❤️ Cardiovascular Risk Prediction System

AI-Powered Cardiovascular Disease Prediction using AutoML + Explainable AI

🌐 Live Demo

Try the Application Here ➡️ https://cardiovascular-risk-prediction-using.onrender.com

🎥 Project Demo

Interface.Demo.Video.mp4

❤️ Predict • Prevent • Protect

An intelligent healthcare prediction platform that combines
Machine Learning + H2O AutoML + Explainable AI + Streamlit + Docker Deployment

📖 Overview

Cardiovascular diseases remain one of the leading causes of death globally.

Early prediction and preventive healthcare can significantly improve patient outcomes and reduce mortality.

This project presents an AI-powered Cardiovascular Risk Prediction System capable of:

🧠 Predicting cardiovascular disease risk in real-time
⚡ Automatically training multiple machine learning models using H2O AutoML
🔍 Explaining AI predictions using SHAP Explainable AI
🌐 Delivering predictions through an interactive Streamlit dashboard
☁️ Deploying seamlessly using Docker on Render Cloud Platform

The system transforms raw health metrics into meaningful medical intelligence.

🎯 Problem Statement

Traditional healthcare prediction systems face several challenges:

❌ Manual diagnosis processes
❌ Limited real-time prediction systems
❌ Black-box AI models with no explainability
❌ Lack of accessible healthcare intelligence platforms
❌ Difficulty interpreting ML-based medical decisions

This project solves these challenges using:

✅ AutoML for intelligent model selection
✅ Explainable AI for transparency
✅ Real-time prediction system
✅ Cloud-based deployment
✅ Interactive healthcare dashboard

💡 Proposed Solution

The proposed system integrates:

🧠 Intelligent AI Prediction Engine

Uses H2O AutoML
Automatically trains and compares multiple models
Selects the best-performing algorithm
Generates accurate real-time predictions

🔍 Explainable AI Module

Using SHAP (SHapley Additive Explanations):

Explains why predictions are high-risk or low-risk
Shows feature contribution importance
Improves transparency in medical AI systems

📊 Interactive Health Dashboard

Built using Streamlit:

Real-time cardiovascular prediction
BMI calculation
Health metric visualization
User-friendly interface

🏗️ Detailed System Architecture

flowchart TB

%% USER LAYER
A[User Inputs Health Metrics]

%% FRONTEND
A --> B[Streamlit Frontend Dashboard]

%% PREPROCESSING
B --> C[Data Preprocessing Pipeline]

C --> C1[Missing Value Handling]
C --> C2[BMI Calculation]
C --> C3[Feature Encoding]
C --> C4[Data Transformation]

%% AUTOML
C --> D[H2O AutoML Engine]

D --> D1[Random Forest]
D --> D2[Gradient Boosting Machine]
D --> D3[XGBoost]
D --> D4[Deep Learning]
D --> D5[Stacked Ensembles]

%% MODEL EVALUATION
D --> E[Leaderboard Evaluation]

E --> F[Best Performing Model]

%% PREDICTION
F --> G[Prediction Engine]

G --> H{Cardiovascular Risk Classification}

H -->|Low Risk| I[Healthy Prediction]
H -->|High Risk| J[High Risk Alert]

%% EXPLAINABILITY
G --> K[SHAP Explainable AI]

K --> K1[Feature Importance]
K --> K2[SHAP Visualization]
K --> K3[Prediction Interpretation]

%% OUTPUT
I --> L[Final Dashboard Output]
J --> L
K --> L

%% DEPLOYMENT
L --> M[Docker Container]

M --> N[Render Cloud Deployment]

🧠 Explainable AI Architecture

flowchart LR

A[Prediction Model] --> B[SHAP Explainer]

B --> C[Feature Contribution Analysis]

C --> D1[Age Impact]
C --> D2[Blood Pressure Impact]
C --> D3[BMI Impact]
C --> D4[Cholesterol Impact]

D1 --> E[Prediction Explanation Dashboard]
D2 --> E
D3 --> E
D4 --> E

🧰 Tech Stack

Category	Technologies
Programming Language	Python 3.10
Frontend	Streamlit
Machine Learning	H2O AutoML
Explainable AI	SHAP
Data Processing	Pandas, NumPy
Visualization	Matplotlib, Seaborn
Deployment	Docker, Render
Version Control	Git & GitHub
ML Algorithms	GBM, Random Forest, XGBoost, Ensembles

📂 Dataset Information

The system uses a cardiovascular healthcare dataset containing approximately 70,000 patient records.

📊 Dataset Features

Feature	Description
Age	Patient Age
Gender	Male / Female
Height	Height in cm
Weight	Weight in kg
ap_hi	Systolic Blood Pressure
ap_lo	Diastolic Blood Pressure
Cholesterol	Cholesterol Level
Glucose	Glucose Level
Smoke	Smoking Status
Alcohol	Alcohol Consumption
Active	Physical Activity
BMI	Body Mass Index
Cardio	Target Variable

🔄 Machine Learning Workflow

1️⃣ Data Preprocessing

Tasks Performed

Handling missing values
Removing duplicates
Outlier filtering
BMI feature engineering
Data normalization
Feature encoding

2️⃣ H2O AutoML Training

The system automatically trains:

🌲 Random Forest
📈 Gradient Boosting Machines
⚡ XGBoost
🔀 Stacked Ensembles
🧬 Deep Learning Models

AutoML Handles

Hyperparameter tuning
Cross-validation
Model ranking
Leaderboard generation

3️⃣ Prediction Pipeline

User Input
   ↓
Feature Engineering
   ↓
H2O AutoML Leader Model
   ↓
Risk Prediction
   ↓
SHAP Explainability
   ↓
Dashboard Visualization

📊 Features

🧠 AI Prediction System

Real-time prediction
AutoML-powered intelligence
Leader model selection

📊 Health Analytics Dashboard

BMI calculator
Blood pressure monitoring
Cholesterol evaluation
Glucose analysis
Risk score generation

🔍 Explainable AI

SHAP feature importance
Transparent AI decisions
Risk interpretation

☁️ Cloud Deployment

Dockerized architecture
Render cloud hosting
Production-ready deployment

🖥️ Application Workflow

Health Data Input
        ↓
Streamlit Dashboard
        ↓
Data Preprocessing
        ↓
H2O AutoML Model
        ↓
Prediction Engine
        ↓
SHAP Explainability
        ↓
Risk Visualization Dashboard

📈 Model Performance

Metric	Score
Accuracy	~73%+
AUC Score	~0.80
Best Models	GBM / Ensemble Models

📊 Visualizations Included

📈 ROC Curve
🔥 Correlation Heatmap
📉 Histograms
📌 Scatter Plots
📦 BMI Distribution Plot
🧠 Feature Importance Graphs
✅ Confusion Matrix

⚙️ Installation

1️⃣ Clone Repository

git clone https://github.com/Keerthishreekesavan/AutoML-and-XAI-for-Cardiovascular-Risk-Prediction

2️⃣ Navigate to Project Directory

cd AutoML-and-XAI-for-Cardiovascular-Risk-Prediction

3️⃣ Create Virtual Environment

python -m venv venv

4️⃣ Activate Virtual Environment

Mac/Linux

source venv/bin/activate

Windows

venv\Scripts\activate

5️⃣ Install Dependencies

pip install -r requirements.txt

6️⃣ Run Streamlit Application

streamlit run app.py

📦 requirements.txt

streamlit
pandas
numpy
matplotlib
seaborn
h2o==3.46.0.6
shap
scikit-learn
xgboost
lightgbm

🐳 Docker Deployment

Build Docker Image

docker build -t cardio-risk-app .

Run Docker Container

docker run -p 8501:8501 cardio-risk-app

☁️ Render Deployment

Deployment Steps

Push project to GitHub
Create new Render Web Service
Select Docker Environment
Connect GitHub repository
Deploy application 🚀

🐳 Dockerfile

FROM python:3.10-slim

WORKDIR /app

COPY requirements.txt .

RUN apt-get update && apt-get install -y default-jdk

RUN pip install --no-cache-dir -r requirements.txt

COPY . .

EXPOSE 8501

CMD ["streamlit", "run", "app.py", "--server.port=8501", "--server.address=0.0.0.0"]

⚠️ Known Issues & Fixes

❌ H2O Java Error

Cannot find Java

✅ Solution

Install Java runtime or use Docker image with OpenJDK.

❌ H2O Version Mismatch

Found version X, running version Y

✅ Solution

Use identical H2O versions for training and deployment.

❌ Streamlit Styler Error

Avoid:

applymap()

Use:

map()

🚀 Future Enhancements

📱 Mobile Application
🧬 Deep Learning Integration
⌚ Wearable Device Support
🏥 Hospital EHR Integration
📊 Personalized Healthcare Recommendations
🌍 Multi-Disease Prediction Platform
☁️ CI/CD Pipeline Integration

⚠️ Disclaimer

This project is intended for:

Educational purposes
Research applications
Health awareness

🚫 This is NOT a substitute for professional medical diagnosis.

Always consult certified healthcare professionals for medical advice.

👨‍💻 Author

Keerthishree Kesavan 🌷

🔗 Links

🌐 Live Demo: https://cardiovascular-risk-prediction-using.onrender.com
💻 GitHub: https://github.com/Keerthishreekesavan/AutoML-and-XAI-for-Cardiovascular-Risk-Prediction
ℹ️ LinkedIn: https://www.linkedin.com/in/keerthishreekesavan/

📄 License

This project is licensed under the MIT License.

See the LICENSE file for more information.

⭐ Support

If you like this project:

⭐ Star this repository
🍴 Fork the project
🐛 Report issues
🚀 Contribute improvements

Name		Name	Last commit message	Last commit date
AutoML and XAI training.ipynb		AutoML and XAI training.ipynb
Dockerfile		Dockerfile
GBM_2_AutoML_1_20250624_202543		GBM_2_AutoML_1_20250624_202543
LICENSE		LICENSE
README.md		README.md
app.py		app.py
cardio_train.csv		cardio_train.csv
requirements.txt		requirements.txt
selected_features_cardio.csv		selected_features_cardio.csv

Latest commit

History

Folders and files

Repository files navigation

❤️ Cardiovascular Risk Prediction System

AI-Powered Cardiovascular Disease Prediction using AutoML + Explainable AI

🌐 Live Demo

🎥 Project Demo

❤️ Predict • Prevent • Protect

📖 Overview

🎯 Problem Statement

💡 Proposed Solution

🧠 Intelligent AI Prediction Engine

🔍 Explainable AI Module

📊 Interactive Health Dashboard

🏗️ Detailed System Architecture

🧠 Explainable AI Architecture

🧰 Tech Stack

📂 Dataset Information

📊 Dataset Features

🔄 Machine Learning Workflow

1️⃣ Data Preprocessing

Tasks Performed

2️⃣ H2O AutoML Training

AutoML Handles

3️⃣ Prediction Pipeline

📊 Features

🧠 AI Prediction System

📊 Health Analytics Dashboard

🔍 Explainable AI

☁️ Cloud Deployment

🖥️ Application Workflow

📈 Model Performance

📊 Visualizations Included

⚙️ Installation

1️⃣ Clone Repository

2️⃣ Navigate to Project Directory

3️⃣ Create Virtual Environment

4️⃣ Activate Virtual Environment

Mac/Linux

Windows

5️⃣ Install Dependencies

6️⃣ Run Streamlit Application

📦 requirements.txt

🐳 Docker Deployment

Build Docker Image

Run Docker Container

☁️ Render Deployment

Deployment Steps

🐳 Dockerfile

⚠️ Known Issues & Fixes

❌ H2O Java Error

✅ Solution

❌ H2O Version Mismatch

✅ Solution

❌ Streamlit Styler Error

🚀 Future Enhancements

⚠️ Disclaimer

👨‍💻 Author

Keerthishree Kesavan 🌷

🔗 Links

📄 License

⭐ Support

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages