Structure (WP's)

WP1 Management and coordinaton (AUEB-RC)

  • Provide overall management of the consortium by liaising with the steering committee and participants to allow them to meet the project objectives and achieve the deliverables on time
  • Ensure consortium compliance with EU regulations together with contractual and legal requirements
  • Establish a Scientific Advisory Board (SAB) and a Patient Advisory Group (PAG) to provide strategic advice and guidance
  • Organise annual meetings of participants, SAB and PAG members
  • Monitor and coordinate financial resources, including producing financial reports for the EC

WP2 Data management and research infrastructure (MAASTRO)

  • Prepare the data management plan
  • Unify datasets from different sources and curate and preprocess clinical, imaging, and genomic data
  • Publish the training repository according to FAIR guidelines
  • Perform federated external validation of the prediction models

WP3 Explainable AI prediction models and AI platform Development (HES-SO)

  • Develop novel methods to embed explainability into ML models.
  • Develop ML-driven dose prediction models.
  • Provide a proof-of-concept validation for making the ML model training process distributed through federated learning techniques for the case of geographically constrained data.
  • Integrate data, ML models with explainability into a user-friendly platform.
  • Use concepts from Transfer learning to explore generalizability of AI models to other cancer types.

WP4 Clinical trial design and implementation for validation of the AI solution (UNICANCER)

  • Write study documentation and gain ethics and regulatory approval in each participating country
  • Recruit 724 breast cancer patients into the trial over a 12 month recruitment window, randomising half to the control and test groups
  • Collect baseline medical, genetics and imaging data to enable individual risk prediction via WP3 models
  •  Provide the doctors and patients in the test group with individual risk predictions, also by means of the app developed in WP5
  • Conduct follow-up clinics at end of radiotherapy and one and two years later
  • Carry out the final data analysis in order to determine how risk prediction affects treatment and outcomes

WP5 Stakeholder co-design and communication package (CNR)

  • Apply, test and assess the co-design process with the communication package for predictive models.
  • Consider traditional evaluation metrics for "classical AI systems" and pursue additional ones pertinent to trustworthy- AI by following (i) principles of respect for human autonomy; (ii) principles of prevention on harm; (iii) principles of fairness; (iv) principles of explicability.
  • Involve different stakeholders e.g.: AI experts (ML programmers), data experts (physicians), AI novices (patients), psychologists and behavioural scientists to reach a holistic vision of different needs and requirements.
  • Foster AI transparency, user trust, bias mitigation, privacy awareness.

WP6 Fair, Transparent and Ethical AI (CENTAI)

  • Ensure that the project abides by key ethical concerns, including avoiding racial bias in the AI algorithms.
  • Define, implement and apply a pipeline for assessing potential fairness issues in clinical AI decision systems based on medical imaging.
  • Perform thorough evaluation of AI algorithms and models for the delivery of documentation, guidelines and good practices regarding fair, transparent and ethical decision support systems for the prediction of side effects of breast cancer radiotherapy treatment.

WP7 Health Economics (UM)

  • Assess the potential patient value of AI-assisted development of therapeutic strategies in clinical care for breast cancer patients, and explore that value for other cancer patients
  • Assess costs of AI-assisted development of therapeutic strategies and implementation in clinical practice.
  • Assess the cost-effectiveness of AI-assisted development of therapeutic strategies and its implementation in the clinic.
  • Develop roadmaps towards further development and implementation to support the most promising use of AI-assisted development of therapeutic strategies for breast cancer and other cancer patients.

WP8 Communication, Dissemination and Exploitation (MAASTRO)

  • Review and record project publications in scientific journals and presentations in conferences.
  • Develop and maintain project website with a section restricted to participants and a public section.
  • Disseminate project outcomes to important stakeholders including radiation oncologists, patients, patient advocacy groups, and the public and health policy makers.