🏥 Five Scottish Cancer Centres feed PROSECCA dataset

PROSECCA will link pseudoanonymised EHR, imaging, radiotherapy plans and outcomes for up to 15,000 Scottish prostate cancer patients treated with radiotherapy (2010–2022) across five Scottish Cancer Centres, aiming to build AI/ML models that predict toxicity and treatment response. The study protocol (trial registration NCT06714630) received NHS Scotland Public Benefit and Privacy Panel approval on July 1, 2024, enabling use of unconsented record data.

Why It Matters To Your Practice

• Many prostate cancer patients live 10+ years; preventing permanent GU/GI toxicity after radiotherapy is increasingly central to quality of survivorship.

• Current care cannot reliably identify which patients will have poorer outcomes or higher side-effect risk after radiotherapy—this project is designed to close that gap using routinely collected data.

• Practice impact could be broad because the dataset spans primary, secondary and tertiary care plus radiotherapy-specific data, not just a single clinic or trial cohort.

Clinical Implications

• Risk stratification: models could flag patients at higher predicted risk of late toxicity or poor response before treatment starts, supporting more individualized consent discussions and follow-up intensity.

• Treatment adaptation: predictions could inform dose/plan personalization (e.g., balancing tumor control vs. normal-tissue constraints) rather than “one-size-fits-most” planning.

• Workflow reality check: any eventual tool will depend on data completeness and interoperability across EHR systems, imaging archives and radiotherapy planning platforms.

Insights

• Scale and linkage are the differentiators: PROSECCA’s value proposition is connecting routine longitudinal records with radiotherapy planning and outcomes to discover de novo predictive biomarkers of radiation response.

• Governance is a key enabling step: the July 2024 HSC-PBPP approval is what makes population-level, unconsented pseudoanonymised linkage feasible in Scotland.

• Expect model-development challenges clinicians will care about: dataset shift across centres/time, missingness, and the need for calibration and clinically usable uncertainty reporting before deployment.

The Bottom Line

• PROSECCA is building a real-world, multi-centre Scottish radiotherapy dataset (up to 15,000 cases) to develop AI/ML tools that predict prostate radiotherapy response and toxicity.

• If successful, it could move radiotherapy planning toward individualized dosing and earlier mitigation of life-limiting side effects—especially important in long-surviving patients.