Organizers
- Domenico Amato, Dipartimento di Matematica e Informatica, University of Palermo
- Salvatore Calderaro, Department of Physics and Chemistry Emilio Segrè, University of Palermo
- Tiziana Currieri, Department of Biomedicina, Neuroscience and Advanced Diagnostics, University of Palermo
- Giosuè Lo Bosco, Dipartimento di Matematica e Informatica, University of Palermo
- Francesco Prinzi, Department of Biomedicina, Neuroscience and Advanced Diagnostics, University of Palermo
- Salvatore Vitabile, Department of Biomedicina, Neuroscience and Advanced Diagnostics, University of Palermo
Abstract
While significant advances have been achieved in the design of machine learning and neural network models for predictive medicine, several methodological challenges remain in translating algorithmic advances into reliable and clinically meaningful predictions. This special session aims to address methodological and application-oriented aspects of predictive artificial intelligence, where topics of interest range from classification models to predictive learning approaches and how they can be rigorously evaluated, compared, and validated when applied to complex biomedical data.
The session aims to bring together researchers interested in neural network–based methodologies for predictive modelling. Consideration should be given to performance evaluation, robustness, generalization, and learning under uncertainty. Research should be submitted for innovative methods for biomedical data analysis, including medical image analysis, feature learning and representation strategies, and multimodal learning frameworks integrating imaging, clinical, and biological data.
Further, it is possible to investigate methodological challenges like reproducibility, dataset or distributional shift, imbalance in classes, or predictive confidence estimation. This special session emphasizes methodological rigor and application-driven evaluation, complementing architecture-focused and purely application-driven research and fosters a deeper discussion on the development of effective predictive AI methods for preventive and early diagnostic medicine.
List of Topics Covered in the Special Session
Possible topics of interest include (not limited to):
- Predictive AI and Classification Models for Preventive and Early Diagnosis
- Innovative Machine Learning Methods for Biomedical Data Analysis
- Methods for Medical Image Analysis in Predictive Modeling
- Feature Learning, Feature Extraction and Representation Strategies for Medical Imaging
- Multimodal Learning for Integrating Imaging, Clinical, and Biological Data
- Explainable and Interpretable Classification Models for Healthcare
- Uncertainty-Aware and Confidence-Based Predictive Models