WP1 provides the overall scientific and management framework for LUMINA. It coordinates activities across the consortium, monitors progress towards the project’s objectives and ensures that the work packages remain scientifically and technically aligned.

The work package covers project governance, financial and administrative coordination, risk management, quality control and internal communication. It also oversees the project’s Data Management Plan and defines common approaches to data stewardship, FAIR data, data protection, access control and the reuse of research outputs.

By maintaining a shared overview of the project’s scientific, technical and organisational activities, WP1 helps the consortium manage interdependencies, address risks and deliver consistent, high-quality results. 

Key activities

• Scientific and operational coordination across the consortium
• Monitoring of progress, risks, milestones and deliverables
• Quality assurance and internal review of project outputs
• Data management, stewardship and compliance
• Coordination with the European Commission.

Lead

Aalborg University.

LUMINA is part of a broader portfolio of projects developing generative AI technologies for cancer diagnosis and treatment. WP2 coordinates LUMINA’s contribution to this portfolio and identifies opportunities for collaboration, knowledge exchange and joint innovation.

The work package supports common portfolio activities, stakeholder engagement, strategic planning and the sharing of research facilities, expertise and best practices. It also contributes to portfolio-level analysis of technological performance, competitiveness, intellectual property and potential routes towards future exploitation.

Through WP2, LUMINA will ensure that its data structures, models, interfaces and knowledge can contribute to a wider, collaborative ecosystem for AI-enabled cancer research. 

Key activities

• Participation in portfolio governance and working groups
• Collaboration with other projects in the Pathfinder Challenge
• Joint communication, dissemination and stakeholder activities
• Comparative technological and economic assessment
• Portfolio-level innovation and exploitation planning

Lead

Aalborg University and Politecnico di Milano.

WP3 brings together multimodal clinical data from participating hospitals in Denmark, Italy, Poland and South Korea. The purpose is to create a high-quality, privacy-compliant and clinically relevant data foundation for developing and testing the LUMINA models.

The participating clinical centres will retrieve, validate and prepare data from sources such as CT scans, pathological images, electronic health records, laboratory tests and other clinical information. Medical experts will annotate imaging data to create reliable reference information that the AI models can learn from.

WP3 will also identify relevant public datasets and assess whether they can be responsibly integrated with the clinical data collected by the consortium. Regulatory approvals, ethical requirements, data quality and patient privacy are addressed throughout the process.

Key activities

• Regulatory, ethical and privacy compliance at each clinical centre
• Retrieval and quality control of multimodal patient data
• Clinical annotation of CT scans and pathological images
• Collection and assessment of relevant public datasets
• Preparation of standardised data for subsequent AI development.

Lead

Maria Skłodowska-Curie National Research Institute of Oncology (MSCNRIO), Region Hovedstaden and Aalborg University.

Clinical data often differ across hospitals, countries, systems and medical modalities. WP4 addresses this challenge by developing the LUMINA Data Suite, a secure and interoperable framework for organising, harmonising and integrating multimodal lung cancer data.

A shared data schema will establish a common structure and terminology for the different data types used in LUMINA. CT scans, pathology images and clinical information will be processed and standardised so that they can be analysed together.

WP4 will also develop generative AI models capable of producing realistic synthetic multimodal data. These models will help the project investigate relationships across data modalities and address limitations caused by scarce, incomplete or difficult-to-share clinical data. Clinical experts will review the synthetic outputs as part of the development process.

The resulting Data Suite will provide structured, AI-ready information to the BRONCHI Digital Twin models developed in WP5. 

Key activities

• Development of a unified multimodal data schema
• Harmonisation and standardisation of clinical data
• Generative AI models for multimodal data synthesis
• Clinical review of synthetic data
• Development of the LUMINA Data Suite framework and pipelines.

Lead

Aalborg University and Humanitas Research Hospital.

WP5 develops the scientific and technological core of LUMINA: the BRONCHI Digital Twin methodology. It's purpose is to model how lung cancer may develop over time and how an individual patient may respond to a particular treatment.

The methodology combines longitudinal, multimodal patient data with a Physics-Informed Attention Network. This approach integrates data-driven AI with knowledge of biological and physical processes related to tumour development and treatment response.

The first models will focus on non-small cell lung cancer and immunotherapy. The methodology will then be extended to other lung cancer subtypes and treatment strategies. Additional models will examine patient-level risks such as metastasis, recurrence and long-term survival.

Explainable AI methods will be incorporated to help identify which data and clinical features contribute to a prediction. The aim is not only to generate predictions, but also to make the reasoning behind them more transparent and clinically meaningful. 

Key activities

• Set-up of high-performance computing environments
• Development of physics-informed attention models
• Prediction of cancer progression and treatment response
• Extension to additional lung cancer subtypes and treatments
• Patient-risk modelling
• Explainability and interpretation of AI-generated outputs.

Lead

Aalborg University and Cracow University of Technology.

WP6 evaluates the BRONCHI Digital Twin prototype in a controlled laboratory setting. The work package will assess whether the models perform reliably across different patient groups, clinical scenarios and data sources.

The validation process will combine existing multi-centre patient data, synthetic data and a prospective test dataset. Synthetic digital representations will be used to investigate how the models perform under different conditions, including variations in cancer type, treatment strategy and imaging protocol.

Clinicians will participate directly in the validation process. Radiologists, oncologists and other medical specialists will review the prototype’s outputs and assess their clinical relevance, interpretability and potential usefulness. Their feedback will be used to refine the underlying models and methodology.

WP6 will also document the validation process and examine alignment with relevant regulatory requirements, supporting the project’s longer-term pathway towards clinical research and development. The scope of LUMINA itself remains laboratory-scale validation at Technology Readiness Level 4. 

Key activities

• Definition of validation criteria and success measures
• Technical validation using synthetic data and digital representations
• Prospective clinician-in-the-loop validation
• Assessment of interpretability and potential clinical relevance
• Documentation aligned with regulatory requirements.

Lead

Region Hovedstaden and Aalborg University.

WP7 brings the central technological components of LUMINA together in an open and interoperable platform. It connects the LUMINA Data Suite with the BRONCHI Digital Twin prototype and establishes the infrastructure required to test the complete data and modelling workflow.

The prototypes will be deployed on Aalborg University’s CLAAUDIA computing platform. The technical architecture will use containerisation and orchestration technologies to support portability, reproducibility and scalability.

Secure interfaces will enable communication between the Data Suite and the BRONCHI Digital Twin. WP7 will then perform end-to-end testing to demonstrate that data can move through the system and be processed by the relevant AI and digital twin components.

The platform is designed to support collaboration and future testing within the Pathfinder Challenge portfolio. It will provide a laboratory-scale environment for exploring, testing and benchmarking the technologies developed by LUMINA. 

Key activities

• Establishment of the deployment environment on CLAAUDIA
• Deployment of the BRONCHI Digital Twin prototype
• Deployment of the LUMINA Data Suite prototype
• Development of secure Data-to-Digital-Twin interfaces
• End-to-end integration, documentation and testing.

Lead

Cracow University of Technology, Humanitas Research Hospital and Aalborg University.

WP8 ensures that knowledge and results from LUMINA reach the people and organisations that can use them. These include clinicians, AI and machine-learning researchers, hospitals, healthcare technology providers, cancer research communities, regulatory bodies, patients and people at risk of lung cancer.

Communication activities will explain the project’s purpose, progress and broader societal relevance through the LUMINA website, social media, newsletters, videos, media relations and stakeholder networks. The communication will be developed with sensitivity to people affected by cancer and with input from clinical experts and patient representatives.

Dissemination activities will make scientific findings and research outputs accessible to professional audiences through publications, conferences, repositories, workshops and engagement with standards and regulatory organisations.

WP8 will also develop the project’s exploitation strategy. This includes identifying valuable project results, examining intellectual property options and exploring possible pathways for further research, collaboration, clinical translation and innovation after the project. 

Key activities

• Development of the project’s communication and dissemination strategy
• Website, social media, newsletters, videos and media engagement
• Scientific publications, conferences and stakeholder workshops
• Open access to suitable research outputs
• Stakeholder engagement and knowledge transfer
• Exploitation planning, intellectual property and sustainability.

Lead

Politecnico di Milano and Aalborg University.