AI-Powered Health Tool

Know your
stroke risk
before it's too late.

Every 4 minutes, someone dies from a stroke — yet 80% are preventable. Enter a few details and our model estimates your likelihood instantly.

Check My Risk →
97%
Model accuracy
8
ML models tested
80%
Strokes preventable

Interactive Tool

Check your stroke risk

Our AdaBoost model (97.3% accurate) estimates your risk from a few key inputs.

45
100
24

Estimated stroke risk

—

⚠ This tool is for informational purposes only and does not constitute medical advice. Always consult a licensed healthcare professional.

Methodology

Eight models. One winner.

We trained and evaluated eight ML classifiers on the Kaggle stroke dataset. Expand any card for accuracy, or click to see the full breakdown and confusion matrix.

AdaBoost
★ Best
Iterative ensemble that learns from weak classifiers' mistakes.
97.3%
Random Forest
Ensemble of decision trees; robust to noise and overfitting.
~96%
Decision Tree
Flowchart-style splits; highly interpretable and fast.
~96%
SVM
Finds the optimal hyperplane separating stroke vs. no-stroke.
~94%
KNN
Classifies by majority vote of k nearest neighbours in feature space.
~93%
Logistic Regression
Classic binary classifier; strong, interpretable baseline.
79.3%
Bagging
Bootstrap aggregation across many Decision Tree subsets.
~96%
Neural Network
Multi-layer architecture inspired by the human brain's neurons.
~95%

Background

Built in one week.

This project was built during a one-week intensive at AI Camp in 2023. Starting from scratch with Python, we sourced a stroke dataset from Kaggle, trained and benchmarked eight machine learning models, and shipped this site — all in seven days. Our final AdaBoost model hit 97.3% accuracy, one of the highest recorded for this dataset.