Inspiration
I wanted to explore how machine learning and interactive 3D visualization could make cardiovascular risk easier to understand. This inspired me to build CADViz as a solo project for the Multimodal AI Hackathon 2026.
What it does
CADViz estimates overall coronary artery disease (CAD) risk and separate probabilities for the LAD, LCX, and RCA arteries. I combined an interactive 3D heart, dynamic visualizations, and explainable AI to help users understand how clinical inputs influence predictions.
How I built it
I independently developed the entire application using Python, NumPy, pandas, FastAPI, React, Vite, Three.js, and Recharts. I implemented four logistic regression models from scratch, along with model evaluation, probability calibration, and feature contribution explanations.
Challenges I ran into
As a solo developer, I handled machine learning, backend development, frontend design, and deployment myself. One major challenge was implementing the ML pipeline without scikit-learn and connecting predictions to a responsive 3D visualization.
Accomplishments I'm proud of
I'm proud to have independently built a complete full-stack prototype combining machine learning, explainable AI, and interactive 3D visualization. I integrated the individual components into one application with overall CAD and vessel-level predictions.
What I learned
I strengthened my understanding of machine learning algorithms, model evaluation, explainability, API integration, and 3D web development. I also learned the importance of communicating model limitations responsibly, especially in healthcare-related applications.
What's next for CADViz: Cardiovascular Risk Explorer
I plan to improve model validation using larger, independent datasets, refine probability calibration, enhance the 3D visualization, and improve the user experience. I also aim to seek feedback from medical professionals before considering any clinical application.
Disclaimer: CADViz is an educational prototype, not a diagnostic tool or substitute for professional medical care.
Log in or sign up for Devpost to join the conversation.