RAIDO - Reliable AI and Data Optimisation

RAIDO is developing a unified platform for ‘Trustworthy’ and ‘Green’ AI. This platform tackles challenges in data quality, model efficiency, explainability, and energy-efficient deployment. It improves AI training data through automated enrichment, Digital Twins, and diffusion models, ensuring high-quality, unbiased, and representative datasets.

To support efficient AI training, RAIDO introduces data- and compute-efficient models, using techniques such as few-shot learning, model distillation, and continual learning. A novel AI Orchestrator optimises dataset creation and training pipelines, adapting AI workflows to specific applications without compromising performance.

A scalable and future-proof AI framework

For Trustworthy AI, RAIDO focuses on explainability, fairness, and transparency. It integrates explainable AI (XAI) techniques, decentralised blockchain, benchmarking, and feedback-driven improvements to enhance trust and accountability in AI decision-making.

Energy efficiency is at the core of RAIDO’s end-to-cloud (E2C) AI deployment, where an AI Orchestrator automates and optimises processes, reducing energy consumption and environmental impact.

The platform is tested in smart grids, smart farming, healthcare, and robotics, demonstrating its versatility and societal impact. RAIDO builds on advancements from previous AI, digital twin, and sustainability projects, positioning itself as a scalable and future-proof AI framework.

VITO's role in RAIDO

VITO is a key partner in the RAIDO project, leading efforts to ensure privacy-compliant AI development, data governance, and regulatory alignment. Our role includes developing guidelines for responsible data usage, assessing the reusability of personal data for scientific research, and analysing how EU and national legal frameworks impact AI-driven data processing. We also explore the use of synthetic data and digital twins as privacy-preserving alternatives, ensuring compliance with GDPR and ethical AI principles across different jurisdictions.

Additionally, VITO is developing the Pharmacogenomics (PGx) AI system, which supports personalised medicine by predicting drug responses based on genetic and clinical data. Current PGx reports are often static and difficult to interpret, making it challenging for healthcare professionals to use them effectively. 

By leveraging large language models (LLMs), we create reliable AI modules that process PGx reports, whole genome sequencing (WGS) data, and clinical guidelines to provide clinically relevant insights. This system aims to improve drug selection, patient safety, and communication of PGx information in real-world healthcare settings.

Once integrated with RAIDO’s explainability (XAI), bias detection, privacy-preserving learning, and monitoring components, this use case will serve as a real-life example of AI-driven pharmacogenomics. By embedding AI-assisted PGx reporting into a broader trustworthy AI framework, VITO helps make personalised medicine more reliable, accessible, and impactful for healthcare professionals and patients alike.

Project partners

Meet all the project partners on the project website.