in Medical Image Analysis

Speaker: Assistant Professor Francesco Prinzi, Department of Biomedicine, Neuroscience and Advanced Diagnostics at the University of Palermo, Italy
As machine learning models become increasingly integrated into clinical workflows, their "black-box" nature poses a major barrier for a real adoption in clinical practice This talk provides a concise overview of current strategies for explaining deep neural networks in medical imaging, including saliency-based and radiomic approaches. I will introduce Rad4XCNN, a novel framework that combines the high predictive power of convolutional neural networks with the transparency of radiomic features. Unlike traditional visualization methods, Rad4XCNN offers global and meaningful interpretations without sacrificing accuracy, providing a step forward in building trustworthy AI for healthcare.

Title: Towards Resilient AI in healthcare

Speaker: Prof. Paolo Soda, Unit of Artificial Intelligence and Computer Systems, Dept. of Engineering, Università Campus Bio-Medico di Roma; Dept. of Diagnostics and Intervention, Umeå University
We need AI systems able to operate in challenging, noisy, and uncertain real-world settings: while this may occur in many fields, it is particularly demanding in healthcare. This talk will present and reflect on our research on resilient AI, spanning from medical imaging to clinical data, reflecting also on the paths forward towards multimodal learning and generative AI to shape a future where personalized healthcare will provide every individual with the right type of care in the right way and at the right time.

Call for Submissions

We invite submissions on any aspect of Robust AI in Healthcare. We welcome research contributions related to the following (but not limited to) topics: Models, Strategies improving robustness; Adversarial techniques; Advanced optimization; Learning; Interpretability; Uncertainty.

Prospective authors are invited to submit complete papers of no more than eight (8) pages in the conference proceedings format, according to the IJCNN Paper Submission Guidelines. Papers must be submitted through the IJCNN 2025 CMT System and will undergo the same rigorous evaluation process as regular papers.

Instruction and Submit Your Paper

Workshop on Robust AI for Healthcare (RAH)