Cardiovascular Disease Dashboard

Cardiovascular Disease Dashboard with Predict Risk

Comment

Inspiration

Cardiovascular disease is the leading cause of death worldwide, yet early risk detection is often limited to clinical environments. I wanted to explore how machine learning can make cardiovascular risk prediction more accessible and understandable.

My goal was to build an interpretable AI-based system using real biomedical data and turn it into an interactive tool that demonstrates how predictive modeling can support preventive healthcare.

What it does

Cardiovascular Disease Dashboard is a machine learning-powered web application that predicts cardiovascular disease risk based on patient health metrics.

The system:

  • Trains a Random Forest classifier on 70,000 real patient records
  • Achieves a ROC-AUC score of 0.787
  • Identifies key predictors such as systolic blood pressure and cholesterol
  • Provides an interactive web-based risk calculator

Users can input health parameters (age, blood pressure, cholesterol, glucose level, lifestyle factors), and the application returns a predicted cardiovascular risk probability.

How I built it

I built the project using:

  • Python
  • Pandas for data preprocessing
  • Scikit-learn for machine learning
  • Random Forest classifier
  • Matplotlib for feature importance visualization
  • Joblib for model serialization
  • Streamlit for deploying the web interface

Steps:

  1. Cleaned and processed a semicolon-separated biomedical dataset
  2. Removed irrelevant features (such as ID)
  3. Split the data into training and testing sets (80/20)
  4. Trained a Random Forest model
  5. Evaluated performance using ROC-AUC
  6. Analyzed feature importance
  7. Saved the trained model and deployed it in a Streamlit web app

Challenges I ran into

Some of the main challenges included:

  • Handling non-standard CSV formatting
  • Debugging feature mismatches between training and deployment
  • Ensuring the model received exactly the same input structure in the web app
  • Understanding how small preprocessing mistakes affect predictions

These challenges significantly improved my understanding of real-world ML system design.

Accomplishments that I'm proud of

  • Successfully training a predictive model on 70,000 real medical records
  • Achieving a solid ROC-AUC score of 0.787
  • Building a complete ML pipeline from preprocessing to deployment
  • Creating a working AI-powered cardiovascular risk calculator
  • Implementing feature importance for interpretability

What I learned

Through this project, I learned:

  • How critical proper data preprocessing is in machine learning
  • The importance of model interpretability in healthcare applications
  • How to deploy a trained model into a functional web application
  • The practical challenges of aligning training and inference pipelines

I also gained deeper insight into how cardiovascular risk factors interact and influence predictions.

What's next for Cardiovascular Disease Dashboard

Next, I plan to:

  • Perform hyperparameter optimization to improve performance
  • Implement SHAP-based explainability for individual predictions
  • Conduct fairness analysis across demographic groups
  • Categorize predictions into low / medium / high risk levels
  • Deploy the application to a scalable cloud environment

My long-term goal is to develop a fully interpretable, accessible AI-based health risk assessment platform.

Built With

Share this project:

Updates