AI for Data Science &​ Advanced Analytics​

Data science and advanced analytics form the backbone of modern AI-enabled science. This review examines how AI transforms the scientific data lifecycle, from data acquisition and preparation to analysis, sharing, interpretation and knowledge discovery. The review covers machine learning, multimodal AI, knowledge graphs, agentic workflows, federated analytics, scientific data spaces and advanced decision-support approaches, with a focus on how AI enables the generation of actionable scientific knowledge from increasingly complex and distributed data ecosystems.

• Agentic data acquisition and analysis​
• Multimodal AI and foundation models​
• Intelligent data discovery and metadata generation​
• Federated and distributed intelligence​
• Knowledge graphs and explainable AI​
• Scenario-based analytics, simulation and digital twins​

• Data quality and interoperability challenges​
• Governance, privacy and security requirements​
• Explainability, provenance and validation needs​
• Unequal access to AI and computing resources​
• Lack of regulatory and ethical clarity​

AI is increasingly reshaping the entire scientific data lifecycle rather than individual analytical tasks. The review highlights the emergence of agentic orchestration, where AI systems coordinate data acquisition, preparation, analysis, sharing and knowledge generation across multiple stages of the research process. It supports more adaptive, scalable and automated scientific workflows while helping researchers manage growing volumes of heterogeneous and distributed data. The topic is directly linked to SCIANCE activities on data-driven discovery, machine-actionable science, federated infrastructures and future AI-enabled research ecosystems.

European science benefits from a strong ecosystem of data and computing infrastructures, including EOSC, OpenAIRE, European Data Spaces, Copernicus, Destination Earth, EuroHPC and disciplinary research infrastructures. AI Factories and Antennas, Data Labs and the RAISE Observatory are expected to strengthen Europe’s ability to support interoperable, federated and machine-actionable scientific data ecosystems, while enabling large-scale AI adoption across disciplines.

• Agentic orchestration across the entire data lifecycle​
• AI-enabled data spaces and federated ecosystems​
• Multimodal scientific foundation models​
• Knowledge-driven and machine-actionable science​
• Conversational analytics and natural-language interfaces​
• Human-AI collaborative decision support​

Data science is evolving from isolated analytical methods towards integrated AI-enabled knowledge ecosystems. Future progress will depend not only on advances in AI models, but also on trusted infrastructures, interoperable data ecosystems, explainability, governance and human oversight. The greatest impact may come from AI systems capable of coordinating the full scientific data lifecycle while remaining transparent, trustworthy and aligned with scientific values.