Inspiration
NSMP endometrial cancer represents a particularly challenging group for risk assessment and treatment planning, as it lacks the clear prognostic pathways available for other molecular subtypes. ENDO-INSIGHT was inspired by the need to support clinicians managing this specific subgroup, by providing transparent, interpretable tools that combine prediction with probabilistic reasoning.
What it does
ENDO-INSIGHT is a clinical decision-support tool designed for NSMP endometrial cancer. It provides disease-free survival estimates at 1, 3, and 5 years using a survival analysis model, and complements these predictions with a Bayesian Network that enables probabilistic reasoning under uncertainty. The platform supports risk stratification, exploration of how different variables influence outcomes, and interpretation of results in a clinically meaningful way.
How we built it
The backend is implemented in Python using FastAPI and serves two complementary models. A Cox proportional hazards model is used to estimate disease-free survival and stratify patients into relative risk groups. A Bayesian Network models probabilistic relationships between clinical, pathological, and biomarker variables, allowing flexible inference even when some information is missing. A Node.js frontend provides an interactive interface for data entry and visualizes results through survival curves, risk categories, and probability summaries.
Challenges we ran into
Beyond dealing with missing and incomplete clinical data, one of the main challenges was selecting appropriate modeling approaches that balanced predictive performance, interpretability, and clinical relevance. Another significant challenge was designing frontend visualizations that clearly communicate survival estimates, risk groups, and uncertainty in a way that is intuitive and useful for clinicians without requiring technical expertise.
Accomplishments that we're proud of
We successfully combined different modeling paradigms within a single platform, using each model for what it does best. ENDO-INSIGHT integrates predictive survival modeling with probabilistic reasoning and presents results visually in a way that is easy to understand for both technical and non-technical users. The platform emphasizes transparency, interpretability, and uncertainty representation, which are essential for clinical adoption.
What we learned
Through this project, we gained a deeper understanding of statistics and data analysis in the medical domain, particularly survival analysis and probabilistic modeling. We learned how critical model interpretability, uncertainty handling, and clinically meaningful visualization are when developing machine learning tools for healthcare.
What's next for ENDO-INSIGHT - Hack the Uterus!
Future work will focus on improving the frontend functionality and user experience, deploying the platform on a production server, and enabling broader access. We also plan to further refine the models, incorporate additional data sources, and explore clinical validation to move toward real-world deployment.
Log in or sign up for Devpost to join the conversation.