Predict YourHeart Attack Riskwith AI
Built on the Cleveland Heart Disease dataset (303 patients). Input your health parameters and get an instant cardiovascular risk assessment powered by Logistic Regression.
About This Project
Built by Team PulseML as a mini research project, CardioSense AI compares multiple machine learning approaches for cardiovascular risk prediction.
Random Forest
An ensemble of 100 decision trees using majority voting. Handles imbalanced data with exceptional resilience.
K-Nearest Neighbors
Classifies using k=5 nearest neighbors via Euclidean distance. Simple yet the most accurate model.
Support Vector Machine
Finds the optimal hyperplane using RBF kernel. Achieved the highest ROC-AUC score of 93%.
Deep Learning (CNN)
Sequential model built with TensorFlow/Keras. 64→32→16→1 neurons, ReLU + Sigmoid, 50 epochs.
How It Works
Input Health Data
Preprocessing
KNN Model
Risk Prediction
Risk Assessment
Enter patient data to predict heart disease risk using our trained ML model
📊 Model Information: This prediction is based on a Logistic Regression model trained on the Cleveland Heart Disease dataset (303 patients, 13 clinical features) with approximately 85% accuracy.
Model Performance
Logistic Regression trained on Cleveland Heart Disease dataset (303 patients, 13 features)
Training Accuracy
Test Accuracy
Precision
86%
Recall
81%
F1-Score
83%
AUC-ROC
89%
Dataset Overview
303
Total Patients
13
Clinical Features
~85%
Overall Accuracy
📊 Dataset Information: Cleveland Heart Disease dataset from UC Irvine Machine Learning Repository. The model uses 13 clinical features including age, sex, chest pain type, resting blood pressure, cholesterol, fasting blood sugar, resting ECG, maximum heart rate, exercise-induced angina, ST depression, ST slope, number of major vessels, and thalassemia type.