Our group applies AI to domains characterized by high data complexity, social relevance, and operational impact. From medical diagnostics and genomics to public health monitoring, business workflows, and policy intelligence, we study AI system design supporting analysis, interpretation, inference and decisions upon heterogeneous information sources. These scenarios often involve unstructured documents, noisy web data, or high-stakes decision-making, requiring robust and explainable models. We combine machine learning, natural language processing, and semantic technologies to address these challenges, supporting intelligent decision-making in sensitive and data-intensive interoperable contexts.
Deep Neural Modeling of Medical Problems
We design AI systems that integrate deep learning, natural language processing, and computer vision to analyze and interpret complex medical and administrative data. Our expertise includes semantic processing of digitized medical documents, enabling intelligent access to structured information from scanned texts. We develop end-to-end pipelines for document quality assessment, text restoration, information extraction, and semantic indexing, making data accessible via natural language queries and semantic search. These systems are capable of identifying key medical concepts—such as symptoms, pathologies, or administrative metadata—through robust understanding of both visual and linguistic content. Our work enables large-scale reasoning and monitoring over document archives, supporting decision-making and governance in healthcare and public administration.
AI for Epidemic Intelligence
Our group leverages AI to support the early detection, monitoring, and classification of emerging public health threats. We design semantic analysis tools and large language model-based systems that can interpret open web content, scientific articles, and news data to identify patterns and events relevant to epidemic surveillance. This includes building interpretable, automated pipelines for topic detection, risk categorization, and information triage. Our expertise merges language technology with decision support, enabling timely and data-driven responses in public health contexts.
Complex predictions in Genomics
We develop advanced AI models for genomic and transcriptomic data analysis, focusing on tasks such as cancer classification, gene marker selection, and phenotype prediction. Our work bridges deep learning with explainable AI to ensure interpretability in high-stakes applications like precision medicine. We are experienced in handling noisy, high-dimensional, and heterogeneous biological datasets, applying representation learning techniques to identify meaningful patterns in genetic sequences. These models support biomedical research and clinical decision-making with a focus on reliability and transparency.
ML for Business Processes Modeling
We apply machine learning to optimize and automate complex business processes, particularly in data-rich and multilingual environments. Our research focuses on document intelligence, semantic search, and workflow analysis, combining NLP and predictive modeling to enhance process visibility and decision support. We build intelligent systems that adapt to domain-specific knowledge and organizational goals, ensuring that models remain interpretable and efficient. This enables businesses to scale analytics and automate information flow in a sustainable and strategic way.
Predictivity over Web Data
We design machine learning and NLP systems capable of extracting predictive signals from large-scale web data, including social media, news articles, and user-generated content. Our models handle the dynamic, informal, and multilingual nature of web texts to support tasks such as sentiment forecasting, trend detection, and opinion mining. This area combines temporal analysis, text understanding, and domain adaptation to enable early detection of emerging phenomena in digital communication. We emphasize ethical and robust modeling for real-world, high-variability data sources.
Big Data Analysis for Tourism
We leverage AI to analyze tourism-related big data, providing insights into visitor behavior, destination reputation, and trend prediction. Our work integrates text mining, predictive analytics, and automated classification of user feedback and media content to support tourism management. We develop scalable, explainable models that assist decision-makers in identifying strengths, weaknesses, and opportunities for growth. By combining structured indicators with unstructured online content, we contribute to smarter, data-driven tourism policies and sustainable destination planning.
Knowledge Integration in the Nuclear Energy Domain
We specialize in applying AI techniques—especially NLP and knowledge extraction—to the processing and management of complex, high-value technical documentation. Our methods support semantic indexing, regulatory information retrieval, and domain-specific vocabulary modeling, enabling intelligent knowledge management systems in safety-critical contexts. We develop neural and symbolic approaches for extracting, organizing, and representing knowledge in ways that support transparency and traceability. This research fosters AI solutions that enhance decision support, compliance, and operational efficiency in sensitive sectors.