PREDI-Care
A multimodal oncology prototype comparing rectal-cancer treatment scenarios, with €100,000 allocated by the foundation to begin prototyping.
· Medical AI, Multi-Agent Systems, Python, Streamlit, Hackathon
Research-to-product prototypeCopy link to section
We built PREDI-Care during the A.R.CA.D 2026 hackathon to explore how clinical, radiological, and biological information could be presented together when comparing rectal-cancer treatment scenarios.
What the prototype demonstratesCopy link to section
The Streamlit application compares surgery and watch-and-wait scenarios, displays interactive risk and survival visualizations, and exports individual or cohort summaries. Its engine combines an optional medical LLM with a heuristic fallback.
FundingCopy link to section
Following the A.R.CA.D hackathon, the foundation announced an allocation of €100,000 to the project to begin prototyping, confirmed to our team in April 2026.
Current limitsCopy link to section
This is a hackathon demonstration, not a clinically validated decision system. The fallback scores are heuristic and the SHAP-like explanations are simulated feature contributions. Real clinical use would require trained models, calibration, external validation, and medical oversight.