CardioPulse-AI is an end-to-end healthcare analytics and machine learning platform designed to analyze cardiovascular health data and predict heart disease risk.
The project combines data preprocessing, exploratory data analysis, statistical validation, machine learning, and interactive business intelligence dashboards to transform raw clinical data into meaningful healthcare insights.
Developed as an academic team project, CardioPulse-AI simulates a real-world healthcare data science workflow used in predictive analytics and clinical decision-support systems.
- End-to-end healthcare analytics pipeline
- Heart disease risk prediction using Machine Learning
- Advanced data preprocessing and feature engineering
- Exploratory Data Analysis (EDA)
- Statistical validation using Chi-Square Testing
- Logistic Regression and Random Forest models
- Feature importance analysis
- ROC-AUC and Confusion Matrix evaluation
- Interactive Power BI Dashboard
- Healthcare-focused visual analytics
- Analyze real-world cardiovascular healthcare data
- Perform comprehensive Exploratory Data Analysis (EDA)
- Discover clinical risk factors associated with heart disease
- Validate statistical relationships between variables
- Train and compare machine learning models
- Evaluate predictive performance using healthcare ML metrics
- Generate actionable healthcare insights
- Build an integrated cardiovascular analytics system
Heart Failure Prediction Dataset (Kaggle)
| Metric | Value |
|---|---|
| Initial Records | 918 |
| Records After Cleaning | 836 |
| Target Variable | HeartDisease |
| Positive Class | Heart Disease Present |
| Negative Class | No Heart Disease |
- Age
- Sex
- Chest Pain Type
- Resting Blood Pressure
- Cholesterol
- Fasting Blood Sugar
- Resting ECG
- Maximum Heart Rate
- Exercise-Induced Angina
- Oldpeak
- ST Slope
- Missing value handling
- Invalid value correction
- Outlier detection using IQR
- Feature scaling using StandardScaler
- One-hot encoding of categorical variables
- Histograms
- Boxplots
- Correlation Heatmaps
- Distribution Analysis
- Target Variable Analysis
Chi-Square testing was applied to determine statistically significant relationships between categorical clinical features and heart disease occurrence.
Used as the baseline classification model.
Used as the primary predictive model for heart disease classification.
Evaluation metrics included:
- Accuracy
- Precision
- Recall
- F1-Score
- ROC-AUC
- Confusion Matrix
- Cross Validation
| Model | Accuracy | ROC-AUC |
|---|---|---|
| Logistic Regression | 89% | 0.953 |
| Random Forest | 92% | 0.959 |
Random Forest achieved the highest predictive performance and demonstrated superior generalization capability on unseen data.
The most influential features included:
- ST Slope
- Maximum Heart Rate
- Cholesterol
- Age
- Chest Pain Type
The interactive Power BI dashboard enables:
- KPI summaries
- Risk factor exploration
- Clinical trend analysis
- Correlation visualization
- Model performance monitoring
- Feature importance interpretation
The dashboard allows healthcare data exploration through dynamic filtering and visual analytics.
Data Collection
↓
Data Preprocessing
↓
Exploratory Data Analysis
↓
Statistical Testing
↓
Model Training
↓
Model Evaluation
↓
Power BI Dashboard
↓
Healthcare Insights
- Python
- Pandas
- NumPy
- Scikit-learn
- Logistic Regression
- Random Forest
- Matplotlib
- Seaborn
- Power BI
- Jupyter Notebook
- Git
- GitHub
CardioPulse-AI
│
├── data
│ ├── processed
│ └── raw
│
├── diagrams
│ ├── DFD_Level_1.png
│ └── System_Workflow_Diagram.png
│
├── frontend
│ ├── models
│ │ ├── random_forest_model.pkl
│ │ └── scaler.pkl
│ │
│ ├── static
│ │ ├── css
│ │ └── js
│ │
│ └── templates
│ └── index.html
│
├── notebooks
│ └── HeartDisease_DS.ipynb
│
├── powerbi
│ └── CardioPulse_AI_Analytics_Dashboard.pbix
│
├── report
│ └── CardioPulse_AI_Data_Science_Project_Report.pdf
│
├── app.py
├── README.md
├── LICENSE
└── .gitignore
- Healthcare Data Analytics
- Data Cleaning & Preprocessing
- Exploratory Data Analysis
- Statistical Testing
- Feature Engineering
- Machine Learning Model Development
- Model Evaluation & Validation
- Power BI Dashboard Development
- End-to-End Data Science Workflow
A complete project report documenting methodology, analysis, model development, evaluation, and conclusions.
Location: report/CardioPulse_AI_Data_Science_Project_Report.pdf
Interactive analytics dashboard for exploring cardiovascular health patterns and model insights.
Location: powerbi/CardioPulse_AI_Analytics_Dashboard.pbix
Contains workflow and DFD diagrams describing the architecture and analytical pipeline.
Location: diagrams/
- Abdul Ahad
- Muhammad Wafa Abbas
- Abdul Rehman
Department of Computer Science
git clone https://github.com/ahadbuilds/CardioPulse-AI.gitcd CardioPulse-AIpip install -r requirements.txtjupyter notebook- Deep Learning Models
- Explainable AI (XAI)
- Real-Time Prediction Dashboard
- Cloud Deployment
- Model Monitoring
- Healthcare Recommendation System
This project is intended solely for educational, research, and portfolio purposes.
It is not designed for medical diagnosis, treatment planning, or real-world clinical decision making.
Suggestions, improvements, and collaboration opportunities are always welcome.
Feel free to connect, contribute, or provide feedback.
This project is intended for educational, research, and portfolio purposes.