Chronic Disease Prediction using Classical Machine Learning

An AI-powered health screening system that predicts risk for three major chronic diseases (diabetes, heart disease, and hypertension) using interpretable classical machine learning algorithms. The system provides real-time risk assessment with explainable predictions through a user-friendly Streamlit dashboard.

---

Overview

This project is a production-ready, interpretable machine learning system for predicting the risk of three major chronic diseases:

• Diabetes
• Heart Disease
• Hypertension

It uses classical machine learning algorithms (Random Forest, Gradient Boosting, Logistic Regression) and features a user-friendly, privacy-focused Streamlit dashboard for real-time, explainable predictions.

---

Key Features

• Multi-disease Prediction: Risk profiling for diabetes, heart disease, and hypertension.
• Explainable AI: Shows which factors (e.g., glucose, cholesterol, family history) drive each prediction.
• Offline Capable: Works fully offline once setup is complete.
• Privacy-First: All processing is local. No patient data is sent externally (HIPAA/GDPR aligned).
• Automated Setup: One-command pipeline for model training and app launch.
• Ready for Clinics and Research: Suitable for point-of-care screening, rural health, and researcher use.
• Mobile-Responsive Dashboard: Streamlit web app works on desktop and mobile.

---

Datasets Used

• Diabetes: PIMA Indian Diabetes Database (UCI)
• Heart Disease: UCI Heart Disease Dataset
• Hypertension: Synthetic dataset structured to reflect CDC BRFSS standards

---

System Architecture

1. Data Collection & Ingestion
2. Preprocessing & Feature Engineering: Imputation, scaling, encoding, BMI calculation
3. Model Training & Validation: Grid search with cross-validation
4. Model Serialization & Storage: Pickle/save best models locally
5. Streamlit Dashboard Interface: Interactive web UI for input/output
6. Real-time Prediction & Explanation: Risk score + feature importance visualization

---

Performance Summary

Disease	Model	Accuracy	F1-Score	ROC-AUC
Diabetes	Random Forest	0.73	0.55	0.82
Heart Disease	Gradient Boosting	0.72	0.73	0.81
Hypertension	Logistic Regression	0.82	0.73	0.86

---

Setup Instructions

1. Clone the repo

   git clone https://github.com/your-username/your-repo-name.git
   cd your-repo-name
   

2. Install dependencies

   pip install -r requirements.txt
   

3. Train the models

   python train_models.py
   

4. Run the Streamlit dashboard

   streamlit run app.py
   

5. Open your browser

   • Navigate to http://localhost:8501

---

Future Enhancements

• Integration with electronic health record (EHR) systems
• SHAP value-based advanced interpretability
• Support for additional conditions (e.g., kidney disease)
• Mobile app deployment
• Clinical validation studies with healthcare partners

---

Citation

If you use this project in academic work, please cite: Vasishta Nandipati EMAIL: vasishtavj@gmail.com