🧠 MedPredict – AI Disease Risk Prediction System

MedPredict is a Machine Learning powered healthcare web application that predicts the probability of major diseases based on patient medical data.

The system currently predicts the risk of:

• 🧪 Diabetes
• ❤️ Heart Disease

Users input medical parameters through a web interface and the system uses trained machine learning models to estimate disease risk.

This project demonstrates the integration of Machine Learning, Flask Web Development, and Cloud Deployment.

---

🌐 Live Demo

🔗 Live Application

https://medpredict-nlff.onrender.com

---

🚀 Features

• Diabetes Risk Prediction
• Heart Disease Risk Prediction
• Real-time ML predictions
• Risk probability calculation
• Risk level classification (Low / Moderate / High)
• Interactive web interface
• Cloud deployment using Render

---

🏗 Project Architecture

User (Browser)
      ↓
HTML Forms
      ↓
Flask Backend (app.py)
      ↓
Machine Learning Models (.pkl)
      ↓
Prediction Result

---

⚙️ Technologies Used

Programming Language

• Python

Machine Learning

• Scikit-learn
• XGBoost
• Pandas
• NumPy

Backend

• Flask

Frontend

• HTML
• CSS

Model Storage

• Joblib

Deployment

• Render

---

📂 Project Structure

MedPredict
│
├── datasets
│   ├── breast_cancer.csv
│   ├── diabetes_dataset.csv
│   ├── heart.csv
│   ├── kidney_disease.csv
│   └── synthetic_heart_disease.csv
│
├── models
│   ├── diabetes_model.pkl
│   └── heart_disease_model.pkl
│
├── notebooks
│   ├── eda
│   │   ├── Diabetes_EDA.ipynb
│   │   └── HeartDiseases_EDA.ipynb
│   │
│   └── model_training
│
├── src
│   ├── diabetes_model_testing.py
│   └── heart_disease_testing.py
│
├── static
│   ├── images
│   └── style.css
│
├── templates
│   ├── home.html
│   ├── diabetes.html
│   ├── heart.html
│   └── result.html
│
├── app.py
├── requirements.txt
├── .gitignore
└── README.md

---

🤖 Machine Learning Models

The project uses trained classification models saved as .pkl files.

Diabetes Model

The diabetes model predicts whether a patient is diabetic using features like:

• Gender
• Age
• Location
• Race
• Hypertension
• Heart Disease
• Smoking History
• BMI
• HbA1c Level
• Blood Glucose Level

Prediction is based on probability threshold:

Threshold = 0.45

---

Heart Disease Model

The heart disease model uses a pipeline model which handles preprocessing automatically.

Input features include:

• Age
• Weight
• Height
• BMI
• Hypertension
• Diabetes
• Hyperlipidemia
• Family History
• Systolic Blood Pressure
• Diastolic Blood Pressure
• Heart Rate
• Fasting Blood Sugar
• Total Cholesterol
• Gender
• Smoking
• Alcohol Intake
• Physical Activity
• Diet
• Stress Level

Prediction threshold:

Threshold = 0.30

---

🔄 Application Workflow

1️⃣ User opens the website

Home Page

Options available:

• Diabetes Prediction
• Heart Disease Prediction

---

2️⃣ User enters medical information

Example (Diabetes):

Gender
Age
Location
Race
Hypertension
Heart Disease
Smoking History
BMI
HbA1c Level
Blood Glucose Level

Example (Heart Disease):

Age
Weight
Height
BMI
Blood Pressure
Heart Rate
Blood Sugar
Cholesterol
Lifestyle Factors

---

3️⃣ Flask receives form data

Example:

age = float(request.form["age"])
bmi = float(request.form["bmi"])
hbA1c_level = float(request.form["hbA1c"])

---

4️⃣ Data converted to DataFrame

input_df = pd.DataFrame([input_data])

---

5️⃣ Model prediction

prob = model.predict_proba(input_df)[0][1]

---

6️⃣ Risk classification

if prob < 0.30 → Low Risk
if prob < 0.60 → Moderate Risk
else → High Risk

---

7️⃣ Result displayed to user

Example output:

Disease: Diabetes
Prediction: Diabetic
Probability: 67%
Risk Level: High Risk

---

📊 Model Evaluation Metrics

Accuracy

Measures overall prediction correctness.

Accuracy = Correct Predictions / Total Predictions

---

Precision

Measures correctness of predicted positives.

Precision = TP / (TP + FP)

---

Recall

Measures how many real positive cases were detected.

Recall = TP / (TP + FN)

---

Confusion Matrix

             Predicted
             No   Yes

Actual No    TN   FP
Actual Yes   FN   TP

---

💻 Installation (Run Locally)

Clone Repository

git clone https://github.com/pawan0110/MedPredict.git

---

Navigate to project

cd MedPredict

---

Create virtual environment

python -m venv venv

Activate environment (Windows)

venv\Scripts\activate

---

Install dependencies

pip install -r requirements.txt

---

Run application

python app.py

---

Open browser

http://127.0.0.1:5000

---

☁ Deployment

The application is deployed on Render Cloud Platform.

Deployment workflow:

GitHub Repository
        ↓
Render connects to repo
        ↓
Install dependencies
        ↓
Run Flask application
        ↓
Public URL generated

Live App:

https://medpredict-nlff.onrender.com

---

⚠ Disclaimer

This project is developed for educational and research purposes only.

It should not be used as a medical diagnostic tool or substitute professional healthcare advice.

---

👨‍💻 Author

Pawan Kumar

GitHub https://github.com/pawan0110

---