Alessandro Ceresi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Alessandro Ceresi has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Alberto Belmonte-Hernández
Grupo de Aplicación de Telecomunicaciones Visuales (GATV), Universidad Politécnica de Madrid, 28040 Madrid, Spain
2 shared publications - Anaida Fernández García
Grupo de Aplicación de Telecomunicaciones Visuales (GATV), Universidad Politécnica de Madrid, 28040 Madrid, Spain
1 shared publication Borja Arroyo Galende
1 shared publication- Borja Arroyo Galende
Information Processing and Telecommunications Center, ETS Ingenieros de Telecomunicacin, Universidad Politcnica de Madrid, Madrid, Spain
1 shared publication - Federico Alvarez
Universidad Politecnica de Madrid, Madrid 28040, Spain
1 shared publication Federico Álvarez
1 shared publication
Latest Publications
- Open Access
- 27 August 2026
- Published by ScienceOpen
BNNPath: Uncertainty-Aware Hypothesis Generation for DrugRepurposing using Bayesian Neural Networks
Drug repurposing methods based on biomedical knowledge graphs typically prioritize drug– disease associations but provide limited insight into the biological mechanisms supporting their predictions. This work presents BNNPath, an uncertainty-aware framework that formulates mechanistic drug repurposing as a learned shortest-path problem over a heterogeneous biomedical knowledge graph. A Bayesian neural network assigns non-negative, query-conditioned traversal costs to graph edges using node, relation, direction, drug, and disease representations. These costs guide bidirectional Dijkstra search to produce discrete and inspectable mechanism-of-action paths. The model is trained with complementary global path-ranking and local edge-selection objectives using expert-curated mechanisms, while Bayesian regularization enables posterior uncertainty estimation. On the held-out test set, the combined model achieved an exact path-match hit rate of 0.7436, outperforming variants trained with only local supervision or only path-level supervision. Analysis of non-exact predictions showed that they frequently preserved the expert path’s entity-type sequence and selected intermediate nodes that were substantially closer to the reference mechanisms in embedding space than randomly generated alternatives. Monte Carlo sampling was further used to quantify confidence in predicted paths from edge-cost variability. The proposed framework provides an efficient and interpretable foundation for uncertainty-aware mechanistic hypothesis generation, expert review, and experimental prioritization in drug repurposing.
- 27 August 2026
- Published by ScienceOpen
Open AccessBNNPath: Uncertainty-Aware Hypothesis Generation for DrugRepurposing using Bayesian Neural Networks
Drug repurposing methods based on biomedical knowledge graphs typically prioritize drug– disease associations but provide limited insight into the biological mechanisms supporting their predictions. This work presents BNNPath, an uncertainty-aware framework that formulates mechanistic drug repurposing as a learned shortest-path problem over a heterogeneous biomedical knowledge graph. A Bayesian neural network assigns non-negative, query-conditioned traversal costs to graph edges using node, relation, direction, drug, and disease representations. These costs guide bidirectional Dijkstra search to produce discrete and inspectable mechanism-of-action paths. The model is trained with complementary global path-ranking and local edge-selection objectives using expert-curated mechanisms, while Bayesian regularization enables posterior uncertainty estimation. On the held-out test set, the combined model achieved an exact path-match hit rate of 0.7436, outperforming variants trained with only local supervision or only path-level supervision. Analysis of non-exact predictions showed that they frequently preserved the expert path’s entity-type sequence and selected intermediate nodes that were substantially closer to the reference mechanisms in embedding space than randomly generated alternatives. Monte Carlo sampling was further used to quantify confidence in predicted paths from edge-cost variability. The proposed framework provides an efficient and interpretable foundation for uncertainty-aware mechanistic hypothesis generation, expert review, and experimental prioritization in drug repurposing.
- Open Access
- 26 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Deep Generative Models Meet Federated Learning: A Healthcare-Centered Review
The integration of Artificial Intelligence (AI) in healthcare holds significant promise, yet remains constrained by data scarcity, privacy and ethical concerns, and the inherent complexity of medical tasks. This review explores the intersection of deep generative modeling and federated learning, two complementary paradigms that enable collaborative and privacy-preserving innovation. We survey recent trends in deep generative models applied to healthcare, with a particular focus on federated settings. The review is structured around model architectures, providing a taxonomy-driven analysis that emphasizes solutions tailored to the heterogeneity of healthcare data. Although this review initially intended to cover multimodal applications, the current literature is largely limited to single-modality models. Among deep generative approaches, Generative Adversarial Networks dominate the field, particularly in domains such as medical image synthesis and data augmentation. In contrast, alternative architectures, including variational autoencoders, diffusion models, and autoregressive models, remain barely explored in federated healthcare scenarios. Notably, the number of related publications has increased substantially in recent years, rising from 1 in 2020 to 16 in 2024, reflecting the growing interest and momentum in this area. This paper highlights key challenges, open research questions, and future directions for the development of trustworthy, distributed generative AI in healthcare.
- 26 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessDeep Generative Models Meet Federated Learning: A Healthcare-Centered Review
The integration of Artificial Intelligence (AI) in healthcare holds significant promise, yet remains constrained by data scarcity, privacy and ethical concerns, and the inherent complexity of medical tasks. This review explores the intersection of deep generative modeling and federated learning, two complementary paradigms that enable collaborative and privacy-preserving innovation. We survey recent trends in deep generative models applied to healthcare, with a particular focus on federated settings. The review is structured around model architectures, providing a taxonomy-driven analysis that emphasizes solutions tailored to the heterogeneity of healthcare data. Although this review initially intended to cover multimodal applications, the current literature is largely limited to single-modality models. Among deep generative approaches, Generative Adversarial Networks dominate the field, particularly in domains such as medical image synthesis and data augmentation. In contrast, alternative architectures, including variational autoencoders, diffusion models, and autoregressive models, remain barely explored in federated healthcare scenarios. Notably, the number of related publications has increased substantially in recent years, rising from 1 in 2020 to 16 in 2024, reflecting the growing interest and momentum in this area. This paper highlights key challenges, open research questions, and future directions for the development of trustworthy, distributed generative AI in healthcare.
- Open Access
- 19 August 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 271, 109019
Radiomics feature analysis for survival prediction in multiple myeloma: An automated PET/CT approach
- 19 August 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 271, 109019
Open AccessRadiomics feature analysis for survival prediction in multiple myeloma: An automated PET/CT approach