PROstate cancer Multimodal Explainable Transferable Holistic Expert for Universal Stratification (PROMETHEUS)

PROMETHEUS is a European research consortium developing next-generation decision support for precision medicine in prostate cancer.
The project integrates imaging, clinical data, and analytical methods into a unified framework for personalized diagnosis and treatment selection. PROMETHEUS is funded by the European Innovation Council (EIC Pathfinder).

Multimodal decision support for personalized prostate cancer care
PROMETHEUS combines MRI, PSMA PET/CT, and longitudinal health records to generate patient-specific disease profiles. By integrating these data into disease trajectory models, the project enables more accurate identification of aggressive disease and supports the selection of appropriate treatment strategies.

The system provides transparent risk assessments with quantified uncertainty and clear links to underlying evidence. The approach aims to reduce unnecessary biopsies and overtreatment, and is validated across large European clinical cohorts

Work packages

WP1: Integration of multi-cohort clinical data
Harmonizes imaging, clinical, and outcome data from multiple centers into a unified dataset.
Creates the foundation for robust analyses by ensuring data quality, consistency, and interoperability across sites.

WP1.1: Establish compliant data access and governance
WP1.2: Harmonize and curate multimodal datasets

WP2: Multi-modal agent fundamentals
Defines the technical and methodological foundation for combined data analysis across modalities.
Enables consistent integration and comparison of imaging, clinical and longitudinal patient data.

WP2.1: Establish benchmarking and experimental infrastructure
WP2.2: Define architecture for integrated data analysis

WP3: Knowledge-based decision support
Links clinical data to guidelines and evidence to support structured decision-making.
Ensures that analyses are aligned with established clinical knowledge and current best practice.

WP3.1: Build a structured knowledge base from guidelines and studies
WP3.2: Connect clinical data to evidence-based decision pathways

WP4: Clinical decision support interface
Develops tools for presenting patient data, risk assessments and evidence in a clear clinical context.
Facilitates interpretation and use of results in routine clinical decision-making.

WP4.1: Integrate results into a clinician-facing dashboard
WP4.2: Enable interactive exploration of patient data and evidence

WP5: Diagnostic modelling
Develops models to identify clinically significant prostate cancer and support biopsy decisions.
Focuses on improving detection of aggressive disease while reducing unnecessary procedures.

WP5.1: Detect and classify lesions across imaging modalities
WP5.2: Estimate risk of clinically significant disease

WP6: Diagnostic evaluation and optimisation
Validates and refines diagnostic performance across datasets.
Ensures robustness, reliability and clinical relevance of diagnostic models.

WP6.1: Optimise model performance and robustness
WP6.2: Evaluate accuracy and clinical utility

WP7: Treatment stratification
Develops models to support selection among treatment strategies.
Enables prediction of outcomes to guide personalised treatment decisions.

WP7.1: Model patient trajectories and outcomes
WP7.2: Compare outcomes across treatment options

WP8: Treatment model evaluation
Assesses performance and usability of treatment stratification tools.
Ensures that models are clinically meaningful and applicable in practice.

WP8.1: Validate across clinical cohorts
WP8.2: Evaluate impact on treatment decisions

WP9: External validation
Tests performance and generalisability in independent real-world cohorts.
Confirms applicability across different populations, healthcare systems and data sources.

WP9.1: Validate diagnostic models externally
WP9.2: Validate treatment models externally

WP10: Synthetic data generation
Develops methods to generate realistic data for development and validation.
Supports data sharing and method development while preserving patient privacy.

WP10.1: Generate synthetic imaging and clinical data
WP10.2: Assess quality and representativeness

WP11: Model optimisation and sustainability
Improves efficiency and evaluates resource use for clinical implementation.
Ensures that methods are scalable, cost-effective and suitable for long-term use.

WP11.1: Optimise computational performance
WP11.2: Assess environmental and operational impact